Open threads at the Open Thread tab every Sunday and Wednesday

Bicameral Reasoning

[Epistemic status: Probably not the first person to think about this, possibly just reinventing scope insensitivity. Title with apologies to Julian Jaynes]

Non-American readers may not be familiar with the history of the US House and Senate.

During the Constitutional Convention, a fight broke out between the smaller states and the bigger states. The smaller states, like Delaware, wanted each state to elect a fixed number of representatives to the legislature, so that Delaware would have just as much of a say as, for example, New York. The bigger states wanted legislative representation to be proportional to population, so that if New York had ten times as many people as Delaware, they would get ten times as many representatives.

Eventually everyone just agreed to compromise by splitting the legislature into the House of Representatives and the Senate. The House worked the way New York wanted things, the Senate worked the way Delaware wanted things, and they would have to agree to get anything done.

This system has continued down to the present. Today, Delaware has only one Representative, far less than New York’s twenty-seven. But both states have an equal number of Senators, even though New York has a population of twenty million and Delaware is uninhabited except by corporations looking for tax loopholes.

To me, the House system seems much fairer. If New York has ten times the population of Delaware, but both have the same number of representatives, then Delaware citizens have ten times as much political power just because they live on one side of an arbitrary line. And New York might be tempted to split up into ten smaller states, and thus increase its political power tenfold. Heck, why don’t we just declare some random farm a state and give five people and a cow the same political power as all of California?

But despite my professed distaste for the Senate’s representational system, I find myself using something similar in parts of my own thought processes where I least expect.

Every election, I see charts like this:

And I tend to think something like “Well, I agree with this guy about the Iraq war and global warming, but I agree with that guy about election paper trails and gays in the military, so it’s kind of a toss-up.”

And this way of thinking is awful.

The Iraq War probably killed somewhere between 100,000 and 1,000,000 people. If you think that it was unnecessary, and that it was possible to know beforehand how poorly it would turn out, then killing a few hundred thousand people is a really big deal. I like having paper trails in elections as much as the next person, but if one guy isn’t going to keep a very good record of election results, and the other guy is going to kill a million people, that’s not a toss-up.

Likewise with global warming versus gays in the military. It would be nice if homosexual people have the same right to be killed by roadside explosive devices that the rest of us enjoy, but not frying the planet is pretty important too.

(if you don’t believe in global warming, fine, having a government that agrees with you and doesn’t waste 5% of the world GDP fighting it is still more important than anything else on this list)

Saying “some boxes are more important than others” doesn’t really cut it; it sounds like they might be twice, maybe three times more important, whereas in fact they might literally be a million times more important. It doesn’t convey the right sense of “Why are you even looking at that other box?”

I worry that, by portraying issues in this nice little set of boxes, this graphic is priming reasoning similar to the US Senate, where each box gets the same level of representation in my decision-making process, regardless of whether it’s a Delaware-sized box that affects a handful of people, or a New York sized box with millions of lives hanging in the balance.

I was thinking about this again back in March when I had a brief crisis caused by worrying that the moral value of the world’s chickens vastly exceeded the moral value of the world’s humans. I ended up being trivially wrong – there are only about twenty billion chickens, as opposed to the hundreds of billions I originally thought. But I was contingently wrong – in other words, I got lucky. Honestly, I didn’t know whether there were twenty billion chickens or twenty trillion.

And honestly, 99% of me doesn’t care. I do want to improve chickens, and I do think that their suffering matters. But thanks to the miracle of scope insensitivity, I don’t particularly care more about twenty trillion chickens than twenty billion chickens.

Once again, chickens seem to get two seats to my moral Senate, no matter how many of them there are. Other groups that get two seats include “starving African children”, “homeless people”, “my patients in hospital”, “my immediate family”, and “my close friends”. Obviously some of these groups contain thousands of times more people than others. They still get two seats. And so I am neither willing to reduce chickens’ values to zero value units per chicken, nor accept that if there are enough chickens they will end up able to outvote everyone else.

(I’m not sure whether “chickens” and “cows” are two separate states, or if there’s just one state of “Animals”. It probably depends on my mood. Which is worrying.)

And most recently I thought about this because of the post on California water I wrote last week. It seems very wise to say we all have to make sacrifices, and to concentrate about equally on natural categories of water use like showers, and toilets, and farms, and lawns – without noticing that one of those is ten times bigger than the other three combined. It seems like most people who think about the water crisis are using a Senate model, where each category is treated as an equally important area to optimize. In a House model, you wouldn’t be thinking about showers any more than a 2008 voter should be thinking of election paper trails.

I’m tempted to say “The House is just plain right and the Senate is just plain wrong”, but I’ve got to admit that would clash with my own very strong inclinations on things like the chicken problem. The Senate view seems to sort of fit with a class of solutions to the dust specks problem where after the somethingth dust speck or so you just stop caring about more of them, with the sort of environmentalist perspective where biodiversity itself is valuable, and with the Leibnizian answer to Job.

But I’m pretty sure those only kick in at the extremes. Take it too far, and you’re just saying the life of a Delawarean is worth twenty-something New Yorkers.

This entry was posted in Uncategorized. Bookmark the permalink.

485 Responses to Bicameral Reasoning

  1. Alice Monday says:

    Fun fact: 2^^^…^^^2 = 4 regardless of how many up arrows there are.

    • Sniffnoy says:

      And in less relevant nitpicks, Vermont was not at the Constitutional Convention; it didn’t exist yet as a US State.

    • anodognosic says:

      I guess I don’t quite understand the notation. Could you explain?

      • L says:

        I tried to link to it, then realized my comment got filtered as spam.

        Search for “Knuth’s up-arrow notation”, which is what it is intended as a short-hand for. The reference is from LessWrong, search for “lesswrong dust specks” or something similar.

        • anodognosic says:

          I remember the dust specks post, but the up-arrow notation just registered as an arbitrarily large number. This helps, thanks.

        • Sniffnoy says:

          I think an easier-to-read formulation is Hyperoperation; H_n(2,2)=4 regardless of n (so long as n>0, of course). But yes just a different notation for the same thing.

      • Anonymous says:

        The notation is wrong. It should be 2↑↑↑…↑↑↑2

        • Izaak Weiss says:

          Using ^ is standard when writing in plaintext on a normal keyboard.

          • Furrfu says:

            Using ^ is standard when writing exponentiation in plaintext on a normal keyboard; for example, in BASIC, bc, TeX, LaTeX, units(1), and on at least TI graphing calculators; or when writing the XOR operation in plaintext on a normal keyboard; for example, in C, C++, Java, JavaScript, Python (which uses ** for exponentiation) and Perl (which also does). If you want to communicate with other people, I recommend that you avoid these ambiguities by using ↑ for Knuth’s iterated exponentiation operator, even if you have to copy and paste it.

            (It’s true that ↑ was also once used for exponentiation, and indeed makes quite a bit more sense for that meaning than ^, which inherited this function (in BASIC) from ↑ along with its code point when ASCII-1963 (the one without the lowercase) gave way to ECMA-6 and ASCII-1967, a process which was not instantaneous; the ZX Spectrum used ↑ at that code point and, I believe, for exponentiation in its BASIC, as late as 1992. But these uses are so old and rare that they will not cause confusion with Knuth’s notation.)

          • Eric says:

            (replying to Furrfu)

            Can you give an example where it would be ambiguous?

      • Izaak Weiss says:

        Basically, x^^y means “raise x to itself y times”, so 2^^3 becomes 2^(2^2), and 8^^4 becomes 8^(8^(8^8)).

        x^^^y means “do x^^x y times”, so 2^^^3 becomes 2^^(2^^2) and 8^^^4 becomes 8^^(8^^(8^^8)).

        so, we can show that 2*2 = 2^2 = 2^^2 = 2^^^2 and on and on as you add more ^’s.

        • Anonymous says:

          “x^^^y means “do x^^x y times”, so 2^^^3 becomes 2^^(2^^2)”

          Are you sure that’s right? Subbing in numbers to your sentence:

          2^^^1 means “do 2^^2 1 times” = 2^^2
          and following on from this,
          2^^^2 means “do 2^^2 2 times” = 2^^(2^^2)
          but you have
          2^^^3 means “do 2^^2 3 times” = 2^^(2^^2)

          Also by my logic 8^^^4 should become 8^^(8^^(8^^(8^^8)))

          • g says:

            It should be “with y copies of x”, so:

            x*y = x+…+x with y copies of x

            x^y = x*…*x with y copies of x

            x^^y = x^(x^…(x^x)…) with y copies of x

            (it’s only from x^^y onwards that we need to put the parentheses in, because addition and multiplication are associative).

    • Edward Scizorhands says:

      2^^2 = 16, so I guess I’m missing something.

      • Anonymous says:

        Sorry, but Alice Monday is actually right. Start from the recursive definition, and consider H_n(2,2). Assume n>0, and

        H_n(2,2) = H_{n-1}(2,H_n(2,1))

        = N_{n-1}(2,H_{n-1}(2,H_n(2,0))). Assume n>=3 (for this inner quantity), and get

        = H_{n-1}(2,H_{n-1}(2,1)).

        Now, look at the second item in the first line and the last item. The only difference is that the second slot went from H_n to H_{n-1}. Thus, we can proceed inductively. We have to be careful, because we assumed n>=3 for this inner quantity. Throw in an inductive variable i, so that

        H_{n-1}(2,H_n(2,1)) = H_{n-1}(2,H_{n-i}(2,1)), for (n-i)>=2. To proceed further, we have to actually compute

        H_{n-1}(2,H_{2}(2,1)) = H_{n-1}(2,2).

        Now, it looks like H_n(2,2) = H_{n-1}(2,2), and we can just perform induction all the way down to zero, but we assumed way up top that n>0, because there’s a special rule for that case. So, the furthest we can induct down is to H_n(2,2) = H_1(2,2). Then, it’s trivial to show H_1(2,2)=4.

        I don’t know if it’s clear, but we have an “inner” induction and an “outer” induction that have different assumptions on the base case.

        • Edward Scizorhands says:

          I botched my calculation. That will teach me to post without coffee. Carry on.

        • Raemon says:

          Can you explain this like I’m five?

          • Anonymous says:

            a*n means to add a to itself n times: a+a+…+a+a where the number of a’s is n. Thus a*2=a+a and 2*2=2+2. Similarly a↑n=aⁿ means to multiply a by itself n times: a*a*…*a*a. So 2↑2=2*2=2+2. Knuth’s notation involves more arrows. a↑↑n means a tower of exponents n steps high, all of the numbers being a: a^a^…^a. So n↑↑2 is a tower of exponents of height 2: nⁿ. In particular 2↑↑2 is a tower of 2s of height 2: 2². Thus 2↑↑2=2↑2=2*2=2+2=4.

            So that corrects Edward’s calculation. And it suggests the pattern Alice claimed. To explain that may be harder. In general, a↑↑…↑↑n with k arrows means to repeat n times the operation with k-1 arrows and all the inputs a. More precisely, it means to use the a n times and to insert the k-1 arrow operator n-1 times in the middle. So if n=2 then 2↑↑…↑↑2 with k arrows equals something using k-1 arrows to join a’s. How many a’s? Just 2. So a single k-1 arrow operator. But if also a=2, then the 2 a’s are 2 2s, so it just reduced the number of arrows by 1 while keeping the numbers being 2.

            (With a chain of addition or multiplication, it doesn’t matter how you parenthesize the operations but with these operations it does. So this isn’t a precise definition. (a^a)^a is not a^(a^a). But if the chains are always length 2, this ambiguity doesn’t come up.)

  2. L says:

    In my mind I thought of this as “people cannot correctly weight values in their (current) environment”. This is definitely in the same basket as scope insensitivity, but it’s important (not equally as important as all biases, perhaps more important than others by a lot!)

    Some similar examples are annoying when they come to mind, like how people think 9/11 is one of the worst events in American History, ignoring that the death toll from it is in fact very low compared to a lot of trivial things, diseases you have never heard of, deaths from car accidents, and so on. It’s given so much weight that it overpowers so many other concepts and allows the government to radically change their laws to keep people ‘safe’ from more 9/11-like events, which also seem to have a low probability of happening to begin with.

    • Ken B says:

      This seems a flaming non sequitur. A thousand auto deaths is about a thousand events, 9/11 was one event. Comparing a large disparate collection of events to one event is a straight up error. Hiroshima wasn’t a big event, because more Japanese die of old age in a year or a decade? Of course 9/11 was one of the worst single events in American history.

      • L says:

        But we can compare them in some ways when we can decide between them, to an extent.

        You’re definitely right in some instances, I should express an example of what I was thinking of:
        what if we spent all the money spent on anti-terrorism efforts post-9/11 to help combat things like deaths from smoking and driving, via education, advertising campaigns, etc. I would imagine there’s some paths that could lead us to save a lot more lives than spending that money with the TSA, NSA et al.

        • Randy M says:

          Well now, if you are looking at how “the money could be better spent on xyz because there are more deaths there” (that is, if we are already using utilitarian calculus) then you also need to consider the benefits for trying to prevent populated cities having skyscrapers knocked down and replaced with smoking craters, etc., which I’d wager is significantly more than replacing 5,000 autos or what have you.

          • LTP says:

            And also the psychological toll that terrorism has on people vs. relatively banal causes of death.

          • Levi Aul says:

            Mind you, effective solutions to skyscrapers getting knocked down don’t involve the TSA (adding air marshals and reinforcing cabin doors seemed to be enough), and effective solutions to the psychological toll of terrorism involve the opposite of getting everyone paying attention to the terrorism. (Effective solutions for that probably involve something like the policy in journalism to avoid publishing stories on recent suicides, to avoid causing even more suicides.)

            Effective spending isn’t actually about how bad the problems are; it’s about how good the solutions are. Or, to put it another way: problems don’t have QALYs; interventions against problems have QALYs. A problem just is. If all the “solutions” to a problem do nothing (or even make you worse-off) with probability 1, then even if it’s the worst problem in the world, it’s not optimal to throw money at these “solutions” compared to trying to solve something else.

            Thus the phrase “low-hanging fruit”: sometimes a problem that isn’t that bad can be the best problem to solve, because solving it is cheap.

          • Alsadius says:

            The concern there is that, while 9/11 only happened once, there was no particular reason to believe it(or something equally bad but different) couldn’t happen again. A 9/11 per year, or per month, starts to make driver education look like a bad investment.

            This is why the question so often boils down to how well organized and effective our enemies are. As it stands, some of what we did seems silly. Frankly, the way the TSA was set up would be silly even if the threat of frequent planejackings was totally real. But a world where terrorism is a House-sized threat instead of merely a Senate-sized threat is quite plausible, or at least was for a couple years.

          • Mary says:

            9/11, unlike car accidents, was the handiwork of people with a motive to keep escalating.

      • Michael vassar says:

        Bundling is the essence of politics. Play that game anyway? Alright. Turn lots of car accidents into GM dismantling public transit systems, or any bit of legislation that did so.

    • John Schilling says:

      Deaths from car accidents are a constant background. There’s things we can maybe do to whittle away at that number, though we’ve picked most of the low-hanging fruit already. There’s no Car Accident Demon that punishes us with extra car accidents until we take the matter as seriously as he thinks we ought to.

      There very definitely was an International Terrorist Demon who was going to punish us with as many terrorist deaths as he could manage until we took the matter as seriously as he thought we ought to, if not necessarily in the specific manner he wanted. Worse, the power of Al Qaeda to kill Americans is strongly correlated with the past successes of Al Qaeda and negatively correlated with American response to Al Qaeda. “Join us and strike a devastating blow against the Great Satan, then party with all the cool people!” is a much better recruiting pitch than “Join us and hide in a cave, and maybe you won’t be killed in a drone strike this week”.

      If an agent proposes to cause you harm until you respond, then unless you are certain he will never be able to cause you a harm you cannot comfortably ignore, the rational course of action is to respond to that agent as soon as you A: notice and B: can spare the time. Doesn’t matter where that agent’s present harms rank on the list of ills confronting you.

      • Paprika says:

        On the other hand, if possible harm caused by the enemy is probably less than the loss caused by taking an action, it is better to ignore the enemy. I am not sure if that is the case with terrorists but it might well be considering the huge negative effects of going to war…

        • vV_Vv says:

          On the other hand, if possible harm caused by the enemy is probably less than the loss caused by taking an action, it is better to ignore the enemy.

          Not necessarily, costly retaliation may be an deterrence effective strategy: If somebody robs a liquor store, society can let them go free or it can try to apprehend them and then put them in a jail for years.
          The cost of this punishment on the society is much higher than the value of what had been stolen in that specific occasion, but you have to compare it with the counterfactual world where all liquor store robberies go unpunished. If the value loss in the conterfactual world is higher than the punishment cost in this world, then the deterrence strategy is cost-effective.

          It works the same for terrorism and warfare.
          For instance, the Soviets had (have?) a “dead hand” nuclear warfare system that could be programmed to automatically (or semi-automatically) launch a retaliatory strike against NATO countries if the USSR (specifically Moscow) was hit by a nuclear strike.

          The system was costly, and it does you no good to attack the NATO once you are dead, but as long as the NATO knew of the system, they were deterred from attacking.

          • ad says:

            IIRC, the soviet system you are thinking of – Perimeter – was unknown to the West, and therefore had no deterrent value whatsoever.

            Rather reminds me of the Doomsday Machine: “Why didn’t you tell the world!”

          • John Schilling says:

            It was also an optional thing that could be turned on in a crisis, e.g. if a radar station saw what might be a salvo of ICBMs bound for Moscow but they weren’t quite sure, and only on the command of the highest authorities. Not something that was intended to be left lying around in the armed state in peacetime.

            The Soviet Union didn’t have as many redundant early-warning and command-and-control systems as the United States, and they had a shortage of human commanders outside of the Kremlin inner circle that they could trust with nuclear launch authority, so this sort of thing made some sense for them. As you note, it would have worked better if they had told anyone about it.

          • RCF says:

            “IIRC, the soviet system you are thinking of – Perimeter – was unknown to the West, and therefore had no deterrent value whatsoever.”

            The US may not have had specific knowledge, but it was reasonable to suspect that the USSR had it. There’s a bit of a paradox here: even if the USSR had told the US about, what would that have meant? The rational thing to do would be for the USSR to claim to have it, but not actually spend money on it. And given that the US would realize that this would be the rational thing, there would be no way to convince them that they had it. The USSR claiming to have it doesn’t really add much to the situation.

          • Nornagest says:

            And given that the US would realize that this would be the rational thing, there would be no way to convince them that they had it.

            The US and Soviets had more communication channels than what they explicitly said to each other, many of them covert: satellite photos, wiretaps, defector reports, old-fashioned human intelligence, and so on. It’s certainly possible to lie to your opponent across those channels (a few years earlier, for example, the Americans and Brits had engaged the Germans in a massive and spectacularly successful fake-out campaign over the eventual invasion of France), but it’s a lot harder than sending diplomatic misinformation, and at some point it becomes easier to just build the damn system.

            There’s also the usual problem with crying wolf: it’s impossible to fake anything big forever (Operation Fortitude only lasted a few months), and once a few of your doomsday claims have been debunked, your opponents will likely be tempted to dismiss any new ones. Since the whole point is deterrence, this is the opposite of what you want.

      • imuli says:

        As of 2001 we very clearly hadn’t picked all the low hanging fruit, as 2009-2013 saw about 9000 fewer deaths per year than 1998-2007.

        • Hari Seldon says:

          I don’t have a source, but I seem to remember reading that a non-trivial portion of that reduction was due to gridlock in major cities and suburbs. People simply aren’t going fast enough to die in accidents.

        • Adam says:

          Given the timeline, it seems feasible as a hypothesis that this is partially due to all the highway funding after ARRA. If I recall correctly, most traffic deaths occur at intersections and elevated highways don’t have intersections.

          I say this without evidence, of course, and may be tremendously wrong.

      • RCF says:

        “There very definitely was an International Terrorist Demon who was going to punish us with as many terrorist deaths as he could manage”

        Well, gee, if you’re going to hide your claim in metaphor, then that does mean that no one can pin you down enough to actually dispute it. If you’re going to anthropomorphize terrorism, you can do the same with car accidents. The consequence for not taking anti-car accident steps is to have more accidents.

        A harm being caused by an agent does warrant a greater response, but it doesn’t just any level of response. On September 11, 2001, about one third of American deaths were due to terrorism. Why are we putting such massive resources towards something that, in that absolute worst case, was responsible for only one third of deaths? Al Qaeda spent years planning 9/11. Even if they managed to pull of one of them a year, that would be about one tenth of one percent of deaths.

        And if you’re so hung up on agency, during 2001, terrorism made up about one sixth of homicide deaths.

        • John Schilling says:

          Is it even possible to anthropomorphize Osama bin Mohammed bin Awad bin Laden? I mean, he was an actual, specific, individual human being before anyone here knew or cared what he was. Might as well talk of liquifying the ocean.

          • RCF says:

            I said “anthropomorphize terrorism”, not ” anthropomorphize bin Laden”.

          • Other says:

            @RFC I think what you meant to say there was “I’m sorry. I didn’t pick up on that when I read your original post.”

            It doesn’t matter whether you said “anthropomorphize terrorism” or “anthropomorphize bin Laden.” Semantic arguments don’t actually change the facts.

            In point of fact, there was an actual human (and more importantly an actual organization) specifically recruiting people to engage in terrorism. Moreover, Schilling’s original post used particular phrases that clearly indicated that he was referring to bin Laden and Al-Qaeda. (Referring to the United States as “the Great Satan” being the most non-ambiguous of them.)

            The presence of an actual agent also doesn’t matter. What matters is the causation.

            Due to the way recruitment for terrorism works, in the absence of a response one successful terrorist attack causes more to follow unless something is done to interrupt that causation. Due to the way people respond to car accidents, one extremely destructive accident causes a reduction in car accidents (because people get scared while it’s still on their mind and drive a little more cautiously for a while). One of these two things is a buffered system, which means you can safely ignore it without things getting any worse. And the other is a self-escalating system, which means once it starts propagating in a direction in a way that costs you something, the costs to you grow until you interrupt and reset it. (And the costs to interrupting and resetting the system likewise grow the longer you put off doing it.)

            Your murder argument is a good argument, because murder is also something that self-escalates. The more people get away with murder, the more people will expect to be able to get away with murder, and the more people will commit murder. This arguments involves people again, but that’s not as important as the direction of the causal arrows. (For something that doesn’t involve human agents. If you live in an old house, cleaning a speck of black mold from your shower should be a higher priority than getting rid of the lead pipes in your plumbing even though a speck of black mold is less dangerous than consuming lead. Unless you clean it, the mold will continue to grow until it eventually is a bigger health hazard than the lead. The longer you wait the more of it you will have that you need to clean.)

            Returning to terrorism: You can have all kinds of arguments about whether a particular response fueled the escalation or actually reset things towards a safer condition. (It appears to me that the net effect of the US involvement in the middle east over the past decade and a half is to destabilize the region and make the world less safe. Even if it has prevented further terrorist attacks for the time being, I wouldn’t expect that trend to continue indefinitely.)

    • It’s not unreasonable to include delibaratrnesss in your evaluation of moral badness.

      • vV_Vv says:

        It’s not just an issue of “moral badness”, it’s also an issue of game theory: a terrorist attack is a defection, and if you let a defection slide without retaliating, your enemies are going to do it again, more frequently and more strongly.

        • Michael vassar says:

          Maybe… And maybe that’s why the terrorists really did one over on Canada, which never bombs them. OTOH, India. These group identities are probably the problem.

        • Jesse M. says:

          “a terrorist attack is a defection, and if you let a defection slide without retaliating, your enemies are going to do it again, more frequently and more strongly.”

          Is that a falsifiable hypothesis, or just an intuition?

          • Other says:

            It is certainly a falsifiable hypothesis.

            The evidence supporting it may well just be an intuition.

            Those are two different classes of things, and it is easily possible for something to both just be an intuition and in principal also be a falsifiable hypothesis.

            More specifically (and longer):

            “If you let a defection slide without retaliating, your enemies are going to do it again” is a claim that has been strongly validated by experiments. Someone could argue that the claim “A terrorist attack is a defection” is part of a bate-and-switch argument used to garner evidence from the sort of experiments that can be conducted in a lab, to something more complicated that only occurs in vitro. However, one can design an experiment to test whether that claim is valid in context or a bate-and-switch by observing whether ignoring terrorist attacks or responding to them was more likely to prevent future attacks, since that’s the property of a defection that matters in the context of the quote. So again, it is, in principal, falsifiable. You would have trouble running an experiment to test it in practice because nobody has any spare countries under their control sitting around in the lab.

    • Matt Goldenberg says:

      I think the issue with 9/11 is that it’s a black swan, Nassim Taleb would say it exists in extremistan. Sure, it wasn’t that bad, but terrorists attacks COULD be EXTREMELY bad. On the other hand, car accidents could never be devestating on that scale.

  3. Ken B says:

    An advantage of having a senate too is that it means ideas must pass muster under two different sets of criteria.

    • Drew says:

      I agree. And seeing it as a “must pass 2 sets of criteria” makes it seem a lot more fair.

      The ‘trick’ is realizing that, absent congressional legislation, states are free to set whatever rules they want. So, the answer to, “the majority of the population wants X” is “Great! Let them pass it in their state legislatures!”

      The Senate only becomes relevant when New Yorkers decide that they want to have New York’s preferences imposed on other states. It seems perfectly reasonable to have a second standard there, as a check on the “does this need to be decided nationally?” question.

      • Levi Aul says:

        Mind you, a ton of things have been “imposed on other states” with such sweeping scope that now a state that wants to do its own thing has to get the federal law affecting it changed in order to do so. A state couldn’t just say some FDA or SEC or FCC policy doesn’t apply within its borders.

      • Alsadius says:

        Federalism is a lovely idea in theory. Shame it’s been so thoroughly ignored in practice since FDR got elected.

      • RCF says:

        “The ‘trick’ is realizing that, absent congressional legislation, states are free to set whatever rules they want.”

        No, they are not. The whole reason for the need for government in the first place is coordination problems. Those coordination problems exist between governments as well as between individuals.

    • Deiseach says:

      The idea of equal representation independent of population is so that smaller states don’t get steamrollered by larger ones; this is the same principle in the E.U. (as a smaller state, Ireland is very interested in this).

      If you only have the House model, then New York’s interests outweigh those of Delaware, even for things that have a very bad impact on Delaware. New York could (for instance) vote to make Delaware its giant landfill dumping site and Delaware could just lump it if they didn’t like being turned into the state tip for New York.

      Or as in England, where the Tories are in power to the shock and despair of many (though I can’t see how this was not forecast to be the result) even though they haven’t won a majority of the votes cast. Or as with the recent referendum on Scottish independence, where the “No” vote was going on about how “We’re better together”, yet when it came to the national elections it was scaremongering about the horrors of having Scottish MPs able to be powerbrokers.

      As for liking some policies from column A and some from column B, so what? Nobody is 100% perfectly consistent in all positions (unless you’re filling out those surveys on the Trolley Problem and decide to stick to one position when answering all the questions so as to avoid the “gotcha!” over inconsistency, even if otherwise you’d have more wiggle room in your decisions).

      It really does come down to what you personally judge most important. I’m sure there are people who’d put fighting global warming over starting a war in Iraq, or vice versa. You’re never going to get a government or president or Prime Minister who matches up perfectly on everything you want or consider important. You have to make compromises and settle for “Two out of three ain’t bad” (and remember that campaign promises are not worth the breath taken to utter them; I remember reading with jaw agape the kind of gushing about Obama being a Lightworker the first time round; no, the man is a politican and what’s more, a Chicago politician. He knows how many beans make nine!)

      • DiscoveredJoys says:

        Or as in England, where the Tories are in power to the shock and despair of many (though I can’t see how this was not forecast to be the result)…

        My own suspicion is that the various polling organisations were still using questions valid in previous elections (thinking consistency) without realising that the recent electorate were responding to different circumstances (the collapse of a middling party and the rise of two significant alternatives). You ask the wrong questions, you get the wrong result.

      • Alsadius says:

        But that’s only relevant if you assume that the state is the fundamental unit of political transaction, not people. I mean, why is NYC more likely to screw Delaware than Rochester?

        Also, no party has won a majority of the votes in the UK since 1931, so that’s saying very little. The UK is not the US.

        • onyomi says:

          It’s a vestige of the federalist system the Constitution was supposed to establish: not a single large state with a bunch of administrative districts, but rather a federation among several mostly sovereign states for the purposes of defense, trade, and a few other things. As Deisach says, this is necessary if we conceive of the people of Delaware and the people of New York as inhabiting two different *states* (as in, sovereign entities), rather than two different administrative districts. Otherwise, Delaware would have more to lose than to gain by the federation.

          And if it sounds like the implication is that a strong national government with control over a big, diverse area will tend to abuse the people of sparsely populated areas at the expense of people in the big cities, well, I think that is kind of true. Look at the USSR.

          Increasingly, the “federal government” of the US is becoming a national government, and superseding the authority of the 50 states. To the extent this is true and/or desirable, the House model makes sense. To the extent we want to preserve some sense of state autonomy, the Senate model is also necessary.

          Put another way, if we wanted to be really take this “House” model to its logical extension, the United States government would be run entirely for the benefit of the peoples of India and China. After all, they are effected by its actions, and there are a whole lot more of them. Why shouldn’t they get an equal say?

          • ddreytes says:

            This is not entirely true.

            The US did have a government that was meant to be a federation of sovereign states; it was the Articles of Confederation. It’s certainly true that the Constitution as originally envisioned had a very different balance of power between the states and the federal government, but it was not universally envisioned as a bunch of sovereign states.

            I mean, there was always an argument about it, and there’s always been a tension between different interpretations and a tension between the states and the federal government, so that it’s difficult to say that it ever was or ever was intended to be just one thing. There were certainly people who regarded it that way. But I think people are maybe pushing back a little too hard on the role of the states, and I don’t think we can say that the framers intended it to be regarded that way, or even that it was the dominant understanding of the relationship between states and the federal government at the time.

          • onyomi says:

            Yes, the Constitution is a stronger federation than the Articles, but I think most states ratifying the Constitution didn’t see themselves as wholly ceding their sovereignty to become like French departments, though there were certainly some, even then, who would have liked that.

            And this is one of many reasons I hate political power: the “camel’s nose under the tent” effect.

        • Leo says:

          The UK is indeed not the US, but that leaves open the possibility that when a party doesn’t get the majority of votes cast, they form a coalition. Obviously this is what happened last time in the UK, though there is plenty of precedent for the Tories going it alone this time. In some other European countries having a coalition in power is absolutely normal.

          As an Irish person I’ve thought about the bicameral representation in a federal system question. The only conclusion I’ve come to is that the current structure of the US Senate was absolutely necessary at the time, but may be due for reform now. At this point, the USA has existed for over 200 years. New Yorkers hardly regard Delawareans as the savages out foreign. A common identity has been forged.
          In Ireland right now, and many other EU countries, the older generation at least regard the other states as very much being foreign, feeling no more kinship for the Italians than they do for the Mexicans. We definitely don’t want to cede too much power to people we don’t feel any solidarity with. If the union persists, then in 100 years things might be different. Then people might be able to accept the notion that a tiny country like Portugal shouldn’t have the right to veto legislation approved by 25 other states.

          • Nornagest says:

            New Yorkers hardly regard Delawareans as the savages out foreign.

            This may be true, I don’t know. But Californians (or, rather, the urban Californians that dominate politics and media in the state) do regard Texans as something akin to ignorant barbarians. It’s probably not as great a cultural divide as between e.g. Irish and Italians, but it might be as big or bigger than between e.g. Argentinians and Uruguayans.

        • Charlie says:

          What’s that quote about world peace? Something like “People worry that without armies and guns, some state – say, Pakistan – would see that its neighbor India had no armies and no guns, and so Pakistan would muster up armies and forge guns all over again to take the land of India. The goal of world peace is for this to be as preposterous as the people of Vermont mustering an army to attack the undefended border of New Hampshire.”

          • Tracy W says:

            @Charlie: My parents lived in Massachusetts a bit in the 1970s, and my Dad likes telling a story about conflict between Massachusetts and New Hampshire while they were living there. Apparently New Hampshire had/has much lower sales taxes than Massachusetts, so of course Massachusetts residents would drive over to New Hampshire to do their shopping just across the border. The Massachusetts government didn’t like this, so they sent state troopers to note people’s licence plate numbers in the car parks of the shopping places just across the border. The New Hampshire government didn’t like that, so they sent their state troopers to biff the Massachusetts state troopers out.

        • RCF says:

          “But that’s only relevant if you assume that the state is the fundamental unit of political transaction, not people.”

          Clearly, states are conceived as being a fundamental unit of political transaction. And perception is reality in politics.

      • ad says:

        yet when it came to the national elections it was scaremongering about the horrors of having Scottish MPs able to be powerbrokers.

        Gordon Brown was a Scottish MP when he was Chancellor of the Exchequer and later Prime Minister. So was Allister Darling when he was Gordon Browns Chancellor.

        No one seemed to mind that.

        What people do mind is having Scottish Nationalist MPs as powerbrokers, as people can reasonably believe that they don’t care about those people – 90% of the population of the UK – who are not Scottish.

        • Tracy W says:

          Well, not so much the SNP “not caring about those people”, as the SNP having fundamentally different ideas about what was good for “those people”. A bunch of arguments for Scottish independence were that Scotland would not be subject to Conservative English ideas about fiscal austerity, and the like. If you’re English and voted, or came close to voting, for the Conservatives in 2010 *because* they promised to cut the deficit and the like, then a political party that spends a lot of time talking about its opposition to those ideas is going to be alarming. (Note, personally I voted for the Lib Dems.)

      • Furrfu says:

        The idea of equal representation independent of population is so that smaller states don’t get steamrollered by larger ones.

        Right. To put it a different way, the idea of equal representation independent of population is so that the smaller states ratify the Constitution instead of forming some kind of smaller confederation that excludes Virginia and New York. It makes sense in the context of states as self-interested political entities whose populations have interests that possibly conflict; it doesn’t make as much sense when you can commute across three states by train to get to work. But to get those independent states to give up their independence, you have to be able to credibly commit to continue to respect the promises of special privileges you made ahead of time. (The same logic applies to the treaties with Native American tribes, but maybe because they were militarily weaker than Delaware but more likely because they aren’t white people, those promises have not been kept.)

      • Adam Casey says:

        (though I can’t see how this was not forecast to be the result)

        apparently there’s an effect where people are shy about mentioning to pollsters that they are voting Tory…

        Or as in England, where the Tories are in power to the shock and despair of many

        I wonder why that effect might exist.

    • Anonymous says:

      I immediately thought of Bond v. United States (the 2010 version) and the fact that the US Senate was originally elected by the state legislature, not by popular vote. From Kennedy’s opinion in Bond:

      The federal system rests on what might at first seem a counterintuitive insight, that “freedom is enhanced by the creation of two governments, not one.” The Framers concluded that allocation of powers between the National Government and the States enhances freedom, first by protecting the integrity of the governments themselves, and second by protecting the people, from whom all governmental powers are derived.

      The House was intended to be directly responsive to the people. The Senate was intended to be concerned with the interests of the States. The Executive was intended to have the big picture, leadership view and be the spear point for foreign relations. All of these interests must agree before anything can happen. The system is a lot more feature-rich than just scope insensitivity.

  4. Sam says:

    The obvious counterargument is that politics (= the art of achieving societal consensus in the presence of competing interests) is difficult, so that on any one issue one should expect (to first order) only marginal progress. To within an order of magnitude, the “slope” of the incremental improvement is the same for all issues, and to the extent that’s not true, we have no way of knowing in advance which issues will prove to be more important than others in practice. (Sure, the incremental policy improvement on issue A *might* be 3 times better than that on issue B, but how sure are you of that? And what about all the various compromises that will be attached as amendments to the legislation? Etc.) So, to first order, it’s rational to run down the list of issues and treat them as binary questions, and prefer the candidate with whom you agree most often.

    In practice, certain issues like the decision to go to war in Iraq objectively have substantially greater weight, but these situations arise in a stochastic manner. Overweighting a particular policy arena because its realized importance the last time it came up is actually just fighting the last war, and a priori we shouldn’t expect its importance the next time it comes up to be that much greater than the various competing issues.

    To be clear, I think the above is *an* argument for the “policy plank” approach to evaluating political candidates. I do not think it is a perfect argument. (Indeed, I actually think history shows that foreign policy is the single most important arena in which one should evaluate presidential candidates.) But it does have a grain of truth.

  5. I studied the math of this a while back and even blogged it. It turns out that you can prove (mathematically) that if you treat all the weights as 1, your decisions will be 75% as good as if you had the perfect weights.

    I.e., if you take unit-weights (the senate model) and true weights (congress), the decisions will agree with each other 75% of the time under fairly reasonable assumptions.

    Warning: the example I use to illustrate the point is choosing who to date. A lot of people got mad about that, so don’t read if that will bother you.

    • Brett says:

      Your (estimated) error rate is only 25% for randomly generated vectors. But in the real world, we don’t have politicians who randomly draw their positions from a bucket, those positions are very highly correlated with each other.

    • Dale says:

      Interesting result! Explains why equal-weighting is so frequent in finance – and seems to work so reliably.

      • Other says:

        tldr (of what’s below): As is always the case with math, the results are just a rephrasing of the assumptions.

        These don’t so much as explain why equal-weighting would work as assume that it usually does. The actual math Stucchio did does not really match up with the results he claimed when he described what that math supposedly shows using words, at least in the context of this discussion.

        Stucchio’s calculation assumed the true weights was best described by the uniform distribution. He assumed that the distribution of traits should be represented by giving each element of the sample a fifty percent chance of having each trait (a uniform distribution on {0,1} instead of on [0,1]). The biggest problem with the “randomly generated vectors” is that they were randomly generated with the built-in assumption that the true weights should pretty much have noise-level variation to begin with. (In the real world, a lot of things, especially things having to do with probability, have tendency to scale exponentially, so even a uniform distribution is noise-level — linear is log of exponential.) If he had used a function that assumed more variation to begin with, there would have be a much greater disparity between the two results. As is always the case with something provable with pure math, the results depend entirely on the priors. In this case, the priors really don’t seem reasonable in light of the available evidence.

        In particular, Stott’s post can be read as saying “I notice that giving all policies equal weight really isn’t a good way to decide who to vote for because some policies so completely dominate others that you are probably better off basing your decision solely on the consideration of the one or two policies you care about most than you are by listing out all the policies and tallying them up.” Then Stucchio responded, “No, if you just assume that the uniform distribution does a good job of characterizing the distribution of the weights that you give the policies, you’ll find that you get the correct result 75% of the time by doing the tallying.” But that’s a non-sequitur in context because the whole point was that there’s a huge amount of variation between the weights. The math adds nothing to the conversation. It pretty much amounts to saying “Well, I assume there’s not much variation to the true weights. Actually, I treat them as being distributed according to the uniform distribution (which happens to be a special case of the Dirichlet distribution which simplifies the math).” The most important word being assumed, because it’s just an a priori assumption. If a logistic curve distribution for describing values had been assumed instead (which would come closer to fitting the sort of pattern that Scott’s post is claiming reflect the way he would weight the various positions politicians can take), the results would be completely different.

        (Incidentally, I would consider Stucchio’s assumptions to be much more likely to accurately reflect the House/Senate situation than reflect the values related to choosing a relationship partner, or weighting how much you care about various positions a politician can take.)

        It’s pretty easy to find compelling empirically evidence to support belief in huge variation. For example, the brain continuously prunes connections… which is another way of saying that it frequently learns to set the weights associated with certain values to 0.

    • Scott Alexander says:

      I haven’t been able to get the math you’re using so far, but the practical utility seems limited. Suppose we have two candidates and ten issues, with one issue being a million times more important than the others. Candidate positions on all issues are randomly distributed between “correct” and “incorrect”.,

      I guess that the candidate with the correct position on the Important Issue is more likely than chance to be the candidate with the correct position on the most issues total, but neglecting the million-times-more-important issue some high percent of the time still seems really bad.

      • Deiseach says:

        I like having paper trails in elections as much as the next person, but if one guy isn’t going to keep a very good record of election results, and the other guy is going to kill a million people, that’s not a toss-up.

        I’d still prefer the guy with the transparent and accountable position on how the sausage was made votes were cast, because if I can’t even trust the bastard to tell me honestly how he won, how the hell am I supposed to trust him when he says no, of course he won’t plunge the nation into a pointless war? “He who is faithful in a very little is faithful also in much; and he who is dishonest in a very little is dishonest also in much.” 🙂

        • Anthony says:

          “He who is faithful in a very little is faithful also in much; and he who is dishonest in a very little is dishonest also in much.”

          I’d like to believe that, but people do compartmentalize much better than that. Many a man is utterly honest in business yet cheats on his wife, or vice versa. The problem is to detect the boundaries of the compartments.

      • ryan says:

        The theoretical point of a list like this is to reckon who will make better decisions in the face of new issues we haven’t predicted yet. Here’s what Jack Johnson thinks of 20 issues. Here’s John Jackson’s take on them. Pick which one you want making decisions on a million other potential issues.

        It’s not about weighing the relative importance of each issue. It’s more about “Will the country be led better by the candidate who is Myers/Briggs type ESFP or INTJ?” The list is supposed to let you grock the political soul of the candidate.

        Note I think this formulation helps explain why people feel betrayed by Obama on things like the NSA. He was supposed to have a Good (TM) Soul, but we were all deceived, yada yada yada.

      • RCF says:

        Chris said that there’s a 25% error rate. I don’t think that identifying a situation where an error is likely really refutes that. For instance, suppose someone says “Counting cards only improves your accuracy by X%”. You respond with “Well, what if there’s a one-deck shoe, and all the aces and face cards have been dealt?”

      • Right, see my conversation with Sam just below. I’m assuming all possible weightings are equally likely – i.e., 10%, 60%, 30% is just as likely as 98%, 1%, 1%. So implicitly, I’m assuming many issues matter.

        In math terms, my model assumes that the odds of one issue dominating the others by a factor of a million is approximately 1 in a million.

        If you assume that there is a single dominant issue, then an algorithm aimed at finding it and ignoring all the rest is probably the better one to use.

      • DES3264 says:

        Chris Stucchio’s result says that this holds when the true weightings (h) are chosen at random. With your non-random — but very important! — choice of h, this is not true. (And you say you are bad at math!)

    • Fezziwig says:

      Your assumptions about weight distributions don’t match Scott’s. You assume a normal distribution of weights, where Scott’s proposed distribution is sorta bimodal: 99% “matters but not all that much” versus 1% “vast, crushing importance”. Probably agreement between the house and senate goes to zero as the distance between the two nodes of the distribution goes to infinity, but I haven’t thought rigorously about it.

      And FWIW, your dating example probably would have gone down more smoothly with a different choice of characteristics. Just a thought.

      • Sam says:

        Yes, this is the real difference between your models. Scott’s main point is that distributions of importance often have wildly different scales. He’s therefore arguing that we should use a prior on weights that is very different from your normal distribution.

        More mathematically speaking, we’re talking the difference between light-tailed and heavy-tailed distributions. Gaussians are famously light-tailed in that their tail probabilities fall off exponentially fast, so the biggest of n issues will only be about sqrt(log n) times as big as the typical issue.

        In other words, for n around 20, the biggest issue is only 2 or 3 times as big as the ‘typical’ issue. And you’re completely right that when that’s true, giving them all the same weight will yield approximately the same result.

        On the other hand, as Scott describes, very often one single issue ends up completely dominating all the others. I think the best simple model for this is probably a log-normal distribution with a pretty high variance. In other words, the orders of magnitude of issues take on a normal distribution. If you tried to prove something similar here, you’ll find that it’s just not true if that variance is high enough.

        This is why I pretty much only pay attention to foreign policy and global warming for US Presidential elections. The annoying thing is that foreign policy seems to be the hardest to predict especially from the campaign (compare: Obama’s Nobel speech), and no one seems to be able to do much on global warming.

        • My normal distribution isn’t a prior on weights – the prior on weights is Dirichlet(1,1,…,1) (uniform). The normal distribution was just a cheap hack to get around the fact that I can’t figure out the proof for a Bernoulli distribution (coin flip) on whether the feature was present or not. Simulations for the coin flip case are present also, and they agree with the proof in the normal case.

          But now I’m curious. Dirichlet(1,1,…,1) chooses weights randomly from the unit simplex, but I think Scott wants distributions weighted towards the corners. I.e., one issue dominating.

          To adjust the proof to handle a case when there is a high probability that most of the weights are zero, instead of choosing Dirichlet(1,1,…,1), you’d choose Dirichlet(e,e,…,e) for e small.

          The net result in this case is that the integral in the average case analysis becomes Ne^2(N-1) / [N^2 e^2(Ne+1)] ~ (N-1)/(2N(Ne+1). The term inside the arctan then becomes sqrt[(N-1)/(2(Ne+1)] ~ sqrt[(N-1)/2]. The arctan of this approaches pi/2 as N becomes large, so you’ll get an error rate of 1/2 instead of 1/4.

          So yes, if Scott’s model is heavily weighted towards a single issue dominating everything, then you would be far better off finding that single issue and ignoring the rest than on trying to take all issues into account.

  6. Partisan says:

    It’s tangential to your point, but I’d like to at least mention one good reason for non-proportional representation in making decisions that affect a large country: protecting the interests of minorities from the majority.

    Suppose several low-population states legalized gay marriage in the ’90’s, and same-sex couples flocked to those states. Then some time later the high-population states want to ban gay marriage nationwide – under purely proportional representation they get their way.

    This of course cuts the other way, too – the minority can block popular things. But I think protecting minority views is a good heuristic when the utilitarian calculus isn’t easy to do. Election paper trails vs. Iraq war is an easy one, but in other comparisons we’re not so sure about, it’s probably a good idea to give the paper trails at least some consideration.

    • Scott Alexander says:

      This assumes that minorities (along whatever dimension we’re looking at) correspond to states, which is rarely true.

      Also, I think the opposite can be true if the minorities are in high-population states. For example, suppose the US annexed Mexico as a single state. The Mexicans are obviously a minority who are different from the rest of the populace. In the House, the Mexicans are well-represented, since Mexico is pretty high-population. In the Senate, Mexico gets two seats, just like Delaware.

      I don’t think there’s any rule that minorities are more likely to be in low population states than high ones.

      If you were really interested in this, you might want a system like Iran’s, which I think guarantees one Jewish senator, one Christian senator, and so on.

      • JB says:

        That’s probably true about demographic minorities, but industries do tend to divide up along geographic lines. (For example, coastal states will have fishing and trade, prairie states will have farming, states with oil will have oil, etc). So if a small state’s economy is heavily bolstered by the oil industry and many people in that state work in that industry, then there is a kind of relevant minority that can form and be affected by national policy toward oil.

        • Tracy W says:

          Also climate. I’ve heard a *lot* from one of my uncles-in-law about problems with housing standards being set in Canberra and Sydney for housing in Queensland.

          And environmental issues. Possums are endangered in Australia, a major pest in New Zealand (they eat the bark off native trees, killing them, so even environmentalists hate them). Australian culture treats possums as cute cuddly creates, while in NZ possums send shudders up the spine (seriously, even typing this I’m feeling a bit of disgust). If NZ and Australia were in a federation the totally opposite views about possums would be a major problem.

      • Jack V says:

        I was going to say something like this. I think the point is that NOW, it probably DOESN’T work especially well, but when the system was implemented, it would have done (at least for the people whose interests were taken into account at all). So it might be the answer to “why we have this system” even if not “should we have this system”.

      • Jaskologist says:

        The poli-sci term here is “cross-cutting cleavages.”

        As others have pointed out, the original purpose of the Senate was to safeguard (individual) state interests, and protect federalism. That’s all gone now, but the bicameral structure still does provide another advantage: you don’t merely have to build a majority to pass legislation, you have to build multiple, different majorities. The need to draw supporters from different factions helps protect against one single faction dominating all the others, or the scenario of 51% stomping the faces of the other 49% forever. You can’t just run the country off city votes; you have to convince the sprawling countryside, too.

        This used to be further supplemented by a gentlemen’s agreement not to pass large, sweeping changes without some buy-in from the other party, but that bridge is pretty well burned now.

        Fundamentally, though, it is a system built around *avoiding* making decisions unless you really have to. I think it makes a nice intro to what you really wanted to talk about, but the metaphor doesn’t go very far, and will probably derail most of the conversation towards issues of federalism instead. Which is a shame, because I’m more interested in the method of reasoning myself.

    • Peter says:

      I have a pet hobby horse about bare plurals, here a notable example is “minorities” and “minority views”. This is sort-of ambiguous between “some minorities”, “all minorities”, and a variety of other weird-and-wonderful things (“cognitively central minorities”, maybe).

      Anyway, with the Senate system, the quantifier is definitely “some”.

      The “cynical Brit” answer to the question “what minorities were the founders thinking of, when they came up with various protections for minorities?” I think is “those that would have been majorities had each of the 13 colonies each been capable of becoming independent independently, rather than having to form a union with the other 12”.

    • Doug S. says:

      Yeah, it was certainly a good thing when the Senate protected the rights of slaveowners.

      • suntzuanime says:

        Yes, the problem with protecting the interests of minorities is that sometimes the interests of minorities will not align with your interests. Tyranny has a lot to recommend it.

  7. John Schilling says:

    There’s good reason for giving near/small as many mental “votes” – and maybe real ones – as far/huge. Near/small is where we can actually make a difference, big/far we can mostly just make a statement.

    So, yes, war in Iraq, hundreds of thousands of dead, some people think that was just plain murder, others that it was a necessary step for the preservation of Western civilization, but either way it was huge. And the vote was 297/133 in the House, 77/23 in the Senate; even if Congress had been 80% Blue in both houses, we’d still have been going to war in Iraq. The political forces involved in that one were overwhelming, and the post-9/11 dynamic was strong enough that they all pointed overwhelmingly towards war with any and every pointedly belligerent Middle Eastern dictatorship. Why even two Middle Eastern dictators were willing to sign on for that role is another matter.

    Global warming, as you note, even bigger on both sides. And here the political forces point towards overwhelming deadlock – no matter what, nothing substantial is going to be done. That can’t plausibly be changed any time in this decade.

    Gays in the military, that’s comparatively tiny. But it was something that a comparatively tiny number of gay people and their allies were able to actually do something about. They were close enough to the problem to understand what needed to be done and how it perhaps could be done, and they cared so much more than most everyone else that they were able to get it done against broad but shallow opposition.

    Ethical chicken farming probably also falls into this category, at least until we enfranchise chickens.

    If you’re not going to do detailed ethical calculus every time you vote – and unless you’re a politics geek, you’re really not – this is a simple heuristic that works. In times of Great Crisis, resolving the one crisis is the only thing that matters, and you vote on that issue. Otherwise, everything that attracts enough of your attention for you to have formed an opinion about it, is of roughly equal importance and gets one “vote”. If this leads to a problem being undervalued, then that problem will grow to the point where an increasing number of people will perceive it as a Great Crisis deserving of all their votes until the balance is restored.

    • Wrong Species says:

      Maybe we should separate the government by function. We could vote for one congress to represent the economy, one for social issues, one for foreign policy, etc.

      • NL says:

        Finally, someone else who has thought of this!

        The major issue I’ve thought of with this is having to many elections and politicians to keep up with, which would hurt the public and help special interests. And jurisdiction issues. But it’s a really interesting thought, and we do it at the local level a lot (school boards, etc.).

      • Who wouldn't want to be anonymous says:

        Directly elect the chairperson of reach if the standing committees. A lot less voting for pretty much the same effect.

      • Tracy W says:

        Demarcation disputes are the obvious problem.

      • Alsadius says:

        So to pick an example from Scott’s post, does gays in the military fall under the military government or the social-policy government?

        • Wrong Species says:

          That seems to clearly fall under social policy but your point is well taken.

          • Evan Þ says:

            The conservatives who claimed open homosexuality would ruin unit cohesion would argue otherwise.

      • vV_Vv says:

        Or just have direct democracy.

      • FJ says:

        This certainly happens at the state level; in my city, for example, we have an elected mayor (responsible for, inter alia, the police), an elected district attorney (responsible for prosecuting arrestees), and an elected city council (responsible for, inter alia, the budget for the police and DA).

        So if you want to look at the simplified problem of “crime fighting,” it’s obvious that there are three independently elected “governments” with a role: police tactics are decided by the mayor(‘s appointees), what charges to bring are decided by the DA, and how much money each has is decided by the council. A voter who cares passionately about “de-militarization” in policing, but who wants a large police force and heavy enforcement of the marijuana laws, could tailor her voting for candidates that support each of those policies in the relevant field. We also happen to have elected judges and elected public defenders in this city, so you could get really fine-grained if you wanted.

        Demarcation issues are a problem, as has been pointed out. But given that there are so many elected officials, all of whom have a different responsibility and a different mandate, a more common problem is simple gridlock. Our current mayor ran on a strong stop-and-frisk platform, while our current city council refuses to fund it. The DA and the mayor are frequently at loggerheads, and both hate the council. Like it or not, but most functions of government tend to create externalities for other government functions. Separating government decision-making into various organs merely allows those organs to ignore their externalities.

      • Selenae says:

        Issues don’t fit into separate categories. The issues wouldn’t be controversial if they didn’t involve tradeoffs between two different areas.

        The best foreign policy might be to grow enough food and drill enough oil to be self-sufficient. The best economic policy might be to raise extra cows on our abundant land and trade the beef for oil from countries that can extract it more efficiently. The best environmental policy is to forbid raising cows and drilling oil altogether.

    • houseboatonstyx says:

      @ John Schilling
      There’s good reason for giving near/small as many mental “votes” – and maybe real ones – as far/huge. Near/small is where we can actually make a difference, big/far we can mostly just make a statement.

      Yes, usually. But sometimes the near/small turns out to majorly influence the far/huge. Paper trails on votes was a minor, crankist sort of cause — till 2000 — at which point, if it and other vote-recounting standards had been already in place, it might well have got us a different President.

      In times of Great Crisis, resolving the one crisis is the only thing that matters, and you vote on that issue.

      That gives incentive to politicians to announce or re-affirm some Great Crisis once a month whether we need it or not, or at least once per election.

      • John Schilling says:

        Except that Great Crisis means “vote on that one issue”, not “vote for the incumbents”. Unless you can be very certain how the population is going to feel about the crisis and your proposed solution, you’re probably safer with the status quo, with the slate of issues where your stance as an incumbent has always had 60% of the population happy with 60% of your decisions.

      • RCF says:

        The term “paper trail” refers to electronic voting, and as far as I know, electronic voting wasn’t involved in the 2000 kerfuffle. And just because “X would have led to a different” president, that doesn’t mean that X isn’t a crank cause. Reshuffling the deck changes the result, but it doesn’t necessarily improve the result.

  8. Rangi says:

    The Senate view seems to sort of fit with a class of solutions to the dust specks problem where after the 2^^^2th dust speck or so you just stop caring about more of them…

    You could fix this by making the number of representatives proportional to the square root or even the logarithm of the state population, or some other limiting function. This doesn’t solve the dust-speck hypothetical because you can replace 3^^^3 with (3^^^3)^2 or e^(3^^^3) to undo the limiting function, but that won’t happen with real-world populations.

  9. Thecommexokid says:

    Vermont was not one of the 13 original American colonies, and thus was not represented at the Constitutional Convention. You might want to pick a different example.

    • Eric says:

      I recommend using New Jersey for the small state and Virginia as the big state, as those were the sponsors of the two plans at the original constitutional convention. Alternatively, Wyoming and California for maximum population discrepancy.

    • Anonymous says:

      In one sense, it was represented by the representatives for New York (although this ruins the point). In another sense, it was represented by William Johnson.

  10. Steve Johnson says:

    States have equal votes in the Senate because the Senate was intended to represent the interests of the state governments – that’s why Senators originally were appointed by state governments. The government of the State of New York has as many senators as the government of the State of Vermont.

    • Schmendrick says:

      Yup. As originally envisioned, states really were little sovereign polities – the basic building block of American government and society, with local governments to handle the super-small stuff and the federal government mainly handling foreign relations and truly interstate functions like mail delivery, and customs-collection.

      • randy m says:

        It’s funny, I recall thinking art one point, why do we use the same word State to mean countries, when it already means the rather inconsequential political divisions we have here in America? Thinking about that helped understand what the intention of the state’s role was, one I understood the chronology of the terms.

      • Derelict says:

        So would it be more accurate to call them provinces now?

  11. Gary Jones says:

    Those who speak of protection of minorities as justification for bicameralism have the right of it. One useful example of this is the history of the settlement of the American West after it was taken from the indigenous tribes by force of arms and immigration.

    It isn’t just that there are ethical concerns about such taking, it is also that the invading hordes did such a poor job. Decisions and policies made by ignorant urban legislators and bureaucrats were wildly inappropriate for the distant lands, a blunder that still causes suffering, not least the water problems in California.

    It is an ethical concern to have overpopulated areas dominate large but sparsely populated areas, but it is also bad management. The hordes have no relevant expertise.

    It’s fractal. The same dynamic plays out within states as between them. Populous coastal cities dominate the inland areas, and do so ineptly. There’s a pattern here that careful thinkers have long been aware of. Subsidiarity.

    • Pku says:

      The naive approach to this is trying to split this into things that are better managed locally (e.g. land use) and things that are better off managed globally (e.g. bank regulation), or more nuancedly, finding for each issue whether it’s best managed at the county, state, or federal level. This can resolve some things pretty well if handled reasonably (water use, for example – the state government is determined by majority-population, because cities actually do have more people who need water, but how to use it effectively could be determined at a local level). There are certainly issues with it, but it seems like a good enough system that it’s probably better to take it as a base then adjust for issues than to build a senate-based power separation from the ground up.

      • Gary Jones says:

        “finding for each issue whether it’s best managed at the county, state, or federal level”

        That’s what subsidiarity means.

        The Oxford English Dictionary defines subsidiarity as the idea that a central authority should have a subsidiary (that is, a supporting, rather than a subordinate) function, performing only those tasks which cannot be performed effectively at a more immediate or local level. The concept is applicable in the fields of government, political science, neuropsychology, cybernetics, management and in military command (Mission Command). In political theory, the principle of subsidiarity is sometimes viewed as an aspect of the concept of federalism, although the two have no necessary connection. The principle of subsidiarity plays an important role in the political rhetoric of the European Union concerning the relationship between the EU governing bodies and the member states.

        This isn’t just about ethics, it is about competence. Technocrats have some difficulty in facing their own lack of expertise since their ignorance is often exceeded by their arrogance. They mistakenly assume that their policy prescriptions are the best that can be done.

    • Harald K says:

      Bicameralism does not protect minorities in general. It empowers one particular type of minority, a geographical minority. And true, those minorities can use the power to protect their legitimate interests.

      But they can also use it to protect their special interests, at the expense of everyone else. There is no way to only empower minorities (geographical or otherwise) when they deserve to be empowered.

      Far too many people think that they can get a better deal than political equality. Most of them got to be wrong.

      • Tracy W says:

        I think though that “better deal” depends on how you define the alternative. If the alternative is not setting up the United States of America at all, then an unequal deal might be better for everyone. New York gets Delaware’s money and men to help in the fight against Barbary pirates, Delaware gets to be confident it won’t have its interests overrun by New York.

        (Note I’m not an expert on US political history, this characterisation of their positions is hypothetical).

      • Mary says:

        OTOH, the House means that minority can’t abuse its power without end. Only when it horse-trades.

      • Gary Jones says:

        There is no way to only empower minorities (geographical or otherwise) when they deserve to be empowered.


      • Wrong Species says:

        Instead of promoting the interests of geographic minorities through the Senate, there could be a representative for other kinds of minorities. One of the baltic countries(I think it’s bosnia) has a system where three different ethnic groups trade the presidency.

        • David says:

          One of the baltic countries(I think it’s bosnia) has a system where three different ethnic groups trade the presidency.

          According to this, Bosnia has three presidents at the same time, one for each of the three big ethnic groups.

          But I think you meant ‘former Yugoslav countries’. The Baltic countries are Lithuania, Latvia and Estonia, none of which, to my knowledge, has such a convoluted system.

          • Unique Identifier says:

            Baltic and Balkan often get mixed up. The first was explained above; the second refers to the peninsula roughly delimited by Ukraine, Italy, Turkey and the Mediterranean Sea, and thus includes Greece, former Yugoslavia and more.

          • Wrong Species says:

            Oops, that was a dumb mistake. Good catch.

  12. jaimeastorga2000 says:

    Game theory point: Proportional representation incentivizes a political entity or faction to increase their population by whatever means necessary, effectively turning itself into a utility monster. This can take the form of subsidizing reproduction, importing immigrants, expanding the definition of entities which count for representational purposes, or, in the case of factions, converting people to the faction. Over time, political entities and factions which resort to these methods gain more power than those that don’t, and dominate national politics.

    • suntzuanime says:

      Taking this as an analogy to decision theory, this means that scope insensitivity is a possible mental defense against utility monsters.

      • Interesting idea. I think I disagree though, because of the prehistorical evolutionary environment in which our biases formed. AFAIK, it has rarely been important to humans to be able to mentally deal with quantities of objects beyond what is now considered very small (1-100). Having a number “lots” was usually enough, more would be a wasted mental capacity. There isn’t any obvious utility monsters I can think of in our past either. So I think its more likely we just never evolved the equipment to directly imagine very large quantities (and instead use symbolic systems instead).

        I guess there could be a social element too. I wonder if exposure to large quantities in childhood influences scope insensitivity.

        • “There isn’t any obvious utility monsters I can think of in our past either. ”

          There have always been utility monsters, as any parent of small children could tell you. And we devote large efforts to keeping up their utility.

    • Ive often seen that kind of point put forward in conspiracy theory mode, but in practice entities that reach a certain such encounter fission rather than domination.

    • RCF says:

      That’s not a feature of proportional representation, that’s a feature of democracy.

    • Harald K says:

      Your theory does not correspond well to what has historically happened.

      Ancient Athens had as proportional a representation as you could get – they chose virtually all political positions with lottery.

      Your theory predicts that they would have fought to include more groups in the democracy, groups that could be expected to be favorable to their faction. They did the exact opposite. People with foreign ancestry (metics) who had lived in the city for generations were denied citizenship. Very few citizen wanted to extend citizenship to women or abolishing slavery.

      It wasn’t for lack of factions. The aristocrats and the lower classes (the latter which were disenfranchised virtually everywhere else) were intensely distrustful of each other, and there was bloody class war, involving both the legal system and outside forces.

      Even when one of their most respected citizen argued that metics who had just risked their lives for the liberation of Athens should receive citizenship, did they give it? No.

      Maybe this was their downfall. Democracy, like many institutions, got knocked down a couple of times but got back on its feet. By keeping their citizenship ever more like a privilege to be jealously guarded from dilution, with no concerns for justice, maybe eventually the majority didn’t care enough to revive it.

      The “import of compliant citizens” theory is little more than an anti-immigration conspiracy theory. It would be an pointlessly risky way to try to get power, even for your faction, but especially for personal power.

      • suntzuanime says:

        Ancient Greek and Roman democracy is dissimilar enough to modern liberal democracy that I would hesitate to try to draw surface parallels. A better example might be that the Republican party fought a bloody war to free the slaves and extend them voting rights, and the descendants of those slaves do not seem to be showing them much electoral gratitude.

        • Harald K says:

          The parallels are straightforward enough in this case. The question is just about who deserves a share in citizenship power, and whether trying to include more people to gain voters is a good strategy.

          From the other direction… I note also that both Labour and the Tories in UK were strongly against independence for Scotland. Labour worried that they would be a permanent minority in the rest-UK. (That didn’t work out so well for them, did it?). But even the prospect of permanent majority in the (rest of) UK didn’t tempt the Tories to support Scottish independence.

          • suntzuanime says:

            Who “deserves” a share of power and what strategies for gaining power are safe and effective are both questions that can only be answered in the context of a particular political system. Since the two political systems are quite different, the parallels are not at all straightforward. The question of Scottish independence is a better example.

        • I’m pretty sure that blacks were a reliable Republican voting block for quite a long time after the Civil War. Nothing lasts forever.

          • Yes, exactly. From the Civil War until 1932, essentially every black voter was a Republican. That’s almost seven decades.

            From 1932 to 1964, black voters were divided, giving substantial percentages to both parties.

            That period is generally thought to have ended with Barry Goldwater’s presidential campaign.

            From 1964 to the present, black voters have been strongly Democratic.

      • Import of compliant citizens only works if you can predict that they and their descendants will stay allied to your faction. Natural increase of your faction makes a little more sense, given pretty high correlation between political positions of parents and their children.

  13. Pku says:

    A minor (but often important) distinction to make here is that you need to measure not only the importance of the issue but also the size of the effect on it: for example, the president has much more influence on the Iraq war Than on Global Warming (possibly to the degree that the president could have prevented more greenhouse emissions by not having all those tanks and planes emit gas in the Iraq war than he would by tying to fight domestic global warming).
    A more important example of this is the chicken thing: if you’re trying to decide whether or not to go vegetarian, it doesn’t matter if there are twenty billion or twenty trillion chickens out there, you’re not eating the same number of chickens. There are issues (like effective charity) where these kind of efficiency considerations need to be made. But often they don’t, because the bigger the problem is, the less effect your intervention would have on it.
    (In general when talking about thought biases, I remember Hermione’s line from HPMOR: It’s true that appearances can be deceiving… but usually, they’re not.)

  14. suntzuanime says:

    My take on Senate representation is that humans are not the only participants in a democracy. Each state is intended to consist of a community, and those communities have interests of their own and needs for representation. Senate representation is egalitarian on the level of the community rather than on the level of the individual because communities were, at the time the Constitution was being drawn up, considered important.

    Now it’s not clear that state boundaries actually delineate meaningful communities anymore, especially in places like California that are tearing themselves apart with internal strife. And there are certainly many ethical/legal/social philosophies that discount communities as meaningful objects except inasmuch as they describe collections of individuals. But I think that’s the philosophical basis they were working under when they drew up the Constitution. This does not translate well into a metaphor for the human brain, as far as I can tell.

    • LTP says:

      I think they are meaningful communities in as much as there is a longstanding continuity of democratic government there. This affects the kind of people who move there and the political culture and climate.

      That said, I think some of the larger states probably should be broke up. Chicago and New York should break off with their suburbs and be metro-states, California could be split into at least two and possibly more states, plausibly parts of Texas (San Antonio + El Paso + the rest of southern Texas), Florida (Jacksonville + the panhandle), and Washington (east of the mountains) could be split off, though I’m less familiar with those areas.

    • nydwracu says:

      The philosophical basis they were working under was that the noun phrase “the United States of America” took plural number agreement. This ended sometime around 1865.

      • Douglas Knight says:

        That story is too pat. The timing isn’t wrong, but the plural was never solidly established and it didn’t end very abruptly. google ngrams.

    • Harald K says:

      Each state is intended to consist of a community, and those communities have interests of their own and needs for representation.

      Independent from their members? Do we have some sort of clan morality, where the members exist for the clan’s benefit rather than the other way around? If you go down that path, you are not a humanist any longer, but some ghastly collectivist. And if you don’t, there’s no reason to give an institution that amounts to a tool, a vote of its own.

      Federation is a good idea because my group and your group can reasonably agree that some things are internal matters. Even if 90% of the nation thinks one approach to regulating public use of drugs is a good idea, we can still agree to not impose that idea on the last 10% – provided we think it doesn’t concern us, doesn’t affect us, or at least doesn’t affect us enough that we care.

      But don’t for a moment think that the group or organizational entity has any value apart from the people it serves. The question of which questions are best decided on the local level, cannot itself be decided on the local level. That must be decided on the higher level, and there it should be people that get a say, not organizational units.

      • suntzuanime says:

        I’m not necessarily endorsing non-individualist views, just suggesting that being aware of them might make sense of things that otherwise might seem senseless. I would prefer if we could discuss things neutrally without having to moralize at every step of the way.

      • Tracy W says:

        I think this is just a verbal shorthand for what is really going on. Like saying “Teddy Roosevelt built the Panama Canal.” Virtually no adult believes that Teddy Roosevelt actually went down there and shifted rocks single-handedly, but that quick form implicitly utilises a lot of implicit knowledge about how presidents do stuff, and is easier to remember if you just want to have a sense of time frames than a longer, more-accurate position.

        And some things are indeed community-level interests, eg clean drinking water, sewage systems, fire-fighting.

      • vV_Vv says:

        Do we have some sort of clan morality, where the members exist for the clan’s benefit rather than the other way around?

        That’s actually pretty common among humans, and it could be argued that even modern countries essentially work this way.

        You may disagree with the morality of such system, but it doesn’t mean that there aren’t many people who find it compelling.

  15. Wrong Species says:

    Imagine a federation of Humans, Vulcans and Klingons with a population of 10, 20 and 100 billion people, respectively. Would you still feel like the fully democratic system is more fair?

    • Scott Alexander says:

      An earlier draft of this post used the UN Security Council as a similar example – China outvotes everyone else easily.

      But the solution to that seems to be “don’t enter into such a federation”.

      • Wrong Species says:

        The federation does have it’s advantages though. That’s why we have one. So if you want to have a federation while limiting it’s abuses, then having a bicameral body seems to be a great compromise, hence the name.

      • Harald K says:

        With the UN security council, there’s very little pretense at justice – they joined in that federation because they are worried the others in it may nuke them, not out of concerns for equality or fairness.

        People on the far right worry about a world government, but they don’t realize we’ve already got one, and it’s a junta.

      • suntzuanime says:

        Shouldn’t they outvote everyone else, barring a major opposing coalition? You don’t even have the excuse that the Chinese aren’t human. I don’t really see a “fair” “democratic” argument for not giving the aggregate wishes of the Chinese and the Indians substantial deference.

        • Tracy W says:

          The obvious response is that the Indians and Chinese don’t seem to have figured out yet how to govern as well as the Swiss, judging by such criteria as environmental damage, civil unrest, and incomes per capita.

          If you were having surgery done, would you want the surgeon to be selected based on a fair and representative process, or selected based on track record in previous similar surgeries?

          (Switzerland used as an example because they do have an ethnic and linguistically diverse population).

          • suntzuanime says:

            Yes, I would not want my surgery performed fairly and democratically. For my surgery I would want an aristocracy.

            Maybe we should apportion senators to the states according to environmental damage and income per capita?

          • Tracy W says:

            Personally I think the main reason that democracy works as well as it does as a decision-making procedure is that, it, like markets, doesn’t have a pre-defined measure of what is good. Instead, we as individual voters get to decide what we value the most, and we don’t even have to quantify our weights. So, if one idea gets taken to extremes, well, we can just vote those guys out, and ignore any anal-retentive type complaining about inconsistencies and how we all agreed to maximise paperclips, and what are we doing changing the rules right now just because everyone’s floor is a foot deep in paperclips?

            Apportioning votes toward some measure of environmental damage and income per capita defeats that advantage.

            (Note: I favour democracy and markets in the Churchillian sense of “worst of all systems apart from all the others that have been tried from time to time”).

          • suntzuanime says:

            So what makes this not apply to Indians and Chinese, then?

            See also:

          • Tracy W says:

            Well, in China’s case, it’s not a democracy.

            The second part of my answer is that I think that democracy is only one part of good government. A very important part, but not the only part.
            Other sorts of things, like people’s willingness to agree to compromises, matter too. Eg, take Northern Ireland, as I understand it, peace eventually happened and is kinda-sticking because enough people on both sides were sick of the ongoing violence, not because 51% of the population voted for peace.

            And politicians’ experience at their jobs is another issue, as far as I can tell, being an effective political team is like being an effective surgeon, or an effective soldier, there’s no substitute for battle experience.

          • ” For my surgery I would want an aristocracy”

            Don’t you mean meriticracy?

        • vV_Vv says:

          Self interest: other countries would not enter the UN Security Council if that would mean surrendering their military policy autonomy to China. As a result, there would be no UN Security Council, and the world would be arguably less safe.

        • RCF says:

          Except that the Chinese don’t vote. How about we apportion representation in the UN according to how many actual freely cast votes for those representatives there are?

          There was a similar issue in the US, with whether to apportion representation in the House according to how many people a state had, or how many citizens. The South wanted to count slaves as people for apportioning representation, but not for anything else. Why should the Chinese government get credit for representing 1.3 billion people, when they do not in fact represent 1.3 billion people? Either individuals are the basic political unit, in which case Chinese’s government is illegitimate, or nations are the basic political unit, in which case why does China get any more representatives than anyone else?

          • Irrelevant says:

            Why should the Chinese government get credit for representing 1.3 billion people, when they do not in fact represent 1.3 billion people?

            I’m unconvinced any democracy above the size of Iceland actually manages to do something plausibly described as “representing its people.” And Iceland went off the rails ten years ago.

      • Deiseach says:

        Yes, but “What’s sauce for the goose is sauce for the gander”. You can’t tell the people of Delaware “Well, tough, you don’t have as many people as New York so you will just have to get used to being outvoted by New York because that’s the fairest way to do things”, then turn around and say “Hey, it’s not fair that China gets to outvote us just because they have more people!”

        For myself, I enjoy the Schadenfreude of seeing English politicians protesting about decisions that affect Great Britain being made on law and government by an alien parliament far away that doesn’t speak the same language, share a common culture (or even religion, when that was important) and pursues its own interests and the interests of influential groups over those of Britain, and how the lasw of the national government can be over-ruled by decisions in that foreign parliament, and how unfair and unjust this is.

        You don’t say! What, you mean exactly like how you governed us after the Act of Union of 1801 and the dissolution of our national parliament? 🙂

        • Tracy W says:

          In defence of the current attitude of the English, they did spend the next hundred-odd years after the Act of 1801 learning what a bad mistake they’d made.

          And the British empire has been broadly criticised as bad for the English, as well as the people they were ruling, for centuries, eg Adam Smith:

          Under the present system of management, therefore, Great Britain derives nothing but loss from the dominion which she assumes over her colonies.

          (This chapter is about the Americas, not Ireland, but the arguments made are as applicable).

        • Alsadius says:

          And the Irish choosing to separate peacefully from the UK would have been democratically acceptable, just like the UK doing the same to the EU.

          • Deiseach says:

            We did vote to separate democratically. The Brits then spent nearly forty years bitching about that, before begrudgingly agreeing to a Home Rule Bill which was then put on ice due to the First World War breaking out.

      • John Schilling says:

        It may not be entirely coincidental that UN security council membership maps 1:1 with arsenals of thermonuclear ICBMs, that both lists are codified by international law, and that the nation presently making the best and strongest case for becoming the sixth member of the UNSC is also the nation closest to becoming the sixth thermonuclear ICBM power.

        A veto is a veto, and we’d really rather not have the other sort ever be cast in a political dispute.

        Similarly, each of the original 13 states had an Army, and even a small interstate war would have been an invitation for British reinvasion or other foreign meddling.

      • Salem says:

        Your answer is far too pat. If you don’t enter into such a federation then you’re liable to get assimilated by the Borg. So yes, there are costs, but there are also benefits, and at a certain point it makes sense to form the federation and try and resolve the tricky representation issues.

        And this is the background to the Senate too. Rhode Island wasn’t entirely happy to form a union with far more populous states like Virginia (which was in part why they were the last to ratify) because they were worried about their concerns being overwhelmed by far greater numbers. But they were also worried about their viability if they didn’t ratify. Similarly, Virginia wanted to bring the smaller states along and recognised the need to compromise in order to do so.

        This kind of representational negotiation is a repeated story in the formation (Netherlands, UK), non-formation (Qatar and Bahrain not in the UAE, Singapore not in Malaysia) and dissolution (Yugoslavia, USSR) of polities the world over. It’s not merely a “decision procedure” for the combined entity, it’s a negotiated means of reconciling other entities to the federation – entities which may often have a stronger pull than the federation itself. People and groups need to be able to co-operate in a wide variety of circumstances, not just ones that make proportional representation easy.

      • onyomi says:

        But the citizens of bigger states knew it had to be like that in order to have the federation at all. Presumably a majority of those making the decision thought it was worth it, just as a majority of Delawarian decision makers thought it would be worth it to let the New Yorkers have way more representatives, so long as they got an equal number of senators.

        Also, the US government does things that affect the Chinese, and we are all just people living on the same planet. Why don’t the Chinese get an equal say in what the US government does?

    • Harald K says:

      And what do you propose the humans and vulcans should do, which would be better than democracy?

      Remember, you’ve got to convince the Klingons it’s a good idea too.

      • randy m says:

        It depends a bit on whether there are also Borg etc or not out there, but basically their choices are fight or negotiate, our as the power discrepancies grow, die or submit.

        • Harald K says:

          Let’s say the klingons, for some reason, put democracy on the table.

          “You are not sufficiently worthy to be our enemies. Out of scornful pity, we give you this gracious offer of union: There will be a parliament consisting of a hundred randomly chosen Klingons, twenty vulcans, and ten humans. They will judge in disputes between our peoples and pass appropriate laws to ensure an honorable coexistence. You have ten hours to consider our generous proposal, if you do not accept we will simply put an end to your miserable existence.”

          Sorry if this is not plausible Klingon, I’m not really well versed in the Star Trek universe. So I’d appreciate if we keep the Borg out of this too, I know even less of them.

          How do you propose you get a better deal from fighting? How would you negotiate? How would your negotiation lead to anything better than could be achieved in the klingon-dominated parliament?

    • I think that would be completely fair. I also think there’s no way I’d be supprting that fair system until I knew that Vulcans and Klingons were committed/wanted human survival. Kind of like the status very tall/short people have now in a population of shorter folks – sure they’re a group, but they’re one of us so nobody gangs-up to try to cannibilise them.

  16. ddreytes says:

    The bicameral compromise was, I think, somewhat more elegant than just a compromise between the interests of large and small states. It seems to me that it also represents a compromise between democratic & aristocratic modes of government in order to end up with a mixed constitution, which was a commonplace of republican theory. The Senate doesn’t just represent the small states, it also was meant to represent a more balanced, stable element of government, separated from the direct influence of the people (although that is much less the case now, since we have direct representation of senators). Of course these days we tend to see everything through a democratic lens so the Senate model seems completely outmoded and insane. It is meant to be kind of an aristocratic body.

    So (as others have pointed out, I think) it’s not just a question of different weighting; it’s a question of fundamentally different paradigms, and really, I think, different functions.

    Of course what (if any) relevance this has to a post that seems to have really been about weighting priorities, I don’t know.

    • LTP says:

      That is an excellent point about the Senate. Sometimes I feel like we’d be better off in the 17th amendment was abolished and we returned to Senators being elected by state legislatures. Of course, the potential problem with this is state legislative elections could become about which party you want to represent your state in the federal government, and so state legislative candidates would be elected purely on the voters’ impressions of the *national* political parties and national issues, thus leading to state issues to be ignored in the election season in all but extreme cases.

    • If you’re going to call the Senate aristocratic, what are you going to call the House of Lords?

      It’s stretching a point too far to call an intuition that is democratic and not hereditary aristocratic,

      • Schmendrick says:

        The Senate can be called “aristocratic” in the classical greek typology of political forms – democratic/ochlocratic; aristocratic/oligarchic; monarchic/tyrannic. In this framework, an “aristocratic” system is one that emphasizes rule by a (hopefully virtuous) elite rather than one that takes popular opinion into account since, as the theory goes, the great mass of the people is prone to passionate outbursts and greedy “leveling” impulses. The precise details of the selection method don’t matter a whole lot, so long as the process is insulated from anything resembling a plebiscite.

        • I guess Im just going to have to regard that typology as out of date.

          • ddreytes says:

            I am amused at the name / comment combination.

            In any case, it may or may not be out of date now – I don’t think it is, personally, but whatever. But I don’t think it would have been out of date at all for the Founding Fathers.

      • houseboatonstyx says:

        If you’re going to call the Senate aristocratic, what are you going to call the House of Lords?

        I wondered when someone would mention the Lords. A similar feature is term length. A Lord has tenure so to speak. Senators don’t officially have that, but their terms are quite longer than a Representative’s: 6 years vs 2 years, and the 6 year terms are staggered. So to get Senate approval for any very drastic change*, your faction would have to win quite a few races in quite a few states.

        * And the Senate must approve other important things such as treaties and appointees, and make the verdict on impeachments.

        It’s stretching a point too far to call an intuition that is democratic and not hereditary aristocratic,

        It does often seem to work out that way in the US, though.

      • Adam Casey says:

        The founders explicitly refer to the constitution having a “monarchical element” (president), an “aristocratic element” (senate), and a “democratic element” (house). It’s the division that seemed natural to them.

        • Ok, I’m grokking that now.

          But it still isnt real monarchy/aristocracy.

          Although it still does show that carefully designed democracies try hard to head off the objections to democracy from other systems.

  17. TeslaCoil says:

    Obviously some of these groups contain thousands of times more people than others. They still get two seats. And so I am neither willing to reduce chickens’ values to zero value units per chicken, nor accept that if there are enough chickens they will end up able to outvote everyone else.

    I think the latter issue is different: utilities are assumed to be real numbers. I’ve always found this strange, since my intuition tells me that utilities can easily be non-commensurable. This should not be mistaken for scope insensitivity.

    Consider the following 3 scenarios regarding a nuclear reactor.

    1. Reactor temperature is optimal.
    2. Reactor temperature is 1K above optimal.
    3. Reactor is melting down.

    Scenario 1 is preferable to Scenario 2, and Scenario 2 is preferable to Scenario 3.

    The usual dustspecks dialogue goes something like this:

    Straw Utilitarian: Would you prefer Scenario 3 happening to one nuclear reactor, or Scenario 2 happening to a million nuclear reactors?
    Me: I would prefer Scenario 2.
    Straw Utilitarian: Ok, now what about Scenario 3 happening to one reactor, or Scenario 2 happening to 3^^^3 nuclear reactors?
    Me: I still prefer Scenario 2.
    Straw Utilitarian: Surely you don’t understand big numbers.

    Does this mean I am being scope-insensitive? No! I prefer “Scenario 2 happening to a single nuclear reactor” over “Scenario 2 happening to K reactors” proportionally to K. But that does not overwhelm Scenario 3, for any value of K.

    • JB says:

      That’s a really good example with the reactors, and quite interesting… but I’m still not entirely sure that there isn’t a value of K for which it’s better to have one reactor melting down. Presumably the reactor has some kind of performance metric which varies with temperature (for example, efficiency, power output, safety margins, etc). Being 1K away from optimal might mean only a 0.00001% drop in efficiency relative to optimal, and melting down is really bad. But still there might be some sufficient number K of reactors where their combined decrease in efficiency amounts to civilization-consuming levels of lost power, at which point maybe it’d be better to have one melt down (even given the ensuing chaos) and the remaining K-1 reactors function optimally?

      If being 1K off of the optimal reactor temperature really has no utility penalty at all (like in soccer/football — as long as the ball goes between the goal posts, it’s a goal…), then it really makes zero difference and the big number K has no effect because it’s multiplied by zero.

      That said, I’m not sure that there isn’t some situation which has utility that can be measured in the way you describe. It’s just very hard to think of.

      • SanguineVizier says:

        There definitely is a value for K where it is preferable to have one reactor melting down, even if it is K reactors in scenario 1 (ignore scenario 2 for a moment). Given K reactors in scenario 1, the amount of heat being released into the environment exceeds the ability of the environment to absorb the heat without drastically raising the temperature. It is possible to run out of space to put more reactors before reaching K, but I have not done the calculation to see whether land/resources or thermodynamics is the actual limiting factor.

        I take TeslaCoil’s point, but this was not a great example to make it.

        • JB says:

          Oh, I was assuming these were multiple reactors strung across multiple planets, and K is not limited by available real estate on Earth.

          • SanguineVizier says:

            My argument still applies in that case. Taking the entire universe as the environment to which heat is ultimately rejected from the reactors, there is still some finite value of K scenario 1 reactors that will cause a sufficient amount of heat rejection that in turn causes an unacceptable rise in temperature in the environment, as long as the universe does not have a literally infinite ability to absorb extra heat. There is still a potential resource limitation here, since K is limited by the number of atoms in the universe, and one would probably run into that limit before hitting the limits of the universe’s ability to be a cold reservoir. But TeslaCoil mentioned 3^^^3 scenario 2 reactors as being preferable to a single scenario 3 reactor, which already puts things well above the number of atoms in the known universe, so perhaps we can ignore that limitation for the purposes of the thought experiment. In which case, there definitely exists some value K scenario 1 reactors (and some value L scenario 2 reactors where L < K) that is objectively worse than a single reactor meltdown.

        • RCF says:

          The reactors are in different universes/Everett branches.

    • Murphy says:

      Ok, Why is 1 preferable to 2?

      Presumably there’s a safety issue making it more likely that the number 2 reactor will melt down, or it increases chances of damaging parts or increase the odds of minor accidents or releases of nuclear material, workers being injured by equipment or some other minor bad thing that will happen or may happen or a small chance of a major bad thing happening.

      If your argument relies on the idea that none of those things apply then 1 and 2 are indistinguishable and your multiplier is zero and asking people to choose between 1 and 2 is a fake choice. “choose which is better given this description but now for the rest of the argument all the disadvantages that shaped your choice no longer count.”.

      If one of them does apply then there’s some point when the risk of a bigger accident, the number of curies of nuclear material released, the number of workers killed etc will outweigh the single big disaster.

    • Peter says:

      Let’s invent another straw utilitarian.

      SU: Meh. Much of a muchness to me.
      Other: ???
      SU: Utility is about pleasure and pain, right?
      Other: Well, there’s preference utilitarianism, but I’ll take you at your implicit word that you’re a hedonic utilitarian.
      SU: OK then, so in all three scenarios no-one’s suffering, no-one’s having any pleasure either, it’s all a wash. Nothing of moral interest here at all.
      Other: But but but, nuclear meltdown! Millions suffering long protracted miserable deaths. Communities torn apart, land rendered uninhabitable, billions of dollars that could have been spent on good stuff gone to waste!
      SU: Oh, is that the case is it? Then you should have said!

      One of the troubles of trying to debate utilitarianism is that people (arguing for or against or just trying to explain a point) try to construct “pure” scenarios where one chooses between various distributions of happiness and suffering with no knock-on effects, and, well, I don’t think we ever encounter such scenarios in the real world. If we do, we have no memories of them, because memories are knock-on effects. In other words, just because under utilitarianism happiness and suffering have (positive or negative) intrinsic value, and that value by definition aggregates nicely, it doesn’t mean they don’t have extrinsic (“instrumental”) value too, and there’s nothing that says that that value has to aggregate nicely.

      Let’s continue the discussion.

      Bystander 1: Other – that’s hogwash. The danger of nuclear reactors is grossly inflated. Even Chernobyl wasn’t that bad, wouldn’t have even been that bad if the Chernobyl divers didn’t dive.
      Bystander 2: B1 – hogwash to you too! I don’t know whether you’re being a dupe of the nuclear industry or a vexatiously stupid contrarian, either way, stop it!
      SU: See? As soon as you start having implicit consequences in your moral dilemmas, you get derailed into empirical questions of no philosophical interest.
      B1 & B2, in unison: No philosophical interest? How dare you!
      Other: SU – look, whatever the outcome of their discussion, Scenario 3 is clearly going to have bad consequences however you slice it, Scenario 2 isn’t, it’s not hard.
      SU: You still should have said.
      Other: Who actually talks like you in real life?
      SU: Ah, so we’re in a hypothetical situation are we? I’m not a pure straw utilitarian, I’ve got a rule for situations like this. [ draws gun and shoots Bystander 3, who was minding his own business. ]
      B3 (weakly): what was that for?
      SU: Silly in-joke. Don’t worry, the pain won’t last long, this comment is about to come to an end.

    • RCF says:

      “I’ve always found this strange, since my intuition tells me that utilities can easily be non-commensurable.”

      Your “intuition” has nothing to do with it. “Utility” is defined as being commensurable. If something is non-commensurable, then it’s not utility.

      “Straw Utilitarian: Surely you don’t understand big numbers.”

      Well, clearly you don’t. A meltdown is a stochastic process. There is some point at which the probability of a self-sustaining chain reaction exceeds 50%. This point is called “critical mass”. Below critical mass, the probability of a meltdown quickly goes to nearly zero. But it’s never completely zero. The probability of a chunk of subcritical uranium spontaneously exploding is much, much more than 1 in 3^^^3. If you were to take 3^^^3 nuclear reactors, the number of them spontaneously exploding would be unimaginably large. If you were to just look at how many more spontaneously explode at +1k degrees, versus how many at optimal temperature, that number will still be unimaginably large. In fact, the probability of all the hydrogen atoms in a gram of water fusing into helium, causing a thermonuclear explosion, is probably more than 1 in 3^^^3.

      You really don’t understand large numbers. And/or physics.

      • Anonymous says:

        In fact, the probability of all the hydrogen atoms in a gram of water fusing into helium, causing a thermonuclear explosion, is probably more than 1 in 3^^^3.

        For sure. I guessed some numbers for this and I think the probability of this happening per second is greater than 1 in 10^10^10^10^10^10, which is much much much more likely than 1 in 3^^^3.

        Due to quantum mechanics, any physical process you can imagine (the Moon transmuting to cheese, say) should happen at a rate higher than 1 in 3^^^3 per second, as long as it obeys conservation laws. (And as you point out, the rate of most of these processes should be a strongly increasing function of temperature).

      • suntzuanime says:

        “Utility” is defined as being commensurable. If something is non-commensurable, then it’s not utility.

        This is why I don’t believe in utility.

        • RCF says:

          What do you mean? You don’t believe that people have defined a concept with these properties? You don’t accept it as a useful concept?

          • suntzuanime says:

            I don’t accept it as a meaningful description of human desires. If someone says “my utility is non-commensurable” and you say “utility is by definition commensurable” you’re arguing definitions, not reality. If utility is by definition commensurable we need something different to describe what people actually want.

          • RCF says:

            It’s meaningful as a model. If you reject commensurability, then you’re rejecting that model. If someone’s preferences are completely non-commensurable, then we’re dealing with merely ordinal preferences, and we should abandon quantitative measures, and you should just say “preferences”. If you have different regions that are not inter-commensurable, then it’s more complicated.

      • TeslaCoil says:

        Thanks for all of your feedback.

        The nuclear reactors were merely intended as an example of pure preference, decoupled from human-suffering. I did not mean to imply any physical or human-suffering-related consequences. Unfortunately, it was a horrible choice for that purpose, which detracted from the point I wanted to make.

        Anyway, my intended point, phrased in dust-speck-speech:

        I think one person having a dust-speck in the eye is some non-zero amount of bad (i. e. if I had a choice, I would choose “nothing” happening to a random person over “dustspecks” happening to a random person). I think two people having dust-specks in their eyes is exactly twice as bad, so I am explicitly not scope insensitive. Yet I think that one person suffering torture for 50 years is incommensurably bad.

        My preferences cannot be stated in terms of utility, because utility is defined in a peculiar way. I’d like to know the reasons for defining it like that.

        • Anonymous says:

          I think two people having dust-specks in their eyes is exactly twice as bad, so I am explicitly not scope insensitive. Yet I think that one person suffering torture for 50 years is incommensurably bad.

          OK, here’s a series of choices; which choice do you make in each case?

          (A) Would you rather pay $1, or get a dust speck in your eye every second for the rest of your life?

          (B) Would you rather be tortured for 30 seconds and then given $100 million, or be left alone and given nothing?

          (C) Would you rather one person were tortured for six minutes, or 100,000 people tortured for five minutes?

          (D) Would you rather one person be tortured for seven minutes, or 100,000 people tortured for 6 minutes?

          (E) Would you rather one person be tortured for 50 years, or 100,000 people tortured for 49 years, 364 days, 23 hours, and 59 minutes?

          If you choose the first option in each case–and if you think N people being tortured is N times as bad as 1 person being tortured–we can probably come up with a number of dust specks that outweighs fifty years of torture. If not, in which case do you prefer the second option?

          • TeslaCoil says:

            I choose the first option in each case, and I think N people being tortured is N times as bad as 1 person being tortured

            Yet I maintain that there is no number of dust specks that outweighs fifty years of torture.

            Please try to change my mind! To facilitate this process, I have fixed a utility calculation scheme w.r.t. money, dust specks and torture and uploaded the details to this page. Once our interaction is over, I’ll reveal the password.

          • Deiseach says:

            I think torture is equal and equivalent to torture. I don’t think dust specks are equal and equivalent to torture, unless you’re talking about a zillion gazillion people getting dustspecks in both eyes every minute for the entire rest of their lives.

            One dustspeck in one eye, even one a week, per person of the zillion gazillion does not stack up to the same as one person being shocked every hour on the hour for the entire rest of their lives because pain does not stack up neatly into one huge block of ouchium like that.

          • g says:

            (Reply to TeslaCoil; nesting limit prevents it looking like one.)

            OK, so there are kinda two halves to this, because 50 years’ torture is really large and one dust speck is really small. Let’s begin by trying to convince you that you should prefer 50 years’ torture for one person to something-really-minor for a sufficiently enormous number of people.

            Advance notice, in case you want to give some further thought to your declared utility function ahead of time: some of the torture and dust specks in what follows is probabilistic.

            OK, here we go. The overall approach, of course, is to interpolate from “torture” to “dust specks” by a large number of changes, each of which (roughly) increases the number of victims a lot while reducing the severity a tiny amount. To avoid the conclusion that “torture” is preferable to “dust specks” you must either reject transitivity, or hold that one of those changes actually makes things better rather than worse. (Or, of course, declare that such calculation is unfitting a civilized human being, throw your toys out of the pram, and decline to debate further :-).)

            Define event E(0) to be “one person is tortured for 50 years”. Once we have a definition for event E(k), which will always have the form “N people are tortured for time T”, define E(k+1) to be “1000000*N people are tortured for time 0.999999*T”.

            I conjecture that you prefer E(k) to E(k+1) for each k. The amount of torture per person in event E(22000000) is less than 1 second. If my conjecture at the start of this paragraph is correct, and if your preferences are transitive, then you prefer E(22000000) to E(0).

            OK, now define event F(0) to be the same as E(22000000) but with the time exactly 1 second. (You prefer E(22000000) to this because this is the same but with more torture for everyone.) I’m now going to switch from changing duration to changing probability. So if event F(k) is “N people tortured for 1 second with probability p”, event F(k+1) is “1000000*N people each tortured for 1 second with probability 0.999999*p”. Of course you can translate these probabilities into reduced numbers of people, so I am very confident that you prefer F(k) to F(k+1) for each k. If so, and if your preferences are transitive, then you prefer F(0) to F(42000000), in which the number of people is now greatly increased but the probability of torture for each is less than 1 in a billion billion.

            That’s the first half of the job: Under the plausible assumptions I’ve described above, you should prefer 50 years’ torture for one person to “probability 10^-18 of 1 second’s torture” for each of 10^384000000 people.

            For most people, I think the seeming incommensurability comes from the awfulness of 50 years’ torture more than from the triviality of a single dust-speck in the eye. And personally, I think I would rather have a 1-in-a-billion-billion chance of a second’s torture than a definite dust speck in my eye. If you are like me in this, then we’re already done. But you might be different, so let’s go on.

            Define event G(0) to be a dust speck in the eye every second for 10 years, for each of the 10^384000000 people in event F(42000000). I’m pretty confident you will agree that almost everyone will prefer F(42000000) to G(0). Now, suppose G(k) is “a dust speck in the eye every second with probability p, for N people for 10 years”; let G(k+1) be “a dust speck in the eye every second with probability 0.999999*p, for 1000000*N people for 10 years”. (To be clear, the probabilities are independent for all the dust specks, so when p=1/2 each person gets a dust speck about every 2 seconds, rather than having a 50% chance of a 1Hz stream of dust specks”.)

            I expect you to prefer G(k) to G(k+1) for each k. If so, and if your preferences are transitive, then you prefer G(0) to G(40000000), in which each person’s probability of getting even one dust speck over 10 years is just a little over 1 in a billion.

            Now, in that last scenario each person faces a (tiny but nonzero) chance of getting an awful lot of dust specks, and with the 10^624000000 people we’ve got it’s likely that that will happen a few times. So let’s make a few more changes…

            Event H(0) is the same as G(40000000). And now our events H(k) have the form is “N people get a dust speck every second for 10 years with probability p each time, except that after each dust speck they actually get the time before another can come along is t”. (So when k=0, t = 1 second.) If event H(k) has parameters N, p, t, then event H(k+1) has parameters 1000000*N, p, t + 1 microsecond. (In particular, t never increases by a factor bigger than 1.000001 when k goes up by 1.) I expect you to prefer H(k) to H(k+1) for each k. If so, and if your preferences are transitive, then I expect you to prefer H(0) to H(86400000000), by which time no one gets more than one dust speck per day. (And recall that each person’s probability of getting any dust specks at all is less than 1/30000.)

            And, finally, I expect you to prefer event H(86400000000) to event K, where each of the people involved definitely gets exactly one dust speck.

            OK, let’s recap. E(0) is 50 years’ torture for one person. Trading off time against number of people, we got to F(0), which was 1 second’s torture for an awful lot of people. Trading off probability against number of people, we got to F(42000000), which was a very small probability of 1 second’s torture for an awful lot of people. That, I thought, was obviously preferable to G(0), which was the certainty of getting a dust speck in your eye every second for the same number of people. By trading off probability (of a dust speck each second) against number of people, we got to G(40000000), where an even larger number of people each got a tiny chance of a dust speck each second, each person being very unlikely to get any dust specks at all. By trading off guaranteed holdoffs after getting specked against number of people, we got to H(86400000000), where an absolutely enormous number of people each have that same tiny chance of a dust speck each second, with the further guarantee that the dust specks can’t come faster than one per day. And I thought everyone involved in that scenario would be happy to take those probabilistic specks in preference to certainly getting one speck.

            And of course the final number of people involved is far, far, far less than 3^^^3.

            So, at what point in this progression do you find a step that makes things better rather than worse?

          • TeslaCoil says:

            Thank you for your answer, g! That is a very nice, Morgenstern-esque argument.

            So, at what point in this progression do you find a step that makes things better rather than worse?

            The following agrees with the preference function defined here.

            The first switch occurs at k=10687383. That is, I prefer E(10687384) to E(10687383). I still prefer E(10687384) over E(10687384+k) for all positive k. As expected, I prefer E(22000000) to all F(n).

            I prefer F(42000000) over G(0). In fact, I prefer all F(n) over G(0).

            Since the probabilites are independent for dust specks, I have to calculate with expected time, and therefore I prefer G(6775363) over G(6775362). Beyond that point, every preference is as you expected.


            The basic idea is that dust specks are qualitatively different from torture. Intuitively, most people (especially non-philosophers and non-economists) would agree with this, and so would many of the commenters; so I don’t think this unreasonable. Yet the standard definition of “utility” would not allow for such preferences.

            According to my preferences, torture lasting a long time is also qualitatively different from torture lasting a short time, e. g. torture for 0.1 second is not in any sense a scaled-down version of the “real deal”. Similarly for the case of one dust speck vs dust specks for the rest of your life. The preference switches occur when a variable, such as time, crosses such a qualitative boundary.

            The password is TezlaKoil.

          • Anonymous says:

            @TeslaCoil, OK so there’s some amount of time after which torture becomes infinitely worse (or becomes “really torture”).

            So, concretely, I could have a torture apparatus that directly stimulates the pain nerves in your hand or something. There’s a number T for which you would prefer a 100% probability of T-1 seconds of pain to a 0.0000001% probability of T+1 seconds of pain?

            For me personally I don’t think there’s any such number; everything–my mental state, my emotions, my memories–is a continuous function of T, so my preferences probably should be as well.

            Furthermore I think if you actually try to apply incommensurate values they don’t match up with what we actually want. For example if torture is really infinitely worse than more mundane evils, we should be sacrificing everything (or at least, everything with less-than-infinite value) if we can achieve even tiny reductions in the probabilities that people will be tortured. Similarly placing a literally infinite value on life is inconsistent with activities like driving cars and going swimming. I don’t think people behave as if they had incommensurate values.

          • Adam says:

            Replying to g:

            If you could guarantee me immortality, but the expense is you’d flay and mutilate my hand for 10 hours every year, then regrow it and do it again, for eternity, I’d take that.

            I also think I’d take a permanent dust speck if you could guarantee a 0 probability I’ll ever get tortured in the life I already have. I might even accept completely removing one eye for that.

          • g says:

            (Replying to TeslaCoil.)

            OK, so you have an infinite discontinuous jump in your utility function at 10 hours of torture, and another at 100 hours of dust-specking. I think Anonymous’s question — probing that boundary a bit further — is a pretty good illustration of why that’s even more counterintuitive than preferring TORTURE to DUST SPECKS.

            Or we can look more concretely at the particular case where you first give a “backwards” answer in my series of comparisons. We have some colossal number of people (let’s just call it N) being tortured for just very marginally over 10 hours. Then a genie comes to you and offers to reduce their time of torture by 36 milliseconds, in exchange for which another 999999*N people (who were formerly not going to be tortured at all) will get tortured for just marginally under 10 hours. “Great,” you say, “that sounds like a really good deal. Let’s do that.” Really?

            (Your document mentions that the actual numbers don’t match very well with the numbers that apply in real life. Please feel free to share more realistic numbers and we’ll see what the corresponding reversal looks like then. I don’t think it will be any less awful.)

            I do understand the intuition that says “torture and dust specks are just different kinds of thing, incommensurable with one another”. The point of this sort of exercise is to suggest that taking this idea seriously even in extreme cases is very, very hard to make sense of.

            (Of course in realistic cases where the numbers of people, lengths of time, etc., don’t vary by many orders of magnitude it’s perfectly reasonable to treat them as incommensurable, and there’s obviously nothing wrong with that.)

          • TeslaCoil says:

            (replying to Anonymous and g)

            Thanks again. While I did not change my mind, I think I have gained a better understanding of the “commensurability” point-of-view.

            g wrote: Please feel free to share more realistic numbers and we’ll see what the corresponding reversal looks like then. I don’t think it will be any less awful.

            What I meant was that my real preferences involve many more (possibly infinitely many) categories. For example, I chose the ten hour boundary because it is the approximate upper bound for the “I would do this for money” category.

            Somewhere around that point, my money-to-torture (mutatis mutandis, money-to-dust-speck) conversion ratio becomes unbounded, and I won’t prefer “10+ hours of torture + earn $X” over “nothing happens” for any value of X. I am okay with losing large amounts of money for slight reductions in torture time around the ten hour boundary (In these extremely over-simplified scenario. Really, money utility is strictly not linear, since $10^100 is only marginally more useful than $10^99). This does not mean that I should be unable to decide between “11 hours torture + get $100” vs. “11 hours of torture + get $200”.

            I have other category boundaries for ever smaller amounts of torture, e. g. torture on the order of “1 second” is categorically different from torture on the order of “1 hour”, and so on.

            When (expected) torture times start reaching nanosecond levels, I don’t think these reversals look awful at all.

            Anonymous wrote: There’s a number T for which you would prefer a 100% probability of T-1 seconds of pain to a 0.0000001% probability of T+1 seconds of pain?

            Absolutely, as long as you replace seconds with units, and T is sufficiently small.

          • g says:

            (Reply to TeslaCoil.)

            I’m not sure whether I’m correctly understanding your description of how your actual preferences differ from the deliberately oversimplified ones in your document. Having more (incommensurable?) category boundaries seems to me to make the difficulties worse rather than better.

            You say: “When (expected) torture times start reaching nanosecond levels, I don’t think these reversals look awful at all”. That might be true, but we can get stepwise from TORTURE to DUST SPECKS without going via torture times anywhere near as short as nanoseconds. In fact, we don’t need any times shorter than one second. So if you endorse a principle like “for any time T between 1 second and 50 years, I would rather 1 person be tortured for time T than 1000000 people be tortured for time (T minus 1 microsecond)”, and if your preferences are transitive, and if you don’t care vastly less about one person’s suffering merely because lots of other people are suffering in similar ways, then you have to prefer 50 years’ torture for one person to 1 second’s torture for some suitably enormous number of people.

            (And then we can get from “1 second” to “1 second with tiny probability”, etc., etc., etc.)

            Are you willing to bite this bullet and declare that actually there is some T (greater than 1s) for which you prefer 1000000 times as many people to suffer, over an extra microsecond of suffering for each? If so, can you tell us what T, or are you merely asserting that you’re sure there’s some such T but you have no idea what it is?

            (If the category boundary you say you have somewhere between 1 second and 1 hour is another of these infinite incommensurable ones, then we don’t need to go all the way between 50 years and 1 second; we can just go between 1 hour and 1 second. In that case the number of people involved doesn’t need to vary by quite such an outrageous factor. As I said before: having more category boundaries makes this sort of position harder, not easier, to swallow.)

          • TeslaCoil says:

            Are you willing to bite this bullet and declare that actually there is some T (greater than 1s) for which you prefer 1000000 times as many people to suffer, over an extra microsecond of suffering for each?

            I have such a boundary somewhere around T=10h, as we’ve previously discussed.

            I still don’t understand why you find this strange. You already know that I would accept dust specks for any finite number of people over one person getting tortured for 50 years. This is exactly the same thing, merely replacing “dust specks” with “short torture”.

          • g says:

            Oh, I’m sorry, it wasn’t clear to me that you actually do have a boundary of that kind near 10h.

            Your last paragraph makes me wonder whether you’re one of those perfect reasoners one finds in logic puzzles, to whom every consequence of everything is immediately apparent. The equivalence between the (extremely intuitive and natural-feeling, to almost everyone) preference for any number of dust specks over 50 years’ torture for one person, and the (extremely counterintuitive and unnatural-feeling, at least to me but I suspect to almost everyone) preference for 1000000*N people suffering 10 hours of torture over N people suffering 10 hours + 1 microsecond of torture, is … not something most people find obvious. Hence (what I thought was) the value in giving an extremely explicit and careful demonstration of how to get from one to the other.

            I’m curious, though, as to what you think that sudden infinite discontinuity feels to someone experiencing it. For example: suppose some evil mad scientist prepares a few billion copies of you, and subjects them to torture of duration varying from 9 hours to 11 hours in microsecond intervals, and then asks each how they found it — do you expect some of them to give infinitely more anguished answers than others, with a breakpoint whose sharpness is measured in fractions of a microsecond? Or suppose it’s just one of you; do you think that at some point about 10 hours in something abruptly happens that’s literally infinitely worse than everything you’ve undergone before then?

            Let me attempt to steelman your position a bit: perhaps what happens is that any given torture either does or doesn’t permanently break someone’s mind, and the difference between those two is of infinite importance, and when you take gazillions of people and subject them to gradually longer torture, at (numerous) various points one extra person gets broken, and each of those makes things worse in a more important way than the mere suffering undergone by the others.

            Not obviously crazy. But still crazy, I think, because actually the breaking of minds is a gradual matter, and once again there is a continuum between “perfectly unimpaired” and “permanently disastrously broken”, and the same sort of argument as before can be applied.

        • RCF says:

          “My preferences cannot be stated in terms of utility, because utility is defined in a peculiar way. I’d like to know the reasons for defining it like that.”

          Utility is defined in that way because it’s the only way to make everything comparable. If you want things to not be comparable, then yes, you need something other than standard utility.

          • TeslaCoil says:

            – What are your reasons for including the condition of commensurability?
            – My reason is that it’s the only way to achieve commensurability.

            This is a non-answer. Why do you think commensurability is so important?

    • Adam says:

      Why not just use the dust specks as an example? Some types of suffering, say minor inconvenience and torture endurance, are categorically different, and scaling minor inconvenience to some unimaginably large number of people, even BusyBeaver(googolplex^^googolplex), doesn’t make minor inconvenience more of a moral problem than torture endurance.

      At some point, if we don’t accept that certain types of utility are incommensurate, how do we ever end up coming to any moral conclusion other than we should devote all resources to engineering a planet with extremely specialized sentient creatures that consume only non-sentient food and don’t compete for any resources at all? Really, we should be trying to create sentient solar-powered machines while bringing about the extinction of all animals, humans included, which then spread to the remainder of the reachable universe and do the same thing everywhere else life exists.

      Of course, maybe that’s the conclusion we should be coming to, and AI that kills us is a good thing, provided it does it quickly and doesn’t unnecessarily torture us first.

      • TeslaCoil says:

        Thanks for phrasing all this in a much better way than I would have been capable of.

        Why not just use the dust specks as an example?

        In hindsight, I should have.

      • g says:

        Really, we should be trying to create sentient solar-powered machines […]

        If this is intended to follow from “things as different as torture and dust specks are not absolutely incommensurable” then I think there may be some steps missing in your reasoning.

        In particular, the “torture is better than dust specks” argument only purports to work in the special case where the number of people involved can become unthinkably vast. One can endorse that position while also holding that for “small” numbers of people — say, no more than a billion billion billion — torture is always much worse than dust specks. That seems like enough incommensurability for practical purposes.

        If you try to go from “utilities are commensurable” to “might as well replace the human race with barely-sentient machines”, I think you’re going to need some assumption akin to “we can make more than 10^2000 of these machines”, at which point the argument breaks down because actually there is probably no way for us ever to make more than, say, 10^120 of *anything* because the observable universe doesn’t have enough space or enough stuff in it.

        But maybe I’m misunderstanding how your argument works?

        • Adam says:

          No, you’re correct. I wasn’t making the complete argument. I think I was actually incorrectly stating it as a conclusion following from the commensurability of all sources of utility. It’s more a conclusion of utility maximization as the only ethical value. In practice, even committed utilitarians seem to value things like human utility over machine utility, and think we should aim for a universe in which humans continue to exist, even if human existence is more painful and less blissful than a future universe in which machine sentience is the only form of sentience. And they needn’t be only barely sentient. They can have human levels of sentience but simply not experience pain, fear, regret, never set goals they can’t achieve, whatever it is, that makes them better utility maximizers than humans.

          Similarly, maybe the distress experienced due to dust specks and due to torture can be made commensurate through a normal utility calculation, but the utility loss from torture nonetheless has a different ethical value than the utility loss from having a speck of dust in your eye.

          That is, if for whatever reason, we think things like a universe full of human brains in dopamine drip vats controlled by a super-AI is bad even though it has a greater aggregate utility than a universe of humans in bodies who sometimes hurt each other, kill each other, experience tremendous loss, disappointment, and failure, then maybe we can accept that a universe with no torture, but unimaginably many dust specks, is a better universe than one with no dust specks but torture, even minimal torture.

  18. Buck says:

    I still think that the suffering of chickens outweighs that of humans, because there are three times as many of them and they suffer way more intensely than almost any human alive. Even if I only had a 10% credence in weighing their suffering 50% as much as humans, I’m pretty sure that would outweigh human suffering.

    And preventing chicken suffering is a gazillion times cheaper: my girlfriend mentioned today that by ACE’s admittedly optimistic estimates, she’s prevented more deaths with her donations last year than AMF has in its entire existence. I would love to see your response to this sometime.

    You say at the end that you don’t like how in the extreme case, the Senate is valuing Delawarians (?) 10x as much as New Yorkers. I guess that that’s how I feel about morality compared to you: you always seem much less in crisis mode from the suffering in the world than I am. So I find it distasteful to use Senate-style reasoning in real life at all, but you’re okay with it.

    • Orborde says:

      What is “ACE” in this context? I might care about animal welfare, and I’d like to look them up.

    • Nestor says:

      Once we solve the sufering of chickens by perfecting vat grown chicken meat and eggs, paradoxically we remove the economic incentive to breed chickens in fantastical numbers, so chickens are reduced to anecdotal populations in hobby farms, with most extremely optimized breeds dying out altogether.

      I think this is far preferrable but I believe some utilitarians might disagree…?

      • Buck says:

        Most utilitarians either agree or think that animals don’t matter. I agree with you for sure.

      • I can’t speak for the utilitarians, but if you’re a species preservation consequentialist like I am, the dying-out part looks morally suboptimal at first glance. If the point is survival, large numbers are not intrinsically important however, so the first would be morally fine.

    • Leonard says:

      You’re in “crisis mode” over chickens? You cannot do anything significant about chicken suffering. Isn’t it stressful and yet pointless to be in crisis mode? I don’t get it. How do you enjoy life?

      • Wrong Species says:

        “You’re in “crisis mode” over slavery? You cannot do anything significant about slave suffering. Isn’t it stressful and yet pointless to be in crisis mode? I don’t get it. How do you enjoy life?”

        • Anonymous says:

          That’s a great comparison. The Quakers decided in the late seventeenth century that slavery was the most pressing temporal issue. They proceeded to spend a century enjoying themselves and taking over the world before they found themselves in a position to act.

        • Leonard says:

          There is slavery right now in many countries. Bangladesh. Mauritania. China. Perhaps it exists among the three-headed N’yyurrlap’yy of Titan. And yet I do not feel any of these is a crisis for me. If Chinese slaves are to be freed, I feel that that is a job for someone else who is much closer to the thing to deal with. Presumably, China’s government? China’s people? Furthermore, I think that crisis mode in response is entirely unproductive, since I am willing to do nothing about these problems — slaves! oh my! — and can do practically nothing, and if I did do anything noticeable it would probably worsen the problem.

          So, I enjoy life by keeping what I regard as proper, humble perspective on how powerful I am, what my responsibilities are, and to whom. Distant others have much better keepers than me. I don’t worry about them. I feel no crisis. The slaves of Titan will just have to free themselves.

          So, I have answered your question. Maybe you can answer mine. I still don’t understand how one can be in a “crisis” mode, about something one can do nothing significant about.

          • Buck says:

            I think that it’s easier to help chickens than you think, for example by donating to Mercy for Animals or The Humane League or ACE’s other top charities.

            Chicken suffering is a cheaper problem to solve than human suffering by far.

            There are some other problems, like wild animal suffering, which I don’t feel as personally upset by, because I can do less about them.

      • Anonymous says:

        You cannot do anything significant about chicken suffering.

        Sure you can. I think it’s potentially a crisis because there’s so much you *can* do for chickens. If you spend most of your money on animal welfare you can probably rescue a *lot* of chickens.

        • houseboatonstyx says:

          If you spend most of your money on animal welfare you can probably rescue a *lot* of chickens.

          Enough money might get the factory farming industry shut down, which would prevent a lot of chickens being brought into the world in a physical condition (designed for factory farming) in which they would be unhappy. Ie, factory farm breeds would become extinct, leaving only wild chickens (happy in the tropics) and backyard chickens with (hopefully) kind owners and pleasant conditions. Worth doing, but I question whether ‘rescue’ is the right word. Though maybe there isn’t yet a neat term better than ‘rescue’, even if ‘rescue’ does send allies like me quibbling.

        • Leonard says:

          If you spend most of your money on animal welfare you can probably rescue a *lot* of chickens.

          Probably not.

          There are tens of billions of chickens. If I can “rescue” (whatever that means) one of them for a penny, then I can spend my entire life’s savings and dent the number not even by 1%.

          Bill Gates could move the needle, but that would not affect the underlying “problem”, which is that (1) people have disposable income, (2) people love the taste of meat, and (3) chickens are about the cheapest way we have to produce meat. If Gates cornered the market on chickens, “rescuing” tens of billions of them, what would happen is chicken farmers would breed more chickens. Duh! There’s a market here; supply rises to meet demand.

          • g says:

            I can spend my entire life’s savings and dent the number not even by 1%.

            Why is the percentage what you should care about?

            Imagine there is a child drowning in a pond ten metres away from you. You can easily save them. I guess you do. Congratulations, you just did something valuable. Now, suppose that as you are running towards the pond someone calls out “Stop! There are millions of other children drowning in other ponds!” Does this make any difference to your willingness to save this one?

            For me, it makes very little difference. Sure, it would be nice to be able to feel “I have solved a large fraction of this problem”, but in terms of actual good done you do as much good however large the problem as a whole is.

            supply rises to meet demand

            That is a much more serious objection, but surely all it means is that any kind of “rescuing” to which that would be the reaction is a bad approach, and that if you actually want to improve things for chickens you should aim for things like reduced demand for chicken meat and better treatment of chickens by farmers. I think these are in fact what most people hoping to improve things for chickens are aiming at.

    • Jiro says:

      The problem with this reasoning is that the “10% credence” speculation is something that you just made up. People, when asked to estimate low numbers, don’t say “0.00000001%”. Just the fact that they can even think about the unlikely situation will lead them to name a number of at least a certain size (typically at least 5 or 10 percent). If you then try to do calculations with that number, you can get ludicrous results.

  19. The Anonymouse says:

    Everyone loves the idea of pure democracy… until they count up how many of their heartfelt views are not shared by 51% of their fellows.*

    *Or, updated: … until they realize that their Facebook friends are not a representative sample of the voting public.

    • suntzuanime says:

      But how many of their heartfelt views are shared by 51% of the Senate? Yeah, most people would prefer a dictatorship that shared their views to a pure democracy, but that’s not on offer for most people.

    • Anthony says:

      Worse: until they realize that their facebook friends might actually be representative of (at least 51% of) the voting public…

  20. psychorecycled says:

    I am working on a project which uses computer vision to detect defecation/urination in cows: if you see a cow go to the bathroom, you trigger a machine to give it a reward and make a noise so all of the other cows know someone did a good thing. (The cows have RFID tags in their ears so the machine which gives out the rewards knows not to reward the first cow that comes up.) The goal is to teach cows to only poop in certain places, instead of all over the place, which allows you to house cows in a manner which will make them happier, and to harvest the poop efficiently.

    I feel very good about this job from a moral standpoint because there are a lot of cows with terrible lives.

    This is only related to this post insofar as Scott talked about the moral weight of cows and chickens, but I wanted to share anyway.

    • Deiseach says:

      I can’t decide if you’re serious or if this is a wind-up.

      Slatted housing.

      • PsychoRecycled says:

        It’s actually totally serious. Slatted housing works great…but you wouldn’t want to lie down on it, and cows sleep lying down. At least, they like to. (Lying down is hard for cows at the best of times, actually, especially when they’re pregnant. A cow which has recently given birth which lies down might actually die: they aren’t strong enough to get back up, and the methods of getting them back up cause them enough stress that you’ve definitely taken some time off their lives.)

        Cows don’t have much of a sense of not shitting where they sleep. You solve this problem (currently) by changing the bedding frequently, pumping the cows full of antibiotics so it isn’t a big deal that they’re surrounded by feces, and housing the cows in such a manner that their shit can’t get everywhere.

        There is a growing movement to be nicer to cows: dairy cows have fairly terrible lives right now. The current method of raising cows starts with immediately removing them from their mothers. When they encounter their first other cow three months later, they panic and attack it. Cows which haven’t been socialized don’t learn very well, and are generally screwed up for life. This actually helps with that.

    • houseboatonstyx says:

      Actually someone I knew quite well invented a bolus that would stay inside the cow and beam various information to a nearby scanning device. Including notification when some particular muscle movement occurred. (This was for free range cows, because he did not approve of other situations.)

      • PsychoRecycled says:

        Precision dairy farming is a real and serious thing. They have conferences. There is a reasonable amount of money which goes into the field. You can get a degree. It’s pretty neat.

  21. Harald K says:

    About the casualties in the Iraq war: there has been a decade-long effort to downplay the number of deaths, and use the documented numbers (in the 100000 range) rather than epidemiological estimates – by the government, but also by some journalists who sneer at statistics.

    Early blogger Tim Lambert of Deltoid covered it well when it happened. Were you around when that guy blogged? Great on public statistics, astroturfing, war on science stuff. He’s posted nothing but open threads for many years, though 🙁

    • Alsadius says:

      It’s not a “war on science” to disregard one study that’s a massive outlier in favour of many other independent sources of data that come to very similar conclusions.

      • Harald K says:

        You cannot compare a study that can only possibly determine an absolute lower bound (Iraq body count, counting only media-reported deaths) over epidemiological studies (Lancet, ORB, PLoS). The study I assume you mean is the outlier, ORB, used the same methodology which has been used to estimate excess mortality in other conflicts – strangely enough, none of those were controversial.

        Even the epidemiological studies are likely undercounts, since the fewer surviving members a family has, the less likely it is to be sampled. At the most extreme, households entirely wiped out are not possible to survey.

        “War on science” was in reference to the whole of Tim Lambert’s blogging. He looked at all sorts of politically motivated anti-science efforts, and very often it was the same usual suspects behind it. I recommend the book “Merchants of Doubt” by Naomi Oreskes to get an overview.

        (I’ve also heard Chris Mooney’s “The Republican War on Science” is good, but I haven’t read that so I can’t recommend it).

        By the way… I’m moving. Every book that’s in my bookshelf is a hassle, and I already try to make a habit of giving away books that are better the more people have read them. Anyone want a fine hardcover copy of “Merchants of Doubt”, for free? You need to be comfortable giving me your mailing address, and I’d appreciate it if you’re an established commenter so I can feel a little confident it won’t go straight in the trash. But I’ll pay for shipping.

        • Deiseach says:

          Personal observation: I can’t take seriously anything with titles that rely on “The X War on Y”. Not even if it’s a title that appeals to my prejudices.

          War On Science, War On Women, War On War – No. Especially when you’re writing from a position of “That tribe that is not my tribe” waging the war. I instinctively (and perhaps erroneously) distrust your good faith and that you’re going to give me reasonably unbiased and unfiltered data rather than cherry-picked talking points about “See how evil the Evil Others are?”

          • Harald K says:

            Agree it’s a cliché. But it’s an interesting phrase: I believe it gained currency with Nixon’s “War on drugs”, but the time when “war on” was a good metaphor for “heroic effort and refusal to stand down or be lenient in the face of…” etc. couldn’t have lasted longer than five minutes. The last 40 years, almost no one seems to have taken the words “war on drugs” in their mouth unless the purpose was to condemn it.

            Anyway, first impressions aren’t always reliable. I followed a lot of the stuff they (Tim Lambert, John Quiggin, Chris Mooney) wrote about, and sensationalistic title aside, there really is a deliberate effort to spread misinformation here that isn’t well appreciated.

            Do you want the book, Deiseach? The one that does NOT have “war on” in its title?Shipping to Ireland is cheaper too 🙂 I can be reached at (my first name before the @) if you’re interested.

          • Douglas Knight says:

            Nixon was involved in both the War on Drugs and the War on Cancer. Reagan renewed the War on Drugs. There was an earlier war on cancer that I did not know about.

          • Richard says:

            I’m OK with “War on X” as long as it actually is. War is not when you try to police inner city neighbourhoods more and increase the jail time for drug offences. War is when you send in the US Marines and Apache helicopters to take out the opposition.

            During the Reagan years, I seem to remember “The War on Drugs” involved sending armed forces to Bolivia which gave the term a smidgeon of credibility.

        • g says:

          I would be happy to have a copy of “Merchants of Doubt”, and would not put it in the trash. But (1) I have hundreds of books on my to-read shelves already, (2) I have no trouble affording to buy the books I read, and (3) I think my opinions on this topic are already fairly close to Oreske’s. You may therefore prefer to send your copy to someone who will read it sooner, is less able to buy it for themselves, and/or is more likely to have their mind changed by it. (You may also be influenced by shipping costs. I’m in the UK.)

          I’m not a very frequent commenter but I’ve been around since SSC began.

  22. Alex says:

    Given my libertarian-anarchist tendencies, any system that makes it harder to create new laws has an intrinsic advantage. Now if we could have laws come with an attached expiration date…

    • DavidS says:

      In the UK, we kinda do. Most new legislation has a ‘sunset clause’ which means it automatically stops being law unless Parliament re-affirms it explicitly.

      • Schmendrick says:

        The U.S. frequently does this with laws as well. However our big programmatic laws usually take the form of open-ended mandates for the executive branch to do something about the problem at hand (which usually results in the creation of a new bureaucratic apparatus). When the sunset clause kicks in ten years or so down the road, the institutional inertia of the bureaucracy coupled with attack campaigns from the law’s ideological supporters (“Senator X wants to undo all the progress we’ve made on issue Y! He’s a stooge in the pocket of the [Y] industry who probably kicks puppies!”) means that making any change in the law or allowing it to expire would cost a disproportionate amount of political capital. Because of this, most politicians who don’t have issue Y as their pet cause just vote to extend the laws in question on the theory that we’ll find some way to muddle through.

        • onyomi says:

          Even laws with built-in deadlines are usually eternal in the US. See, e. g., the “debt ceiling.” There will always be a fearmongering campaign to get it renewed, if it is not just quietly renewed with little fanfare.

          And stuff like defense contracts are actually intentionally designed to spread benefits around so that the number of constituencies who will be pissed off can be maximized should there ever be a push to stop the money flowing.

          • Alex says:

            Which at least has the nice side effect of everyone spending time discussing that renewal and not writing more freedom-reducing laws. I’ll take every little bit I can given that we’re nowhere near a place where we can push a reset button to revoke all laws but the Constitution.

          • RCF says:

            The debt ceiling isn’t renewed, it’s increased.

    • Harald K says:

      So you think it should be difficult for politicians to correct their errors? You’re willing to prefer present injustice over the risk of future injustice?

      You realize that you’re making it all the much harder to end e.g. farming subsidies?

      Either way, that would be just your preference. I don’t see why a system with a built-in bias towards your preferences is a good system. I’d want a system with as few preferences as possible built in, including my own.

      • Irrelevant says:

        It should be easy to abolish laws and hard to pass them. What’s unclear about that?

        • Harald K says:

          That still doesn’t explain why you think the system should be biased in favor of people like you who think that laws are usually bad.

          I think the question of how many laws are too many, is itself a question which must be settled on a level playing field. Trying to lock in place some answer, I see as claiming power over those who come after you for no good reason. Why do you think you know better than them?

          • Irrelevant says:

            First, that’s an excellent choice of metaphor! On a level playing field, it takes more force to stop someone than to move, and that force must be kept up constantly. Let’s make the legislative system like that.

            Second, anyone who does not agree that laws are at best necessary evils desires to enslave what little of humanity they do not desire to murder, and should be kept as far from the levers of power as possible.

          • onyomi says:

            Yes, the presumption should not be evenly balanced between freedom and restriction. If you’re going to force me to do something against my will, or to not use my own body and property in the way I wish, then you owe me a compelling reason. I, on the other hand, do not need to justify my right to do everything I do with my own body and my own property.

          • Alex says:

            What Irrelevant says, laws should be hard to pass and easy to abolish. I’m saving that in the quotes archive.

            Re: Farming subsidies. If they had to be renewed every year by now we probably have enough support against them to let them expire.

            As to the number of laws that is optimal I agree on the level playing field. The natural state is no laws at all. Now for every single law that you want passed please justify it, preferably without the threat of force. Sorry, you can’t use any other laws or the status quo as justification as this is currently biased in favor of new laws and is not level at all.

          • Harald K says:

            It’s not a question of justifying laws or not. Obviously you need to have a good justification to convince 50% of the people that a new law is a good idea.

            If you think laws should be easy to repeal and hard to pass, well, build that in to your own voting. For the vast majority of restrictions on freedom that could be proposed, I’d vote against, too. But precisely therefore, we don’t need any built in structural bias. A built in structural bias can only make our decisions worse, not better.

            This is a special case of a more general phenomenon about correcting for bias. It’s not as easy as you think.

            Story time: I followed the computer Go mailing list at the time they invented Monte-carlo tree search. Programs went from being pathetic to being on par with the strongest club players.

            But there were some things that disturbed the developers: The computers played pretty alien. In the endgame, they would casually throw away points, and usually end up winning with 0.5 point. They sucked at handicap games, because they would throw away their advantage, thinking they had a solidly won position.

            The issue was that these bots were based on statistics, unlike chess bots. They maximized ONLY the winning chance, they had no concept of “locking in” points or other tactical advantage. In solidly won positions or solidly lost ones, they simply couldn’t tell good moves from bad – a human would see that this move throws away five points, and that move locks in five points, but a computer would only see that they win (or lose) anyway.

            So some developers got the idea of a built in bias, of overcompensating. Aim for a slightly higher score that you need to win in the endgame, to avoid throwing away points. More points is good, in case you’ve misjudged something somewhere, right? Can you guess how that went?

            It made the bots weaker. True, it made them better at holding onto point in the endgame – but it also made them unnecessarily desperate when the game was actually equal. The bot has its own reasons for thinking the game is won or not. Although the bot may be wrong, although it may be systematically wrong in one direction, unless you can explain to the bot why it’s biased, it won’t be able to play better from that information.

            They tried for years to fix this problem, and every attempt only made the bots weaker. In the end they settled for some careful approaches that only made the bots a little weaker in general, but made them play much more human-like (and slightly better at handicap games – a prior assumption that the opponent will catch up 20 points by the distant future of the game, is less destructive than an assumption that it will catch up two points in the closer parts of the game space where the bot has solidly convinced itself it can’t).

            So you see, if I systematically make the wrong choices regarding which new laws should pass and not, your task is to explain to me why, and when I make those wrong choices. If you try to build in a crude correction, you’ll be much better at sabotaging my correct decisions than fixing my bad ones.

          • Lupis42 says:

            But precisely therefore, we don’t need any built in structural bias. A built in structural bias can only make our decisions worse, not better.

            Why is “laws have ontological inertia” not structural bias, while “laws do not have ontological inertia” is a structural bias?

            Requiring a 70% majority to pass a law but a 51% majority to repeal would be a structural bias.
            A system where all codes had to be reauthorized approximately once every other election cycle should not be a structural bias.

          • Lupis42 says:

            Trying to lock in place some answer, I see as claiming power over those who come after you for no good reason.

            Every law that stays on the books for more than one election cycle meets this description.

        • Jaskologist says:

          Really, there aren’t enough camerals. We need a House of Repeals, too. Give it proportional representation just to mix things up more, and let them strike down any legislative or judicial law rule with 60% of their vote.

          • Jaskologist says:

            I am fine with the filibuster. Why wouldn’t I be?

          • Harald K says:

            It rewards disproportionately those who would sabotage everyone unless they get their way – holding the world hostage. What kind of world do you live in where brinkmansship is good?

            I see avoiding brinkmansship and expensive competition (of the all-pay auction type) as one of the main purposes of government.

          • Lupis42 says:

            It rewards disproportionately those who would sabotage everyone unless they get their way – holding the world hostage. What kind of world do you live in where brinkmansship is good?

            That only holds if the status quo is unacceptable, which is almost never the case.

            I see avoiding brinkmansship and expensive competition (of the all-pay auction type) as one of the main purposes of government.

            By that standard government is a failure, and should be abandoned.

          • Jaskologist says:

            This is overwrought. The world is not held hostage because you didn’t get to pass your bill. You just don’t get to pass that bill, that’s all. This is not a disaster; if the bill really is that essential, you should be able to find 60 people who are at least willing not to block it, even if they won’t vote for it.

            If the bill is super-essential and you still can’t get 60/100 people on board, your real problem is much much deeper, and this is merely a symptom.

          • onyomi says:

            Yes, all that “hostage taking” rhetoric surrounding the debt ceiling made me want to puke.

          • g says:

            Jaskologist( and onyomi), I think there are a lot of people who would say that the debt ceiling brinksmanship was a sign of a “much deeper” problem, and held (or at least stated) views of the Republican Party perfectly consistent with its being impossible to find 10 of them (or whatever the number was) prepared to give up short-term political gain in order not to wreck the US economy.

            Whether they were right about that is another matter entirely, but I don’t think their position was inconsistent in the way you’re suggesting.

          • Harald K says:

            By that standard government is a failure, and should be abandoned.

            Much as the US senate sucks as an institution, there’s still a hell of a lot more brinkmansship going on between your average warring clans and warlords. Some government ought to be abandoned for better government – one measure of better being that it lets us come to agreements without threatening scorched earth and salted fields.

            The world is not held hostage because you didn’t get to pass your bill.

            That depends on the bill. I know lots of people in here are climate deniers, so how about a less contentious topic, the neurological effects of lead? Take the U.S. Clean Air Act that banned leaded gasoline in 1996. If you could get 40% to coordinate to block that bill, you could win a lot of concessions from tender-hearted utilitarians who actually cared about US kids getting brain damage from lead exposure.

            The psychopaths are rewarded. If better government is to mean anything at all, it should at least involve not being rewarded for being a madman.

          • suntzuanime says:

            Better government, if it means anything at all, means whatever gives us better government. If the psychopaths provide us good policy, there’s no need to moralize over whether the sausagemaking process rewards “bad” behavior somehow. Arguably you might not expect the policies supported by the most skillful hostage-takers to necessarily be better than average, but they’re not bad by definition.

          • Harald K says:

            You can expect their policies to be a lot worse than average. If they were at all reasonable, they wouldn’t have to take hostages to get their way.

          • suntzuanime says:

            That doesn’t strike me as remotely true. Think of all the various people who are at all reasonable, then count up the fraction who get their way without taking hostages.

          • Lupis42 says:

            Much as the US senate sucks as an institution, there’s still a hell of a lot more brinkmansship going on between your average warring clans and warlords.

            Are you saying that warlords and clans aren’t forms of government?
            I’m not trying to claim that abolishing government is practical, just that if the standard is ‘brinksmanship’, governments of all types throughout history have been causes, not preventatives.

            If you could get 40% to coordinate to block that bill, you could win a lot of concessions from tender-hearted utilitarians who actually cared about US kids getting brain damage from lead exposure.

            If there were any terderhearted utilitarians there, they should have been blocking the AWB in 1994, or the Patriot Act in 2004, or the AUMF.
            Your argument still depends on the assumption that *doing nothing* is catastrophic hostage taking. I would argue that *something must be done* is a much more dangerous default than *nothing need be done*.

          • onyomi says:

            @ G, it was question begging. I, and I’m sure many Republicans, did not accept the premise that failure to extend the debt ceiling would “wreck” the US economy. I had similar feelings about the “shutdown.” They shut down all “non-essential” services. My question is, why is the government doing things which aren’t essential in the first place?

          • g says:

            onyomi, disagreeing with the premise that not raising the debt ceiling would have a disastrous effect on the US economy is not the same thing as saying that the people who claimed it would were insincere or inconsistent. I was commenting on the latter, not on the former.

            I spend a lot of resources on things that aren’t essential. So does my employer. So does just about any organization. I don’t see any reason why governments, alone among human institutions, should be doing only strictly essential things.

          • onyomi says:

            Because government, alone among human institutions, claims the authority to initiate force.

            If you are in a lifeboat and it’s taking on water, and no one will bail out the water, it’s arguably justifiable to point a gun at them and say “bail!” as that will save everyone’s life. This is a plausible justification for why governmental authority should be allowed to exist at all, assuming you think it is necessary to prevent catastrophic consequences. It is not permissible to then point the gun at people and say “make me a sandwich!” In other words, using force, which is what government law is, to accomplish non-essential goals is immoral.

          • g says:

            So, if I may make what I think is your argument more explicit to check I understand it, it goes thus: Everything a government does, it funds with money raised by taxation. Taxation is (ultimately) taken by threat of physical force. We should make as little use as we can of such threats; therefore government should be minimal.

            I think this is all wrong, I’m afraid.

            Taxation is not the only thing that ultimately rests on the threat of physical force. For example, suppose I am hungry and would like to eat the food in someone else’s house. I can’t — it’s their property, and taking it would be stealing (and burglary, if I break into their house to get it), and my practical difficulty in doing so rests (in the last resort) on the threat of physical force by the owner, the police, and the prison service.

            Notice what this means. If I am not to risk starvation, I had better find ways to get property of my own. In short, I had better work. If I don’t, I may starve. So my employer’s use of my time, like the government’s use of my taxes, is ultimately founded on the threat of physical force: work, or starve, or get thrown into jail. So is my employer also obliged not to have me do anything that isn’t strictly necessary?

            (In most civilized nations these days, the threat of actual starvation isn’t too severe, mostly because their governments provide a safety net. Which is paid for using tax revenues. Which ultimately depend on the threat of physical force.)

            But, in fact, most of the time neither labour nor tax is exactly by the direct threat of violence. No doubt that’s partly because everyone knows that it could come down to violence in the end, and knows who would win. But I don’t think that’s the whole story. No one exactly enjoys paying taxes, but most of us pay them fairly willingly; they are the price we pay for a society in which people mostly aren’t left to starve if they’re out of work, in which people and property are protected by a reasonably effective police force, in which there are tolerably maintained roads to get around on, etc. No one gets elected promising to do away with income tax. And most of us are content with the existence of private property, even though it stops us just taking whatever we want, because it also enables us to have things that other people can’t just seize, and it seems to lead to a more peaceful society than one where Might Makes Right is the whole of the law. And most of us don’t really mind working for a living all that much; after all, we get paid, and some of us even have quite interesting jobs.

            So even though the possibility of physical violence is there in the background, speaking for myself I don’t think it’s mostly what motivates me to do anything I do; and it’s there (likewise distantly) in the background of so many different things that I don’t see any reason to single out government as specially in need of minimalism.

            I’m a consequentialist, and it seems to me that if we want to decide how much a government should be doing we need to look at the actual consequences of the possible answers. That’s really difficult (a nation is a very complicated system) but I see no reason to expect the answer to be that governments should do no more than the barest minimum necessary.

            If you’re in that lifeboat, and decide that gun-toting is obnoxious and no one should be required to bail any more than the minimum consistent with survival, then what happens is one of two things. Either the strongest and/or meanest people do that minimum and everyone else has to (ahahaha) bail them out by working extra-hard, sacrificing a bunch of people to the freeloaders for no net benefit; or else everyone does only the minimum that would suffice if everyone else bailed their hardest, and the boat sinks. Given that you need coercion to get anyone to bail, there is no reason to expect that the optimal level of coercion is also the minimal possible level, and if someone argues too vehemently that it must be I can’t help suspecting that they’re hoping to be one of the ones doing two minutes of bailing per day while the less fortunate exhaust themselves.

            (In one sort of ideal world, everyone would recognize the symmetry of the situation, do some kind of veil-of-ignorance thing, and agree that everyone will do roughly equal amounts of bailing, or amounts depending on how good they are at it, or something. In another, everyone would engage in some kind of market-based bargaining and end up with an optimal allocation of bailing given their abilities, preferences, etc. Unfortunately, the nearest we get to either is a crude approximation and we probably are all better off with the gun-toting government in place.)

          • Nornagest says:

            If you’re arguing that more institutions than the government rest on a threat of physical force, and your argument includes a step where the government threatens physical force, I must confess I find it rather unconvincing.

            I’m not Onyomi, but my take on it is rather simple: government action carries costs, and those costs, if not directly paid by their beneficiaries, rise in proportion to the coercion involved in their funding or their implementation (mainly for defense and law enforcement).

            It may be that we value what a government provides enough to make up for those costs in some cases; I’m not a strict libertarian. But there is a tendency in policy discussion and even political theory to ignore or minimize those costs unless they cross some ethical boundary and consider only the benefits; leftists are more inclined to minimize financial factors, rightists enforcement factors, but objections on both sides usually rest mainly on hard moral boundaries. As a consequentialist, I think that’s the wrong framework to be using.

          • g says:

            My argument includes a step where physical force is threatened. Where there’s a stable government, it’s usually the last arbiter, but without that people would still defend their property with physical force.

            But I don’t think I see the actual problem. onyomi suggests (if I’m understanding correctly) that the government should do as little as possible because the more it does, the more taxes need to be raised with the threat of force backing up the raising. But an exactly parallel argument seems to say that businesses should do as little as possible, because the more they do, the more work needs to be exacted from workers with the threat of poverty backing up the exaction. If for some reason only threats of violence concern you, then indeed that turns into “the threat of violence” (in defence of property), and indeed it may be the government carrying out the violence. But whoever does it, it’s equally the case that for businesses to do more they need more work done for them which requires more people more of the time working to avoid the choice between starvation and theft-and-maybe-prison.

            (There’s an obvious apparent difference here, which I think is enlightening to look at further. As long as scarcity and private property exist, that threat of starvation will be there and people will need to work to eat, which means that if a business does less it doesn’t mean fewer people facing that threat, it means fewer people having an answer to it. Isn’t the situation with taxation different? Well, maybe. If taxes went down, then indeed people would pay less tax and have more money in their pockets. Except for the people employed by the government, some of whom would be out of a job. And the contractors used by the government, ditto. And the people who benefit from the government’s spending in other ways (e.g., using the roads they build, not-starving because of the benefits they pay out). And the estimates I’ve seen (which are for the US; other countries are likely to differ) suggest that actually on net everyone would be poorer if the government collected less tax. I’m not sure whether that’s quite the same thing as being less threatened-with-violence, but as suggested above I’m not too convinced by the whole tax=violence thing anyway.)

            I don’t know who’s ignoring the fact that government action carries costs. Any time any sort of government spending is proposed, it seems to me like everyone’s aware that it’s likely to mean taxation, borrowing or cuts elsewhere, the costs of all of which are pretty obvious. And there is seldom a shortage of people pointing out those costs. So I’m just not seeing this alleged tendency to ignore them; it seems to me like they’re very widely acknowledged. Perhaps it’s different for different sorts of government action; in the UK, where I am, a lot of recent political debate has been around government “austerity” which is all about those tradeoffs.

          • Nornagest says:

            I may have been optimizing for snark rather than clarity. But I was trying to point to a deeper asymmetry, which is that most government action is inherently coercive, either by way of taxes or directly on the enforcement end, while labor transactions may be contingently coercive depending on information asymmetries and workers’ exact circumstances and the general expectations of scumminess floating around.

            Employers, absent direct physical coercion (which has absolutely happened, but also isn’t what we’re talking about), are not sitting in the head of the lifeboat pointing a gun at anyone. They are offering a transaction, something like “you give me a hand warmer from your emergency pack and I’ll give you a ration bar from mine”; if accepted the terms may be backed by force of law, but “no” is always an option. That implies a profound difference in the underlying game theory. Point a gun at someone and force them to bail and they’re not going to be happy about it, but everyone may be better off afterward; offer a transaction at acceptable terms, and both parties are going to be happier. Now, if you’re the only provider of ration bars around, that does give you an opportunity to gouge, a situation that breaks that pattern and is plausibly analyzed as coercive — but I don’t think it generalizes well to the average case.

          • onyomi says:

            I basically agree with Nornagest. Life’s “forcing” you to eat is not the same as an actual person or organization forcing you to do something against your will.

            It is possible to live with no help from anyone else–you can build your own cabin, grow your own food, etc. Most people don’t choose this route because the benefits of trade (not to mention the desire to socialize) are so great. Yet when you enter a voluntary employment contract, you are not “forced” by anyone to do so.

            And even if you were–let’s say it were literally impossible to live on your own–there is an obvious difference between being “forced” by the laws of physics to do something and being “forced” by another thinking human being. Gravity “forces” me to hold myself up all the time. If gravity were a thinking being we might all say it was a real jerk always preventing us from flying. But it’s not, so debating whether or not gravity “should” exert its force is pointless.

            To those who have the power to make choices we ascribe moral agency, and my contention is that moral agents should not use physical force except in extreme circumstances, such as defending oneself against a killer, preventing a lifeboat full of people from drowning, etc. Using force and threats of force to accomplish less pressing goals is wrong, even if those goals seem noble.

          • Irrelevant says:

            I don’t see any reason why governments, alone among human institutions, should be doing only strictly essential things.

            The issue of force has little to do with taxation, nothing to do with starvation, and everything to do with enforcement. All government laws are backed by chains and bullets. Every additional law is an additional threat of chains and bullets. Killing and kidnapping are moral only when necessary, and immoral in all other cases, so there is nothing the government can do which is non-essential but not evil.

          • g says:


            No, different kinds of forcing are different. What I don’t see is how the difference actually leads to the different conclusions you draw in different cases. That’s because I don’t really follow the reasoning that goes from “government action is ultimately backed by the threat of force” to “government action should be minimized”. I completely see the value of minimizing actual use of force, but when the government raises more taxes or gives out more benefits or builds more roads that doesn’t generally appear to result in extra people going to jail or getting shot by armed police.

            The connection between government activity and violent force seems to me too indirect for moral culpability to flow along it in the sort of way that seems required for your argument.

            (Some particular kinds of government activity more directly involve violent force. Waging war, for instance; or putting people in prison for breaking the law. I think your argument has some force if instead of “governments should do only what is absolutely necessary” the claim were “governments should wage war as little as possible, and have as few offences as possible on the books that are punishable with violence or imprisonment”. Even then, though: let’s consider the sort of law libertarians often dislike, a safety regulation that aims to stop non-obviously dangerous products going on sale. If introducing such a law means that a few dozen lives are saved every year, and that once a decade an executive from a company that decided to break the law and sell dangerous products to increase profits gets sent to jail — well, I’m having trouble seeing that that’s not a win, even if in some sense it involves inessential threats of force. How about you?)


            It seems to me that there is a morally significant difference between actually killing or kidnapping (which, indeed, are extreme measures to be employed only when absolutely necessary) and doing other things that ultimately depend, several hypothetical steps away, upon the possibility of killing or kidnapping.

            I own some stuff. I rely (somewhat) on its continuing to be mine rather than being taken away the first time someone comes past who might want it. Ultimately, this too is backed by chains and bullets. (The government’s, if I live in the UK or the US. Perhaps my own, if I live in Somalia.) Am I obliged to do the barest minimum possible with my private property, because my continued possession of it is ultimately dependent on threats of force? I don’t see why.

          • onyomi says:

            @G, you’re missing the key point about government force: it is not purely defensive. Government claims the right to use force to achieve positive goals, whereas the average citizen may use force only in extremes of self defense.

            For example, my claims to property, do, in some sense rely on use of force–that if someone tries to steal from me I will either forcibly stop them or try to get the police to forcibly stop them/and or bring them to justice. The difference is that this is defensive force, whether I do it or the police help me do it.

            Common morality and the law both say that everyone, governmental agent or private citizen, is allowed to use physical force to the extent necessary to protect one’s own life and property. Of course, proportionality matters here too. Hardly anyone would say it’s ethically wrong to shoot a criminal who is shooting at you and your family. It is wrong, however, to shoot someone for trampling your lawn.

            Similarly, I don’t think anyone would say it is wrong for the president to employ the secret service to protect himself. After all, any private individual can permissibly hire armed bodyguards to protect himself if he wants to and can afford it. But if he used those bodyguards not for protection but to intimidate or extort, it would be wrong, even if intimidating people to do things that are for their own good.

            Though society grants to individuals the right to use reasonable force to protect life and property, then, no private individual is allowed to use force the way the government does–in a pro-active sort of way.

            Imagine I decide to start patrolling my neighborhood with a gun, and if I see someone doing something I deem unsafe, I handcuff them and lock them in my basement. Most people would say that’s wrong, even if the people I apprehended really were criminals.

            Now imagine that I knock on your door with a pistol at my hip and say, “just so you know, I have been patrolling the neighborhood and apprehending criminals to make this a safer place to live. For this service, you owe me $100 a month. If you refuse to pay your fair share to support my noble crime fighting initiative, I will have to label you a criminal and lock you in my basement as well.”

            Almost everyone can agree this would obviously be extortion and ethically wrong, even if I really had reduced crime in the neighborhood.

            My problem is with the moral double standard. Why do things which we would label as obviously immoral if regular people did them become moral when government agents do them?

            There are many justifications for this: “we all voted for the government agents,” etc., but imo, they all fall flat. If a majority of your neighbors all agreed to pay the crime fighting fee, would that make it right for them to imprison your for not following suit?

          • Nornagest says:

            let’s consider the sort of law libertarians often dislike, a safety regulation that aims to stop non-obviously dangerous products going on sale. If introducing such a law means that a few dozen lives are saved every year, and that once a decade an executive […] gets sent to jail — well, I’m having trouble seeing that that’s not a win

            This is exactly what I meant earlier when I was talking about ignoring costs. Most of the cost of, and most of the coercion inherent in, a law like this does not come in the form of executives getting sent to jail or even corporations getting fined. It comes mainly from adding a barrier to entry backed by government force: even if you weren’t going to do anything unsafe, compliance needs to be documented, it usually involves enforcing best practices and thus makes it difficult to experiment with new methods, and it often costs money in licensing fees; and large, established organizations are better equipped to pass those hurdles than new, small ones. The taxation needed to fund enforcement is also significant, but to a lesser degree.

            It may be that the regulation’s worth it anyway. But I’d really like to see some kind of accounting of those costs rather than couching the debate entirely in terms of tragic industrial accidents and evil executives going to prison, and if you’re the consequentialist you claim to be, you should too.

          • g says:


            OK, agreed, the use of force that the government (in principle) threatens is not “purely defensive”. But if that’s immediately decisive for you, then what that tells me is that you’re adopting moral axioms I don’t see the justification for.

            The right, or the absence of a right, to use a particular level of force in a particular situation isn’t (so, at least, it seems to me) something handed down by the gods or found encoded in the laws of physics; it’s something a society agrees upon in the hope that it’ll be conducive to peace, prosperity, and so forth. In the actual systems we have in our countries, governments do have the right to use force in situations other than self-defence. I gather you would prefer a system where they don’t; fair enough, but an argument for that would need to look at the likely outcome each way, rather than saying (I paraphrase, unkindly but I think not unfairly) “ordinary individuals don’t have that right, so obviously governments shouldn’t either, because what possible reason could there be for giving different rights to individuals and to governments?”.

            I don’t know what you get if you seriously try to imagine a system in which governments don’t have the right to do anything beyond self-defence. When I do it, though, the results look really terrible; much worse than the way things actually are in most nations with what I’d consider functional governments. Maybe I’m wrong; but until I see good evidence that I am, I’m happy to continue conceding governments those rights.

            Incidentally, in your vigilante example, I might well be strongly in favour of your actions if we didn’t already have police to do that job. The reason why you don’t have the right to lock people up for committing crimes isn’t that arresting people and locking them up is always bad, it’s that we’ve settled on a system where the arresting and locking up are centrally coordinated. I have to say I think the failure modes of this system (although really bad sometimes) are probably preferable to those of vigilantism. Given the paucity of calls to abolish the police force, it seems like this opinion is widely shared.


            I focused on people going to jail because the question at issue was that of whether not-strictly-essential government action is immoral because it involves potentially-violent coercion. Of course such a law will have costs and benefits other than the ones I mentioned there, but there’s a limit to how digressive I am willing to be. And of course when a particular law of this kind is being considered, making a good decision would require looking at all the major costs and benefits, including the ones you list. But, again, that wasn’t the discussion here. So I gestured briefly towards the main gain from such legislation, and towards the cost that was actually being cited as a near-fatal objection to government action, hopefully making it clear that the former can outweigh the latter. Who knows?, maybe it turns out that regulation always ends up harmful on net. (I doubt it, but it could do.) That would be a very interesting and powerful argument against a lot of things governments do — but it would be a completely different argument from the one I thought onyomi and I were discussing.

          • Nornagest says:

            @g — Well, I’ll admit I was digressing a bit. But in this context, I think my objections might be more relevant than you’re giving them credit for.

            Specifically, reducing the libertarian objections to initiation of force to cases where people actually get shot or sent to jail is a bit of a straw man. The objections are much more general, and easily cover cases like the ones I described; if it takes eight permits to run a taco truck in Oakland (it does in at least one case, by the way), the occasional arrest or fine is the least of that policy’s economic and social consequences.

            Libertarians tend to lump that all under “force”, partly for philosophical reasons and partly for rhetorical. I don’t totally agree with that framing myself (though I did briefly discuss coercion); I feel that it’s a bit overwrought and distracts from a realistic accounting of consequential impact, which is why I didn’t use it. But once we get down to tallying real personal and financial damage, that’s mere semantics; the harms are the same either way.

          • g says:

            Nornagest, I think we are agreed that the actual ethical question of importance is: does such-and-such a scheme of government, where the government has such-and-such powers and such-and-such responsibilities, work better or worse than whatever alternative we’re considering? — And that talk of coercion and violence and the rights of a government agent acting as such obviously being the same as those of a private individual, etc., etc., etc., may make effective rhetoric but is probably poor reasoning.

            I have no intention of claiming that all government regulation is beneficial. I guess you have no more intention of claiming that none of it is. So whatever disagreements we have are probably complicated empirical ones that we have no realistic prospect of resolving in the comments here.

          • onyomi says:


            Would you agree that using force when force is not imminently necessary is morally wrong?

            If so, is it not wrong for the government to use force or the threat of force (including taxation) to do things which are not imminently necessary?

            If not, there are, it seems, only two other options:

            1. Everyone can use force in any situation so long as it produces better consequences. I should therefore be permitted to kill my neighbors’ developmentally disabled infant, so long as I am confident it will result in more long-term happiness for the parents and society at large.

            2. Everyone can use force in dire situations, but a special group within society can use it whenever they see fit.

            If 2, what gives that special group that special right the rest of us don’t enjoy? Where does anyone get the right to delegate a right they don’t themselves have? How could something which would be immoral if I did it myself be moral if my representative did it?

          • g says:


            To your first question: no, of course I don’t agree. E.g., suppose the use of force is plainly but not imminently necessary?

            To the follow-up: I don’t agree that those are the only options, and in fact my position is intermediate between the two. Better consequences are, indeed, the key thing; but, people being fallible, one thing that has “better consequences” is setting up a framework within which some kinds of decisions are strongly discouraged even if they seem to have better consequences. E.g., probably most people who think that killing their neighbours’ infant children will improve the world are badly wrong, for which reason doing so gets an extra penalty attached to it.

            More generally, the framework we’ve got heavily regulates the use of force, and having a framework that does so has “better consequences” than (e.g.) just letting anyone use whatever violence they want against whomever they want.

            So, to address your last question. What gives the government the right to use force in ways the rest of us aren’t entitled to? Well, simplemindedly, the law does. One level of “why” further up: what gives them that right while denying it to the rest of us is the fact that a system in which the use of force is regulated produces a happier, stabler, more prosperous society than one in which it isn’t or one in which it’s forbidden in every case except self-defence against imminent threats. (I’m not even sure the latter is possible; who’s going to enforce that prohibition and where do the resources come from to enable them to do it?)

            (Of course you may dispute the “fact” I alleged above; you may think that in fact a system in which there is no government regulation would lead to a happier, stabler, more prosperous society. But I already raised that issue, and you already chose not to pursue it…)

            One final remark. Trying to decide what a system of government should look like purely on the basis of principles about who can use force against whom seems like a really bad idea, unless every other misfortune a person can suffer pales into insignificance beside having force used against them (or even beside the threat of having that happen). I don’t think that’s so; I don’t think it’s close to being so. (And, in fact, I question whether you do. Imagine a world in which taxation works by the government simply taking money from you, in some manner you have no way of preventing. Perhaps they employ wizards or something. Now there is no violence, no threat of violence; but you are no better off for it. Are you sure that your actual objection to taxation is all about the threat of force?)

          • onyomi says:

            No, I *do* think force being used against me *is* the worst things that can be done to me *by other people.* What could be worse than being murdered, raped, kidnapped, robbed, etc.? And why do kidnapping and robbery, which are obviously wrong if done by a private citizen, become right if done by the government? Because “we” decided it was better? Who’s this “we,” and since when does majority decision make right? As to the idea that I don’t *really* think of taxation as being like robbery, you’re wrong there. Every year when I pay the IRS I literally feel the same rage and helplessness I would feel if I were paying protection money to some local mafioso whom I dare not offend. The only difference is it’s more orderly. Orderly robbery is still robbery.

            Society can clearly be wrong on moral question. Pre-nineteenth century America endorsed slavery. That doesn’t mean slavery wasn’t wrong. Slavery was wrong in the 18th century and every century before that. People came to *understand* that it was wrong; they didn’t make it wrong by changing their moral calculus. By your logic, if it had been deemed a net negative for social happiness to abolish slavery then it would not only not have been right, it would have been wrong to abolish slavery. And let’s say there really was a point in history at which the abolition of slavery would have caused more net unhappiness than happiness? Does that mean slavery was right back then and could conceivably become right again in the future? Is slavery only wrong right now because of particular circumstances which obtain?

            I have intentionally avoided trying to argue more fundamental moral principles, because I don’t think I have to. The principles I’ve advocated are already shared by virtually everyone, who would agree, for example, that one can’t just go around kidnapping people and demanding payment for it, or that one can’t demand someone in a lifeboat make you a sandwich at gunpoint even if it increases overall onboard happiness.

            Though I don’t agree with consequentialism, let’s accept your contention that, basically the right is that which maximizes long term, overall social happiness, but certain types of decisions must receive a strong penalty because people are very likely to be mistaken about them. I don’t know who gets to be the judge of which decisions get this “penalty,” but it seems like if anything deserves it, it would be using violence or the threat of violence: i.e. in most cases where you think using force will make things better, you are probably wrong, so err on the side of not using force unless it is clear that doing so will produce vastly better consequences.

            If we accept the above, doesn’t it still imply that the government should err on the side of not using force (i. e. letting things get done in voluntary, market-based ways) unless the need is very pressing, given that you agree that using force often produces unexpectedly long-term negative consequences?

            As to the factual claim that a certain level of government is necessary to prevent catastrophically bad consequences or insure a much higher level of well-being than would otherwise be possible: again, why is anyone justified in using a level of force *beyond* that which is necessary to avert disaster? Unless you think that after the sinking boat occupants finish bailing out the water it is okay for you to continue pointing your gun at them and say, for example, “you there, you look like you have too much money and this other person over here has not enough. Let’s take out your wallet and spread the wealth around.” This seems obviously wrong on any moral system I can think of. Why does it become right when “we” do it?

          • g says:

            Yup, being murdered, raped or kidnapped is very bad. Has the government done any of those things to you? I’m thinking probably not.

            What the government has actually done is to say: “Give us some money so that we can build infrastructure, stop people starving, imprison people for smoking marijuana, prosecute pointless Middle Eastern wars, run schools so every child gets some education, etc. If you refuse then we may use force against you.” I appreciate that you don’t like that, but it’s no worse than just taking your money; so if indeed the use of force outweighs everything else, you should be very happy for the government to take people’s money if that enables them to reduce the amount of violence that happens. (Which, empirically, it sure looks like it does. Would you feel safer in Sweden or Somalia?)

            It seems unlikely that all the principles you’re relying on are shared by virtually everyone, since most people seem content to have a government that actually does things.

            It is my opinion (and I think it will be widely shared among those who take the trouble to think about it for a few minutes) that although a government unable or unwilling to do anything other than avert imminent crises would use less force than actual current governments, having such a government would mean that more force was used overall, because lots of the things governments do that aren’t averting imminent crises help to reduce violence (directly or indirectly).

            I don’t like pointing guns at people and telling them to spread the wealth. I also don’t like seeing people starve to death because they’re disabled and unable to do work anyone will pay them much for. To the best of my knowledge, no one has yet come up with a system that prevents the latter without an element of the former. I also don’t like seeing people unable to get medical care unless they have a large enough pile of money, or children not getting an education because their parents can’t afford to pay, or vital infrastructure like roads being in the hands of profit-taking firms and inaccessible to the poor. Because all those things tend to produce a permanent underclass, which not coincidentally seems to be a thing that happens more in nations with more limited governments. And the members of those permanent underclasses end up in a very bad way even if no one is actively engaging in violence against them. And, again, I know of no effective way to mitigate those problems other than to have a government providing those services, or at least paying others to. Yes, it’s “sharing the wealth around”, and you are free not to like it, but it saves lives and I’m pretty sure it ultimately leaves almost everyone better off.

          • onyomi says:

            I completely disagree with most of your empirical claims, but I don’t really feel like going over them here, since most have been addressed countless times by countless other internet libertarians. See, for example,

            Suffice it to say that globally and historically the correlation is not more government=more peace and prosperity, but rather quite the opposite.

            Regarding “taking” my money to build roads, etc. “Taking without asking permission,” aka “stealing” is a kind of force. I seriously doubt all the criminals put together in the US, even all the criminals who might theoretically come into being if the government were much smaller, could steal as much money as the government currently takes in taxes. And if it’s true that government force is needed to prevent a much greater increase in violence amongst the populace, then why is my tax money going to fund many things which have nothing to do with violence prevention? I’m not even arguing the stronger case that all taxation is immoral. I’m just saying that, assuming some level of taxation is necessary to prevent blood running in the streets, then why is it okay to tax far more than that level?

            Regardless, you seem not to really address the ethical question, which is where the government gets the moral authority to use force against me in cases when almost everyone would thinks it’s wrong for private individuals to do the same? What is this ethical “miracle of aggregation” whereby something which was bad if a private individual did it becomes good if someone with a badge does it?

            If better utilitarian outcomes are all that matters then I should be able to tax or imprison my neighbors if I have good reason to believe doing so will result in better outcomes, as I should be able to kill a healthy patient to distribute his organs to five sick patients. Do you believe only democracies are able to make these better utilitarian calculations, or can dictators, as well? And if the former, are democracies the only morally legitimate governments, and if so, why? Would it be wrong for the doctor to decide on his own to distribute the organs, but right if a majority voted for it?

            The reason most people think taxation is okay but extorting your neighbor is not okay is status quo bias. If they were to consistently apply to government the same moral standards they already hold about private individuals they would not think so.

          • g says:

            I think a sufficient indication of how much the IEF tells us about bigger versus smaller government is the fact that Denmark comes in above the USA and Sweden barely below. (Tax revenues as a fraction of GDP: 49% in Denmark, 46% in Sweden, 27% in the USA.) Canada ranks above the USA. It turns out that the IEF is (of course) a composite of ratings on several metrics, most of which are all about quality rather than size or activity of government — and the two government-size measures (“government spending” and “fiscal freedom”, which actually turns out to mean “low taxes”) correlate negatively with almost all the others. I haven’t run the numbers, but I bet that if you computed a version of the IEF that ignored government spending and “fiscal freedom” you’d see a stronger relation with prosperity.

            … OK, I ran the numbers and it’s true. GDP per capita is better explained by IEF without government spending and “fiscal freedom” than by IEF with them; if I make a model that uses all ten of the IEF metrics and regress log GDP per capita against it, the most important term is government spending, and the way it goes is that more government spending is better. (Note: one needs to be careful here because the IEF metric called “government spending” is actually measuring low government spending!)

            I actually tried slicing things lots of different ways — different sets of metrics from the ones that make up the IEF, raw GDP versus log GDP, etc. Some things come and go (e.g., one can make “investment freedom” look good or bad depending on what else is in the mix, and likewise for “fiscal freedom” a.k.a. low taxes) but everything I tried had low “government spending” scores (i.e., more government spending relative to GDP) going along with higher GDP.

            (I would be cautious about inferring too much from any of this. Lots of other factors affect a nation’s prosperity, none of these models “explains” much more than half the variability in GDP or log-GDP, and this sort of correlational analysis doesn’t tell us what causes what — e.g., you might imagine a model in which for some reason rich nations choose to show off by taking on a burdensome level of government spending.)

            If you choose to classify depriving you of money as “a kind of force” even when no physical violence is involved, then of course that’s up to you. But note two consequences of doing so. First: when you say that of course the use of force is the worst thing anyone could do to you, and give examples like rape and kidnapping, you’re cheating, because however disagreeable it is to have money removed from your bank account or deducted from your pay, it is not disagreeable in anything like the same way as those things are. Second: if it turns out that government spending is good for your prosperity (which my comments above suggest it may be) then slashing government effectively makes almost everyone poorer, which is doing them the same kind of harm as you classify with rape and kidnapping and murder.

            If applying “the same moral standards” to governments as to private individuals means treating governments the exact same way as private individuals, then I think doing so is crazy. There are some things I’m happy for private individuals to do but much less happy for governments to do, because they have so much more power. (Making public statements about what religion everyone should follow. Watching what people do in public places and keeping extensive notes.) There are some things I’m happy for (reasonably competent democratically elected) governments to do but not for private individuals to do, because the governments are subject to better oversight. (Imprisoning criminals.) There are some things that governments can do but private individuals simply can’t. (Making laws, like the ones necessary for anything remotely like a free market to exist.) And so on. Governments and private individuals have very different powers, very different relationships with individuals and with (other) governments, etc. Why on earth should we treat them the same?

            Some of the things that make governments different from private individuals depend only on the fact that they are governments. (E.g., the fact that governments are big enough and visible enough to be able to resolve coordination problems between citizens by making laws. For the avoidance of doubt, not all coordination problems are best solved by governments, but some seem to be.) Some depend on what kind of government — e.g., democratically elected ones run by people who are mostly not idiots will act more often and more effectively in the interests of their citizenry as a whole, compared with totalitarian dictatorships run by megalomaniacs. I don’t see any point in trying to make a black-and-white classification into “morally legitimate governments” and “morally illegitimate governments”; but some kinds of governments systematically produce much better outcomes than others.

          • onyomi says:

            Government is like a parasite. It can grow larger when attached to a bigger host. Doesn’t mean it’s good for the host. Government spending in the US today is much, much higher as a percentage of GDP than it was in 1890 and Americans are richer now than they were in 1890. But America became a better place to live much more rapidly in the 1890s than in the 1990s or 2000s. Comparing apples to apples leads to the conclusion that the level of government we had in 1890 was more conducive to growth and innovation than the level of government we have now.

            Re. you being “happy” to give up your freedoms to a huge organization that solves coordination problems: fine with me. If you and your friends want to pay Barack Obama to be your utility Czar and order you around, I have no objection. But what right do you have to demand I do so? Because a lot of people voted? If two wolves and a sheep vote on what’s for dinner does it make it right?

            Even if we assume utilitarianism, what if I believe social utility is better served by not having a government, even though you and many others think it’s better served by having one? Why do you get to force your utility judgments on me? Why do I have to do what I believe is wrong just because a lot of people believe it’s right? See

            Or do you just think that whatever a majority of people in society thinks is right is right?

            Moreover, I see no reason to expect that politicians are better judges of morality than the average citizen. If anything, they seem to be mostly power-hungry megalomaniacs, which is the type of person you’d expect to pursue the job. And people tend to vote poorly according to predictable biases. See Bryan Caplan on the myth of the rational voter.

          • The reason most people think taxation is okay but extorting your neighbor is not okay is status quo bias. If they were to consistently apply to government the same moral standards they already hold about private individuals they would not think so.

            Status-quo bias, like risk aversion, might lead to wrong conclusions, but it’s not a fallacy.

            To re-use one of my favorite quotes in this context: “The life of the law has not been logic, it has been experience.” Human beings have spent the last several thousand years working out how to structure their societies. Every country in the world but Somalia has a government that collects taxes and such, not because of some airy-fairy theory that governments have a greater moral right to order people around, but because it works.

            Moreover, I see no reason to expect that politicians are better judges of morality than the average citizen. If anything, they seem to be mostly power-hungry megalomaniacs, which is the type of person you’d expect to pursue the job.

            Now, as a politician myself, I may not be the best judge of this, but to my knowledge, no one has ever suggested that I was a power-hungry megalomaniac, and I don’t see that quality in the other politicians I know well, either Red or Blue.

            I suppose it depends how “power-hungry megalomaniac” is defined. Megalomania has been displaced in clinical use by “narcissistic personality disorder,” another term I am confident doesn’t apply to me. On the other hand, perhaps words abandoned by medicine, like idiot, imbecile, or hysteria, have been freed from having any very specific meaning.

          • onyomi says:


            I understand that status quo bias does not, by itself, imply that the status quo is wrong. It’s just reason to question one’s logic more carefully when one finds it supports the status quo.

            Similarly, the fact that everyone has a government does not, by itself, imply that it works. There are many other reasons I can think of why this might be the case, not the least of which being that elites like to centralize power.

            Wikipedia’s description of Narcissistic Personality Disorder:

            “a personality disorder in which a person is excessively preoccupied with personal adequacy, power, prestige and vanity, mentally unable to see the destructive damage they are causing to themselves and others.”

            Can you honestly say this isn’t a good description for John Boehner, Nancy Pelosi, Harry Reid, or John McCain? As for your contention that it doesn’t apply to you and the politicians you know, what level of office have you been elected to? It may apply somewhat less to City Councilmen and the like than, say Senators and Presidents, but as for the politicians I see on the news, they all strike me as extremely vain, status-seeking, etc., and I can think of more than a few mayors and city councilmen to whom it applies as well.

            I do think power is more corrupting and more attractive to the narcissists the higher up you go. Who but a narcissist would think they knew better than 300 million people how to run their own lives? This is one reason why much smaller governments do better. There is a better chance of policy actually reflecting the desires of a strong majority of the residents when you don’t have to form massive coalitions. Give me Hong Kong, Singapore, Switzerland, etc. over the USSR, or, indeed, the US, any day.

          • g says:

            onyomi, larger GDP does not obviously imply that government can get larger as a fraction of GDP. There are so many big differences between the US 100 years ago and the US now that I find it hard to believe that you are actually making a serious argument that if living conditions improved faster then than now then it shows that smaller government is better. I don’t even know by what metric you’re saying things improved faster then than now; considering (inflation-adjusted) GDP as a rough but quantifiable measure, it went up by about 2%/year (on average) in the 1890s, which is slightly less than in the 1980s and slightly more than in the 1990s. (And quite a lot less than the 1960s, and much less than the 1940s — though of course they were rather a special case.) If life in the US improved much faster in the 1890s than in the 1990s, I guess it was because of improvements in technology and medicine and suchlike (most notably, I guess, the first burgeoning of the electrical power system) and if you want to claim that those were caused by small government, well, show me your evidence.

            You ask questions like “what right do you have to demand I [allow Barack Obama to order me around]?” as if you actually have a realistic alternative in which people don’t get ordered around. But I don’t believe you have one. In particular, abolishing government (which really seems to be what you’re calling for; otherwise we still get a “utility czar”, still have to pay taxes, etc.) would not prevent it; it would, at most, mean a wider variety of people ordering others around.

          • onyomi says:

            I have offered a perfectly reasonable alternative: hold government agents to the same moral standards as everyone else.

            What you have failed to offer, however, is an ethical argument as to why we should do otherwise. And since you are arguing that government agents belong to a special category, the burden is on you to provide one; otherwise, it’s just special pleading.

          • g says:

            No, you haven’t offered an alternative, you’ve described one feature of an alternative. What do you think society would look like after making your change (which amounts, so far as I can see, to abolishing government altogether)?

            I have described some ways in which I think what we have now is better than what we’d get then. (And that is an ethical argument, whether you accept it as such or not.) You seem almost completely uninterested in what the consequences would be.

            Incidentally, while I’ve said a fair amount about how I see ethics, so far all I can tell is that you think the use of force is bad and that you regard taking away a person’s property as a species of force. Which is hard to square with your indignation at the idea that rights might be socially constructed — if anything is socially constructed, property is. But never mind consistency for now: surely that can’t be the entirety of your ethical system?

            The most plausible system I can think of that looks like that says this: the only thing that really matters ethically is that no one should have anything done to them without their consent. I think this has two serious problems: first, it’s not at all clear what counts as “done to them” (e.g., apparently modifying the charge levels in some tiny capacitors inside computers at your bank’s headquarters can constitute doing something to you, and I suspect that any non-gerrymandered definition broad enough to encompass that is also going to be broad enough to include things you won’t want, such as consuming a good you also use and thereby increasing the price you pay); second, it isn’t merely not consequentialist (that’s fair enough) but completely ignores consequences, which in turn has two problems. The first is that an action may, while not in any way applying any force to you directly, have consequences that include other people later doing so. (Obvious example: an already-existing government decides to make something you want to do illegal. Less obvious example: someone abolishes the government where you are, and in the ensuing anarchy you get forced to do things by people with more guns or more minions than you have.) The second is that it seems to commit you to preferring a world in which you have $10k and no one takes it from you over an otherwise similar world in which you have $100k and then someone takes $10k of it, which seems to me like a very of preference.

            [EDITED to add: of course the last few paragraphs are criticizing not what I know your opinions to be, but one guys at what they might be, so they may well not be directly applicable. But I guess that your actual opinions, if you choose to share them here, are likely to be subject to similar criticisms.]

          • onyomi says:

            You are correct that I would prefer to eventually abolish government entirely, though I could see that happening gradually, as part of a wave of successively smaller secessions and/or privatizations of the functions government now. I do not, of course, think this would be a less pleasant world, but rather a much nicer and more functional one. For reasons why this may be the case and examples of how it might plausibly come about, see, for example, David Friedman’s book, or the second half of Michael Huemer’s book on political authority.

            My ethical view is not consequentialist or utilitarian, but that doesn’t mean it ignores consequences entirely. As I have said before, IF we assumed that not having a government would produce disastrous consequences (I don’t believe that, but if), then I would concede that a government might be morally justifiable in the same way that stealing is wrong, but if you’re literally starving, and there is no other non-coercive way to obtain food (say, you are lost in the woods, haven’t eaten for days and pass an empty cabin with a refrigerator full of food and have no way of contacting the owner to obtain permission), then, in those extreme circumstances, it may be right to steal food.

            What does not follow, however, is that, having established that it’s okay to steal food when you’re starving and are absolutely out of options, therefore it is okay to steal food whenever you want food.

            What I am saying is that IF government, whose use of coercion, like stealing food, is prima facie morally suspect, is necessary to prevent much greater moral ills than the government itself perpetrates, then and *only to the extent necessary to prevent those greater ills* are its actions morally defensible.

            Let’s say having a taxation-funded police force which coerces people and throws them in jail is absolutely necessary to preventing a much greater level of violence and coercion. And let’s say that private security forces or voluntarily-funded community policing are just not a viable option for whatever reason (I think they are, but for the sake of argument). Then, in that case, it would be permissible for the government to tax and establish a police force. It does not follow, however, that because the government needs a police force to prevent violent crime, that they can therefore heavily tax cigarettes to discourage bad habits, require that hairdressers have a license to work, kick out immigrants who have committed no violent or coercive crimes, etc etc.

            Maybe some of these laws make society marginally better (though I think they usually make it worse), but they are obviously not required to prevent blood from running in the streets. They are not required to prevent disastrous consequences or to deal with extreme situations, and so the use of force for these purposes is not morally justifiable, just as forcing someone to make you a sandwich at gunpoint is not justifiable.

            So clearly, I DO care about consequences. It’s just that consequences are not the only thing I care about. I think killing one healthy patient against his will to save five sick patients, for example, is obviously wrong.

            My own ethical view is a species of realist, rational ethical intuitionism, but I don’t know if I want to get in a debate about the specifics of that, as I have previously described the view at length in older threads. The basic idea is that some things really are right or wrong, and we may perceive and debate that rightness or wrongness with our rational faculty. Morality is not a pure social construct, because if it were that would imply that if most people in a society think something is right, then it is right, which is obviously false.

            Ethical intuitionism can take consequences into consideration, but it does not attempt to reduce ethics to that. The view is rather that what makes something right or wrong is complex and that attempting to systematize a grand moral system based on just one parameter is likely to fail, precisely because everyone judges the success or failure of a moral system against their own intuition of what is right and wrong, and intuition takes into account many factors. When a system produces absurd consequences (say, that you should kill one patient to save five), then adherents generally make concessions or logical contortions to square it with the intuition (well, maybe killing one to save five is a net negative because it makes society feel uneasy, etc.) rather than accepting the counterintuitive conclusion.

            So given that everyone ultimately contorts the logic of their chosen ethical system to square it with their intuitions, why not simply say that moral intuition IS the basis by which we must judge morality? It may seem too subjective (though I am claiming it is perceiving an objective quality), but it still allows for rational debate, and is, in any case no less subjective than utilitarianism, which requires an arbitrary determination of whose utility judgments are better.

          • g says:

            I haven’t read Friedman’s book, but I have read Scott’s review of it; his concerns after reading the book overlap a lot with my concerns before reading it, which I take as an indication that the book doesn’t address them convincingly. Since Scott is at pains to point out that DF is smart and thorough and attempts to rebut objections, that doesn’t seem like a good sign that the things that look to me like near-insurmountable problems will all turn out fine. Perhaps Michael Huemer does better?

            Thanks for going into more detail about the ethical position you’re arguing from. I’m glad to hear that you do care (and admit to caring!) about other things besides physical coercion and taking of property, and that you do pay (and admit to paying!) some attention to consequences. But I’m having trouble seeing how, in practice, your moral judgements differ much from what they would be if you didn’t. For instance, you concede that maybe in some weird counterfactual world it would be necessary to have a government that applies coercion — but only to avoid disaster, and the only specific disaster you mention is a much greater level of violence and coercion.

            So we differ in (at least) two ways. Firstly, if we have two possible ways to organize a society, one of which has a potentially-coercive government and the other of which doesn’t but will produce somewhat more violence and coercion than the first, I prefer the first: I will generally take less violence and coercion in preference to more, even if it’s not much more, and the fact that the first option involves a coercive government doesn’t change that. Whereas you, I think, will countenance a coercive government only if it saves much more violence and coercion. I don’t think that makes any sense.

            Secondly, I don’t see coercion as infinitely worse, or even vastly worse, than every other misfortune that can befall a person, and I therefore don’t see any reason to say that coercion can only be acceptable when the alternative is catastrophe. I would rather experience coercion than cancer, for instance, and if some government regulation coerces N people in order to prevent 10N people getting cancer then that sounds like a clear net win to me. (Of course it may depend a bit on exactly what sort of coercion and exactly what sort of cancer.) And this applies even if the regulation doesn’t “prevent blood running in the streets”. Such a regulation really doesn’t seem to me very accurately modelled by “forcing someone to make you a sandwich at gunpoint”.

            (Specific example: I’m sure many smokers really like to smoke; I’m sure tobacco company executives really don’t like being obliged to put health warnings on their products, or forbidden to lie about the consequences of smoking; but I will very happily trade their unhappiness, and all the other negative consequences of tobacco regulation, for the tens of millions of premature deaths prevented by tobacco regulation since, say, 1960. I take it you disagree; I would be interested to know whether you deny that tens of millions of premature deaths have been prevented by tobacco regulation, or consider that the harm done by that regulation outweighs them, or just don’t care about the deaths if it takes government action to prevent them.)

            I agree that this isn’t the place for a substantial debate on metaethics, but I will remark that your argument against moral nonrealism is wrong. Nonrealism doesn’t imply that “if most people think something is right then it is right” (since, e.g., the conclusion there appears to presuppose moral realism); rather, what it means is that words like “right” always need (implicitly or explicitly) a specification of whose values you’re considering; if most people think something is right then it’s right-in-their-value-system, but that doesn’t mean it has to be right-in-your-value-system. But I’ve no particular objection to intuitionism — though it seems that your moral intuitions may just be extremely different from mine and it’s not clear how to proceed when that happens.

          • onyomi says:

            I can highly recommend Huemer’s “Problem of Political Authority.” In fact, if I could suggest only one book for all non-libertarians to read, it would be that book. The first half makes an ethical case for why “at most” a night watchman state may be justified, and the second half makes an empirical case for why pure anarchy would actually be a good thing. It also addresses questions like why belief in government’s ethical legitimacy is so widespread if, in fact, it isn’t.

            If you are truly a pure, committed utilitarian then it may be impossible for us to reach any agreement on ethical issues.

            A bit out of left field, but have you ever seen this episode of Star Trek:

            And if so (or less ideally, if you have read the summary), do you think that the ambassador acted ethically? I bring it up because I happened to rewatch this episode recently and it struck me then that it seemed a good example of why I am not a utilitarian (because I found the ambassador’s actions to be obviously wrong, even though we are told he saved many people).

          • g says:

            I am not a strict utilitarian. I am pretty much a strict consequentialist, but among the actions that have consequences are, e.g., committing always to act in a particular way whether or not it seems to maximize utility, and both for game-theoretic reasons and because of human fallibility such actions may sometimes be best.

            I haven’t seen that Star Trek episode. Having read the summary it’s not clear to me whether we are supposed to understand that the ambassador’s actions really did do the good he claims they did. In the real world, people making such claims are very commonly lying or mistaken and one does well to discount them heavily. But — embracing the hypothetical for the sake of argument — if I were fully convinced that his actions predictably prevented (say) a major war at the cost of a couple of lives, and that any less-horrible action would predictably have led to war, then I would be glad he did it; and if I were fully convinced that it was only for that reason that he was willing to do it, I would approve of his doing it.

        • Harald K says:

          Let’s try to explain this with yet another example. You’re hiring for your company, and you find that you are biased against women. From thorough analysis of results after the fact, you find that you underestimate how useful women are to your team. Your current gender balance is 30-70, you find that it should optimally have been 40-60.

          (Parallel: You find, after the fact, that society is too eager to pass laws that shouldn’t be passed, and too reluctant to repeal laws that should have been repealed.)

          Now you try to fix that with a bias. Henceforth, in the interview process, you give a 10% bonus to the scores of women!

          (Parallel: You make it easier to repeal laws and harder to pass them!)

          There’s just a little problem. It turned out that the reason you end up hiring too few women, is that you are bad at evaluating them. For men, you assess competency accurately by many subtle characteristics, which elude you for women. On average, you aren’t biased – you think some women as worse than they really are, but no more than you think women better than they really are. However, since your hiring is risk-averse – you would rather hire a candidate you’re 100% certain of doing a good job, than one with 50% chance of doing a fabulous job and 50% chance of being a disaster, women still suffer from your greater uncertainty.

          Now what happens with your newly implemented crude bias? You’re just as bad as evaluating women as before. And because you were right to be conservative about the risks, your company suffers from it – sure, you hire some more women you would have rejected before due to uncertainty reasons, but you hire a lot more that you would have rejected if they were men (because people bad at the job are more common than people exceptional at the job). The solution would have been to get better at evaluating women, but you didn’t do that did you? Noo, you went for the built in bias. You thought you could outsmart your own rationality, and now you’re paying the price for it.

          • Unique Identifier says:

            As someone who often disagrees with you, and generally doesn’t bother posting except to disagree, I feel I should take the time to say that this is a beautiful post.

            I often tell a similar story myself, in my case about passing rates and economics at a particular university. In my opinion, the key message is that, intelligent agents (people) are good at adjusting to systems and co-opting them. [Your example is more along the lines of, if an extremely simple fix would work, we probably would already be using it.]

            Systems are very simple, relative to human intelligence – a system is designed by a human and it’s generally the point that people should be able to understand them. In contrast, a system never (to date) understands people – most cannot even adjust in any meaningful way. Insofar as what the system is supposed to achieve goes against the self-interest of the people it concerns, people tend to win.

  23. Planet says:

    Delaware is uninhabited except by corporations looking for tax loopholes.

    My understanding is that it has more to do with Delaware having an established and well-understood system of corporate law.

    I like having paper trails in elections as much as the next person, but if one guy isn’t going to keep a very good record of election results, and the other guy is going to kill a million people, that’s not a toss-up.

    If you figure that paper trails are upstream of preserving functional democracy, which is upstream of many other good things, it might look a little less lopsided. (I don’t think this argument is actually particularly strong, I just want people to notice downstream effects more.)

    • suntzuanime says:

      Yeah, the downstream effects of undermining democracy in cases like the Brendan Eich blackballing don’t really seem to register with a lot of people. The point of upholding principles is not so much the principles themselves as the fact that bad things happen when you don’t, things worse than the bad things you could prevent by discarding your principles.

    • Anthony says:

      Having been a former Delaware inhabitant, I have to report that there are real people there. 10% of them work for DuPont. No, not 10% of working adults. 10% of all the people there. There’s a reason that before they ruined the schools, all the school districts were named “DuPont”. (Well, maybe not all, but I was in the Alfred I. DuPont school district, and the neighboring one was the Alexis E. DuPont school district.)

      • Anonymous says:

        10% of 0 is still a small number.

        Where do you get that number? Wikipedia’s figure for the number of DuPont employees is smaller than 10% of the population of Delaware. And that’s not just the employees in Delaware.

        • Anthony says:

          I don’t remember, but it was a while ago. It’s probably an old number – DuPont has shed a lot of employees in the past 20 years.

          • Anonymous says:

            It’s not even the top employer in the state. Less than 1% of the population. Back in 1993, it had 114k employees worldwide. I doubt that your claim was ever true.

    • theotherguy says:

      This is right. While I thought Scott’s original comment about Delaware was funny, it was about as inaccurate as statements can get. (Setting aside the fact that we could, say point out that Alaska has less people than DE) the situation around Delaware corporations is so large scale that it merits more consideration.

      There are basically 2 reasons why being a Delaware Corp is popular among large sophisticated companies.

      1) The Court of Chancery has the most developed corporate case law of basically any court in the country. Hence sophisticated companies and investors know what they are getting into, and by and large they like it.

      2) DE, along with a handful of other states like Nevada have extremely high levels of secrecy. This attracts those interested in activities like tax evasion. Note: this isn’t about loopholes and tax avoidance, but instead evasion.

      3) If we are to include a 3rd reason, its because ‘everyone else is doing it’ / social proof.

      If you are a corporation interested in tax loopholes like Scott indicated, you simply don’t domicile in the US. Instead you do it in Bermuda, BVI, Ireland, Guernsey, and a few others… And if your top holding company is domiciled in the US, no matter what state it is in, you play a game where your intellectual property (and hence as much of your profits as possible) is owned by companies in Bermuda, BVI, Ireland, Guernsey… Delaware really has nothing to do with this.

      • Deiseach says:

        DE, along with a handful of other states like Nevada have extremely high levels of secrecy.

        Delaware, the Switzerland of America? 🙂

  24. Michelle Taylor says:

    Election paper trails is a terrible example because it’s a concern on a different meta-level than the others – if you screw up your election system, then you lose your voice on whether your country should go to war in the next random Middle-Eastern country or not, so it deserves a pretty high position in your reasoning.

    Most of the issues on that slate have that kind of importance behind them – the ones that don’t directly affect hundreds of thousands of deaths affect future potential in hard-to-measure ways which probably add up to a lot more changed lives over time.

  25. Sergei Lewis says:

    And I tend to think something like “Well, I agree with this guy about the Iraq war and global warming, but I agree with that guy about election paper trails and gays in the military, so it’s kind of a toss-up.”

    And this way of thinking is awful.

    As a counterpoint – remember that you’re not just voting for the policies, but also for the person. You are trying to build a model of how they will respond to a given future situation based on what they’ve said they believe and want to do, and also their past behaviour; and of the models you’ve built, you’re trying to select the one most likely to act the way you want. You want to know which of their promises they are likely to sacrifice to get which through; you want to know how they will rule on matters not in their manifesto; you want to know how they will respond to matters arising.

    For *this* purpose, it is surely indeed useful to consider how much they agree with you even on policies that might not be so important in the overall grand scheme of things.

    • DavidS says:

      I’d agree with this. And specifically, I think the issues people often put apparently irrational weight on are ‘value’ issues – it’s not so much about the specific outcome, and more their symbolic importance in working out if someone ‘shares your values’.

      For instance, nobody I knew here in the UK was bemoaning that gay marriage wasn’t being introduced and saying it was a scandal, terrible discrimination etc etc. (we already had gay ‘civil partnerships’, which were in most regards identical and resolved some of the more obvious issues). But I know PLENTY of people who, once the issue had been raised, would completely refuse to vote for an MP who was opposed to it. Not because it had top utilitarian weighting but because they felt it showed those MPs had fundamentally different values to them.

      • Deiseach says:

        One of the reasons why I’m very sceptical about our upcoming referendum on same-sex marriage is that I am not convinced of the conversion to gay rights by Fine Gael. I don’t for a second believe it’s anything like principled conviction; it’s a combination of “These people are going to keep bringing court cases, plus under EU law we’re probably going to have to give in on this in the long run” and easy, cheap, we’ll get a gold star for being progressive and a boost in public popularity after the austerity budgets of the past few years, and it’ll distract from public anger over the mess we’ve made of setting up the utility to charge for water provision.

        So you have government ministers up to and including the Taoiseach, and members of all political parties, running around with “Vote Yes” buttons and turning up for photo-opportunity events pushing a yes vote in the referendum.

        Meanwhile, they can’t provide 24/7 cardiac cover in our local regional hospital – excuse me, our local University College Hospital. Get a heart attack outside of 9-5 Mon-Fri and you’d better hope you can survive the ambulance journey to the next county over (anywhere from an hour to two hours, depending where they send you to the hospitals in the neighbouring cities).

        But hey, at least your same-sex partner will now be able to call themselves your widow/widower at the funeral!

        I’m not joking about the 9-5 bit: there was a case where a woman arrived in at 5:30 p.m. with a suspected heart attack and sorry, too late, unit is closed for the night.

      • onyomi says:

        When hiring a plumber, or even a doctor, does one hire the plumber who shares your fundamental values, or does one simply hire the best plumber?

        I’m not denying that that’s how democracy works, but I am questioning whether it should. I wish people would stop thinking of governmental “representatives” as divine embodiments of the “will of the people” and start seeing them as people there to do a particular job.

        • Jaskologist says:

          But the job we elect them to do is to impose our values on the other half of society.

        • Sergei Lewis says:

          Does anyone see politicians as “divine embodiments”? This is surely a matter of picking the people who can be trusted with the delegation of tasks, not some kind of selection into a priesthood.

          When I’m hiring a plumber, I want them to do a one-off, concrete, well specified job. A better analogy for picking a politician to vote for might be hiring a manager to run a branch of your plumbing business for a few years while you go off grid.

          You need to be very certain that they will make decisions you are happy with over the time you are out of contact before you trust them with your property, budget, staff, and customer goodwill.

          (Arguably delegating someone to take charge of all your stuff while you go out of contact, in an ideal world, ought *not* be a good model for democracy, but that’s the way this world works…)

          • onyomi says:

            Though not explicitly so, I think most people in a democracy see the government as, in some sense, being “us.” The politicians are just our “representatives” whom we empower to do things for us like you’d hire a lawyer to deal with complicated legal wranglings you yourself have not the knowledge, time, or inclination to handle. The democratic process is seen as “us” coming to a consensus.

            My fundamental problem with this is that the politicians in every democracy I know of are not really “representatives” but “elected rulers.” The difference is that one’s “representative” logically can’t do anything you couldn’t do. A “representative” should gain new powers by virtue of representing someone. Politicians in our system have powers which no individual or group of individuals, even large groups of individuals have, such as a monopoly on ultimate decision making, the power to tax, etc.

            They are therefore “elected rulers,” not “representatives.” But most people think of them as “just us.”

          • Doctors and lawyers can do things you cant do, as well.

          • onyomi says:

            Is there anything a lawyer can do for you which you can’t, technically, however ill-advisedly, do for yourself?

            One area where I’d agree is the power to prescribe medicines. But people don’t think of doctors as their “medical representatives.” People realize doctors are empowered in ways others aren’t because of special training. I don’t agree with it, as I don’t agree with any licensing laws, but at least there is not the misconception.

          • Anthony says:

            Ignoring legal restrictions, there is nothing either a doctor or a lawyer can do that you can’t do yourself (except those things which physically require a second person). However, because of their special training, the doctor and the lawyer* are much more likely to do it right without causing (additional) harm. We have laws which restrict what people who are not formally licensed as doctors or lawyers are allowed to do in an attempt to prevent much of that potential harm.

            *Actually, based on my experience, I’m not so sure about the likelihood of the lawyer getting it right.

        • Ben Anhalt says:

          To say “simply hire the best plumber” is to beg the question. How do you decide who is the best plumber? That decision also involves “fundamental values” comparisons. One might value a cheap, good but not fast plumber. Another might value a fast, good but not cheap plumber. To determine what fundamental values a plumber represents, it would make sense to evaluate their responses on certain key questions. Copper vs. PVC? Do they recommend on-demand water heaters?

          • onyomi says:

            But would you hire a plumber whom you know to be competent, fast, and cheap, but who you also know cheats on his wife, or a plumber who is bad at plumbing and expensive, but who has a beautiful family, goes to church every week, etc.?

          • “But would you hire a plumber whom you know to be competent, fast, and cheap, but who you also know cheats on his wife”

            That depends on whether the job you are hiring him for is one that requires you to trust him. If it is, you should give considerable weight to evidence of whether or not you should.

          • Deiseach says:

            Okay, if he’s cheating on his wife because he can’t keep his trousers zipped, that’s none of my business when it comes to “Can he fix my hot water cylinder?”

            If he cheats on his wife as part of a pattern of being untrustworthy, because he loves ‘getting one over’ on people, then maybe he’s likely to pad the bill he gives me for parts and labour, do a poor job, use inferior/the wrong parts and leave me with a bigger mess than when he started.

            Context is everything.

        • Julie K says:

          Or do you assume that someone who doesn’t share your values is (a) an idiot who couldn’t possibly be a competent plumber/doctor and/or (b) a villain who doesn’t deserve to have a job?

          • onyomi says:

            I don’t think you do. Because when people are using their own money to hire someone to do something, they know that they will directly enjoy the benefits of a good job or suffer the costs of a bad job. Therefore, they tend to care most about who does a good job for a reasonable price. It’s not even so much that they will overlook the adulterous plumber, it’s that they don’t care about the plumber’s personal life or values and so never even find out which plumber is sinful. They don’t even bother to research that aspect because it’s largely irrelevant.

            Contrast politicians who are selected on the basis of a vague feeling of “like me”-ness or charisma, and where the voter doesn’t suffer the consequences or enjoy the benefits of an uninformed or informed decision. For them, actual proof of competence, insofar as that is even possible, is often secondary.

            Witness the campaign of Gary Johnson. He basically said, “I was elected as GOP governor of a blue state and finished my second term far more popular than I began, having fixed a number of major problems and saved the state millions of dollars.”

            The message “I pleased voters in a quantifiable way and did what I was hired to do for less money than expected” is apparently not a message that resonates with voters.

          • LTP says:

            Gary Johnson didn’t succeed in the primary because Ron Paul was the default libertarian candidate*. He failed in the general due to structural problems making it hard for 3rd party candidates to get visibility.

            *Which is a shame, if Johnson had gotten the votes Paul had, and the subsequent visibility, he would have been a much better messenger for the ideas.

          • onyomi says:

            Yes, though I think it was also that Ron Paul was the more “sexy” libertarian candidate. You could have Gary Johnson, who merely claimed to be competent, or Ron Paul, who claimed to have a fundamentally different governing philosophy than other candidates. I have to admit that I myself voted for Paul over Johnson in the primary, though I did vote for Johnson in the general.

            So perhaps I can’t blame the voters themselves so much, since, right now, at least, the president is *not* seen merely as someone you want to be a competent administrator, but someone who sets the tone for all kinds of bigger issues and debates. A kind of national avatar who embodies the hopes and dreams of a tribe. So maybe my problem is more with the perception of a what a president is/should be than with how people vote, given the current perception.

    • Julie K says:

      So you’re trying to tell if he belongs to your tribe?

  26. RustyGunner says:

    I am very, very happy for you that you were able to score a stress brain at your recent conference. Sounds like you might need two of them.

  27. David Moss says:

    One reason why the ‘list of boxes of policy positions for the candidates’ thing, actually sometimes makes a bit of sense, even where one or two policies are millions of times more important than the others, is because you don’t get to elect sets of policy platforms, you only get to elect people. And knowing that they endorse a lot of weird, disagreeable policies, even if they are trivial, is potentially a good indication about the virtues or tendencies of the candidate and how they might act in unspecified, unpredicted ways.

    For example, in the UK the Green Party is pretty much the only game in town if you are at all left of centre. And even if they weren’t they definitely get a big tick in the box next to ‘Don’t Destroy the World – Environment’ which possibly outweighs all the other boxes a thousand times over. And yet, I and lots of other leftists find them really off-putting because they’ve also got a series of boxes saying ‘Genetic Modification: bit sceptical’, ‘Nuclear Power: hate”, “Homeopathy: well maybe it’s got its place” and other weird little signs that are inconsequential in themselves, but overall make people think “Well, what other weird things are these kinds of people going to do?” It’s still probably crazily irrational to assign much weight to these things given the overwhelming importance of the other things, but it does make a kind of sense for limited decision-makers making quick decisions. One could imagine the same thing with Obama/Romney, where you think one of them clearly offers the better policies as stated, but worry that if they also offer some objectionable too left/right stuff, even if it doesn’t matter in itself, this might indicate that they will implement a lot of other crazy actions when in power.

  28. DavidS says:

    I think this depends on what the point of democracy is. If it’s ‘make the best decisions’ or ‘perfectly represent the extrapolated intentions of everyone in the country’, then we’re not onto a winner.

    But key benefits are things like
    1. Allow clear, stable transition of power (the ‘Common Sense’ argument that democracy is far more stable than monarchy)
    2. Provide a mechanism to get rid of people who are noticeably dreadful
    3. Stop any group feeling disenfranchised (and thus not identifying with society at best, revolting at worst)
    4. Produce a decision-making process that is as good as possible within these and other constraints.

    (3) is the most relevant one in this case. As well as people/ideologies potentially being disenfranchised, history sugests that States might also say ‘we don’t feel we have any say, we want to secede’. As long as that’s a thing, you want to mitigate against it. Similar things exist elsewhere: e.g. in Northern Ireland the power sharing arrangement prevents either ‘side’ in the historical conflict simply having all the power.

    The downside of this sort of thing is obviously by treating a certain thing as a unit, you give it more solidity and credibility. US politics to my understanding has more pork barrelling than most European systems, definitely than the UK, and this might be part of the reason why. Similarly, if Parliament had been set up in the UK in 1070 as power sharing between Normans, Saxons and Britons, you might find that these would still be highly charged ethnic/cultural factions…

    (4) could be relevant too: people below have noted that the system also adds some creative tension, which can be a very Good Thing. If you have two houses, you kind of have to have different systems of election/appointment, otherwise it’s just a rubber stamp

    • Kevin C. says:

      I see points like these quite a bit, and I increasingly find them overstated, if not overrated. In order:
      1. In terms of peaceful transitions of power, a democracy is unquestionably superior to dictatorships, juntas, and other modern authoritarian regimes. But how often, really, did hereditary monarchies suffer serious succession crises, and how does that comare to the lifespan of the average democratic regime (recalling, of course, that the USA is a fairly extreme outlier)?

      2. This is the one I find most objectionable; just look at Congressional approval ratings vs. re-election rates. Politicians often retain their seats despite poor performance (see the Curley effect for one of many reasons why), or are voted out for things over which they had little or no influence. The public choice literature makes a pretty clear case that the electorate is pretty bad at judging the performance of politicians and of correctly apportioning blame. Further, ratings of competence are often trumped by concerns such as party loyalty, or the ability to distribute “pork” and other spoils.

      Secondly, with regards to “bad kings”, there are several points to be made. First, there were plenty of monarchs who were fairly awful human beings, and yet were successful leaders, or at least not damaging to their nation as a whole (see, for example, the Vasa dynasty of Sweden, the Ottoman sultans, or most of the more decadent Chinese emperors). From what I’ve read of history, the monarchs who do the most damage tend to be the well-intentioned reformers, who attempt too much too quickly (for example, the Qín dynasty).

      In addition, absolute monarchy was a comparatively small part of history (mainly a product of the declining military utility of both castles and heavy cavalry). Most monarchs have been constrained by powerful lords; see King John, the Magna Carta, and the Barons’ War for one example. German Fritz Kern, in his 1939 “Kingship and Law in the Middle Ages”, describes how in Medieval, and especially pre-Christian, Germanic cultures, the customary Law, rather than the king, was considered sovereign:
      “The Law was regarded as sovereign, so far as any sovereign existed at all, throughout the early mediaeval period. The State existed for the realization of the Law, and therefore the Law was primary — the State only secondary. The monarch’s function was to realize the Law in practice, and he was therefore bound to the Law.”
      And, thus, the people had not only the right to rebel against a king who acted against the law, but a duty, as part of the duty to defend the Law:
      “But since it was the right and the duty of everyone to protect the existing law, in particular to protect one’s own personal rights, it was manifestly a right and a duty to resist the king himself if he were to violate that law or those rights…The king and the people did not simply co-exist as partners in a private-law contract. On the contrary, both were bound together in and to the objective legal order; both had duties to perform to God and the Law. The right of resistance, therefore, was not primarily the right of a party whose contract has been violated, nor was it even exclusively the subjective right of a citizen against an unjust ruler; principally it was a duty of resistance which the citizen owed to the objective legal order which has been disturbed by the ruler, and which is now to be restored.”
      These facts did much to limit the powers of a monarch, and thus the damage a “bad” one could do.

      Thus, the frequency, severity, and unremovability of “bad kings” is frequently exaggerated; and the ability of popular elections to remove “dreadful” politicians is a thing more of theory than practice (especially as more of the functions of goverment fall under the purvew of unelected bureaucracies).

      3. This one is perhaps the strongest, but I would also dispute it’s strength. It might be my upbringing and habitation in Alaska, home of the Alaskan Independence Party, but it seems to me that plenty of people can be found who are “not identifying with society” as a whole despite possessing the miniscule scrap of voice that is the franchise, particularly when they, rightly or wrongly, see themselves as a permanent minority, always out-voted. See also plenty of Scots, Basques, Catalans, Ukrainians, and any number of Balkan peoples. You note that power-sharing mechanisms have developed in at least some places to mitigate against these tendencies, but these seem to me to be more the exception than the rule, and I question their long-term effectiveness.

      Secondly, there’s the degree to which democracy deals with the anger of disgruntled groups not by addressing the grievances, but by channelling that anger into symbolic but ultimately meaningless electoral activity. It works to defuse tensions and prevent more forceful activity, but only so long as the illusion is maintained, only as long as the group is kept from truly realizing how their goals never get accomplished no matter how they vote. And in my own personal circle of acquaintances, I find people on both ends of the political spectrum who are increasingly sceptical of the ability to achieve meaningful results through electoral politics.

      Add in that when party politics comes to increasingly match geographic or racial divisions (see “fly-over country”/”Jesusland” and the increasingly white Republican party, for example), it can come to exacerbate divisions and tensions, rather than disperse them.

      Further, this ignores the ability of a monarch, as a non-partisan head of state, to unify groups and provide common identity. See, for example, Thai politics, or Belgium.

      • Douglas Knight says:

        Scott wrote about this here and here.

        I dispute point 1.

        The canonical example of a good king with a notorious personal life ought to be Leopold II.

  29. Jack V says:

    Yeah. With pro-and-con lists I make personally I assume the answer is somewhere between “adding up the lists” and “choose which item is most important and decided based on that one” and once you’re thinking like that it’s usually obvious which.

    But I should do that more with national politics.

    I like the bicameral metaphor, though I’m worried it’s too much overhead to explain to someone. But maybe having a name for the concept is useful enough it’s better to have it.

  30. Peter says:

    Where’s Gerd Gigerenzer when you need him? He’s got this nifty set of heuristics, in particular he’s got this pair. In “Take the best” you find the most important single criterion that distinguishes the two options under consideration, and go by that criterion. There’s another one where there’s a Senate style vote where each issue gets a single vote. His point is that in loads of cases, at least one of these two models gives pretty good results and that unlike more complicated methods they don’t have lots of fiddly parameters to set. (Oh, erm, I see JackV above making more-or-less the same point). Anyway, what would he say about this? It’s been a while since I read his work…

    [It seems I’m a chemist. I’m wondering what a critter ion is…]

  31. Tarrou says:

    Not that the utilitarians will care, but what all this weighing of options misses is two things:

    1: The limits of pre-knowledge. We don’t know the actual results of an action until long after we take it. Sometimes info is bad, sometimes consequences are unforseen. Doesn’t mean we shouldn’t try hard to havethe best decision making possible, and simple democracy hasn’t been shown to be superior in any way to more complicated republican systems.

    2: Blame attribution! It’s not just X number of deaths (or electricities, or utils), it is who is responsible for them. A thousand car crashes are mostly not intentional, and the blame is divided between a thousand (or more) people. A single attack is attributable to at least one group, usually one man, and crucially, is intentional. Utilitarians can deny the basic human intuition that intent matters, but the public will not be convinced. Murder and accident trigger different responses, and always will. And should.

    Scott shows this problem of attribution in his mistaken characterization of the invasion of Iraq as having killed hundreds of thousands of people. The actual invasion killed a couple thousand, perhaps. The occupation killed some tens of thousand more, most of whom were actively trying to kill the occupying forces (understandable on both sides). The issue is that the occupation failed to stop the Iraqis killing hundreds of thousands of their own people, which shifts the conception of blame and what we should have known somewhat. There is plenty of blame to go about, but failing to stop a violent people from massacring each other isn’t quite the same as committing genocide.

    • John Schilling says:

      A further complication is that the invasion did, predictably and deliberately, break the mechanism that was effectively preventing a violent people from massacring each other. A mechanism whose operation involved brutal oppression and tyranny built on a legacy of past massacres, and empowered a militaristic dictator with the means and inclination to threaten the world with far worse in the future, but which was not ca. 2003 killing Iraqis at anything like the rate the subsequent insurgency would.

      And, of course, the invaders had a plan to set up a mechanism that would prevent the massacres without all the other unpleasantness. One which some people beforehand and many in hindsight recognized as little better than wishful thinking, but which the invaders seem to have sincerely and perhaps not unreasonably believed would work.

      • Tarrou says:

        I find nothing to disagree with here…….

        Quick, someone misconstrue something, create an argument ad absurdum and start name calling! 😛

  32. Ben Kennedy says:

    It doesn’t convey the right sense of “Why are you even looking at that other box?”

    Welcome to the world of the committed pro-life voter, where if you think that a million people are being killed every year just in this country, other issues just don’t seem that important

    • Evan Þ says:

      Well… That’s still assuming that there are actual pro-life politicians who would save substantial numbers of babies if they got into office. How many unborn babies did Bush actually save, personally? Sure, several state legislative majorities have done a lot of work, but Bush himself? Versus (for example) how many people got killed in Iraq?

      But these details aside, you’re exactly right in principle.

      • suntzuanime says:

        The key issue on abortion from a federal perspective is the Roe vs. Wade precedent, and the main thing presidents do with regard to it is nominate justices who are more or less willing to agree with that case’s penumbral logic. Given the constraints Bush was working under, that anyone who openly disagreed with the holding would never have been confirmed, I think he did pretty well. Yes, no children have been saved yet by Roberts and Alito, but their presence on the court makes it easier for a future anti-abortion president to finally achieve a majority.

        EDIT: Actually, I think the Supreme Court with Bush’s nominees on it has upheld certain forms of restrictions on abortion that a Court with more creatively-interpretive Justices might not have, so it’s not even clear that he hasn’t saved any children, even if he didn’t quite manage to shut down the slaughterhouses altogether.

        • ddreytes says:

          I’m fairly sure the relevant case on abortion these days is not Roe but Planned Parenthood v Casey, which – among other things – disavows the penumbral logic of Roe (although it still signs on to the logic of substantive due process).

          Both of Bush’s nominees would (almost certainly) be completely willing to overturn Planned Parenthood and/or Roe, although due to the way the procedure of the things work these days neither of them would have explicitly said so (not because it’s politically impossible; there’s no requirement that you have to agree with the logic of Planned Parenthood to get appointed, it’s just the way that Supreme Court confirmations are done nowadays, after Bork). But they absolutely would vote that way. The reason that Planned Parenthood wasn’t overturned is because Bush’s appointees are still in the minority, because (essentially) David Souter and Anthony Kennedy both ended up being votes to defend abortion despite having been appointed by Republicans. And the fear of that happening again was probably the single greatest thing informing Supreme Court nominations in the Bush era – it’s one of the main reasons the Harriet Meiers nomination fell through, because she didn’t have a strong track record of conservative views to reassure conservatives that should would be a reliable vote on Roe, whereas Roberts and Alito do have that record.

          So he might not have saved any children (from the pro-life point of view) but only because of the luck of the draw: he didn’t get to replace enough Supreme Court members to get a majority on the bench. They’ll probably get there if a Republican wins in 2016, though.

          • Randy M says:

            “not because it’s politically impossible; there’s no requirement that you have to agree with the logic of Planned Parenthood to get appointed, it’s just the way that Supreme Court confirmations are done nowadays, after Bork”

            Pretty sure when he said politically impossible he didn’t mean any statue on the books, but the democratic party/media would have caused created a public controversy over any such remarks, and activists would have caused enough personal irritation to make dissuade candidates from being overly honest.

          • ddreytes says:

            @ Randy M:

            Yes, that’s true. I guess my point was that’s true regardless of your opinion on Roe V Wade (or any other issue). Republican activists would do the same thing to Democratic candidates. Or, at least, the possibility would exist, on both sides, and so that’s the way it’s done these days, because it’s just less risky. That’s the process.

            Like… it’s not particularly impossible for someone who openly disagreed with Roe V Wade to get appointed. It’s difficult for anyone with particularly controversial views to get appointed. And there are people who get appointed who do disagree with Roe V Wade, and everyone pretty much knows it.

          • suntzuanime says:

            Yes, I did not mean to imply that the confirmation process is only participated in by Democrats. I was just making excuses for why I could not point to Roberts and Alito as clearly stating their opposition to the ban on abortion bans.

  33. Vanzetti says:

    Scott, it is time for you to accept that innate human morality is a product of evolution. It cannot be made into a coherent world-encompassing system. Therefore, NO AMOUNT of reshuffling ideas in your head will make you STOP FEELING BAD about the system. The feeling is itself based on your genes.

    You just can’t build the Eschaton out of raw monkey neurons. It’s absurd. You (and I) are going to feel bad about those moral paradoxes right up to the point where we literally reach into our brains (be it by uploading or genetic engineering) and make it stop feeling bad.

    • onyomi says:

      Or there are such things as objective right and wrong, but there is no single, objective criterion by which we may judge them. And since thinking otherwise always results in contorting logic to fit one’s intuitions, why not just be an intuitionist?

      • Vanzetti says:

        >Or there are such things as objective right and wrong, but there is no single, objective criterion by which we may judge them.

        Isn’t this just an example of the invisible unicorn? If you have no way to know what they are, why suppose they exist at all?

        • Jaskologist says:

          The unicorn is not invisible. We can see it, it’s just that we only see it through a glass, dimly. Imperfect knowledge is still knowledge.

          • Vanzetti says:

            >We can see it, it’s just that we only see it through a glass, dimly.

            What you see through the glass is something vague. Your monkey brain pattern matching, honed by millions of years of evolution, turns it into a unicorn.

            There is no unicorn.

          • onyomi says:


            So, on your view, what, exactly, are we saying when we assert “x is good” or “y is bad”?

            Does “x is good” just mean “x makes me happy”? Does “y is bad” just mean “y makes society sad”?

            And if so, why is it coherent to say something like, “x makes me sad, but it’s the right thing to do”? or “indulging in y makes people happy, but it’s the wrong thing to do”?

          • Vanzetti says:

            >So, on your view, what, exactly, are we saying when we assert “x is good” or “y is bad”?

            It depends on specific person, time and situation, of course.

            >And if so, why is it coherent to say something like, “x makes me sad, but it’s the right thing to do”? or “indulging in y makes people happy, but it’s the wrong thing to do”?

            It’s not coherent, for the simple reason that our monkey brain isn’t coherent. It was “designed” by the blind idiot god of evolution and basically consists of nothing but hacks upon hacks.

            That’s why it’s hopeless to try and establish a perfect morality system for humanity 1.0.

          • onyomi says:

            But it IS coherent to say “it makes me sad to do x, but it’s the right thing” in a way that it’s not coherent to say “it makes me sad to do x, but it makes me happy to do x.”

            If it depends on the specific case, then when someone says “murder is bad,” what are they really asserting? “Murder makes me sad”? When someone says “helping the less fortunate is good,” what are they really proposing? “Helping the less fortunate makes me happy?”

          • Vanzetti says:

            >But it IS coherent to say “it makes me sad to do x, but it’s the right thing” in a way that it’s not coherent to say “it makes me sad to do x, but it makes me happy to do x.”

            It is incoherent in the sense that you have contradicting emotions about doing something, so you are going to feel BOTH good and bad about it.
            Or perhaps you will just feel bad about something, and yet feel compelled to do it anyway, which is even more incoherent.

            >If it depends on the specific case, then when someone says “murder is bad,” what are they really asserting? “Murder makes me sad”?

            It depends on the specific case. 🙂

          • onyomi says:

            Okay, so you think statements about right and wrong are really just statements about emotions?

            What, then, does the statement “it is good to feel good” mean? “Feeling good is feeling good”? What if you were to say “it is not always good to do what feels good”? Would that not then become “it does not always make one feel good to do what makes one feel good”?

            Most people are able to make judgments about “good” and “bad” without necessarily feeling any emotion, and when they do so, they perceive themselves to be asserting propositions, not just expressing feelings. There is a difference between “yay charity!” and “charity is good.”

    • I haven’t seen Scott deny that the fact of human morality comes from evolutionary forces. Running, logic, singing etc. also come from evolutionary forces – should we stop trying to perfect those too? I think considering moral systems is a very worthwhile use of Scott’s time (and incidentally my own time). And even if you don’t feel strong morality yourself, then it’s still worth considering that the evolutionary of morality, which helped, along with a cluster of other human traits, us to build this amazing human civilization with all its technology and other wonders. If you like them, maybe its worth supporting the perfection of morality for the instrumental benefits it brings.

  34. Mary says:

    The advantage of the Senate is exactly that it curbs the ability of large numbers of people to do what they want. History having shown that what they want to do is often as tyrannical as it gets.

    “Democracies have ever been spectacles of turbulence and contention; have ever been found incompatible with personal security or the rights of property; and have in general been as short in their lives as they have been violent in their deaths “

  35. Gbdub says:

    To me, the big issue of the “check box” issue framing that you miss is that it doesn’t weight how much impact a candidate can have on the issue. The real value of candidate A vs. candidate B is not how big the issues are, but how much utility they can produce on the issues.

    So Kyoto might be a big deal, but it’s also irrelevant because no matter what the POTUS believes, Kyoto and similar agreements are doomed without buy-in from India and China. “The Economy” was a huge issue and always is in the Presidential race, but the POTUS has only minor control over the economy.

    Also, the checkboxes tend to list current positions and past votes, which are only an approximate gauge of their future positions. Obama was against Iraq, but happily bombed the bejeebus out of Libya. This is probably some combination of pandering to his political base re: Iraq and things looking different from the top. Check boxes do not capture this.

    A position doesn’t equal effort – if McCain favored NCLB but was unwilling to expend political capital defending it, it changes the weighting. If Obama was against Guantanamo Bay….

    Finally, the check boxes often reward hindsight. The 2008 election involved something of a referendum on Iraq, which was stupid because the decision to invade had already been made. The dead people were already dead. I’d argue that Obama’s opposition to the initial invasion actually caused a bad outcome, even if he was right: it pushed him to GTFO in a hurry, leaving things ripe for ISIS to rise.

    Regarding the bicameral system, it’s only weird if you don’t think of he States as valid political entities in their own right. When in the past a belief in some state autonomy was definitely the case, and now I think probably ought to be more of the case again. More varied states allows for more experimentation and foot voting. I’d rather not subject the whole country to the whims of Californa, given their historical performance in their own backyard (or you may not want the whole country run like Texas).

  36. Tim M says:

    It seems to me that weighing New Yorkers against Delawareans is a categorically different thing to weighing chickens against people, cows or the environment.

    The problem with chickens is that you have no way of quantitatively comparing the value of one to the value of a person. How many chickens should be equivalent to a person? Well it’s probably not infinite, but on the other hand, it’s hard to reconcile sacrificing a person for any finite number of chickens.

    As for New Yorkers and Delawareans, it is relatively easy to compare compare their value. I am assuming that you consider them to be of precisely equal value* and that you think they should have an equal say in a fair democracy. In which case, if your criteria for judging these things is democratic fairness, then your temptation to say “the House is just plain right and the Senate is just plain wrong” is one you should give in to, and the Senate system is just a sop to Delaware and other small states.

    On the other hand, if you are interested in which system delivers the best social outcomes, then you will have to dispense with democratic purity and look beyond this sort of reasoning into the practicalities of how these systems work.

    * At least to a first order approximation, ignoring any externalities that they might impose on the rest of the world.

  37. This gets brought up by people like Gigerenzer (I first read it in ‘Simple Heuristics that Make us Smart’) pointing out that simple heuristics and decision rules actually do a pretty dang good job of getting the right answer. The explanation they give is that people are very good at recognizing what variables are important, but very bad at integrating that information into a decision – so bad in fact, that unit weights can do better than individual judgment.

    Haven’t looked at it super closely, but here’s a highly-cited relevant paper:
    The Robust Beauty of Improper Linear Models in Decision Making

  38. Jaskologist says:

    The good news is that tribalism routes around the problem entirely! If you agree with Candidate A on a few of your major issues, you probably end up agreeing with him on most of your others, too, because we work hard to bring our views into alignment with the rest of the tribe. That’s why these scorecards are released in the first place; they’re stacking up issue after issue where your Candidate agrees with you. They can implicitly assume that you’ll want to check column A the vast majority of the time.

    • houseboatonstyx says:

      @ Jaskologist
      The good news is that tribalism routes around the problem entirely!

      That may be a reasonable way to decide who to vote for, since a candidate is more or less of a package deal. (Which is an advantage of representative democracy over direct democracy, since a politician needs to show some consistency in what zie will vote for. With direct democracy, each referendum can be won by whichever side of that issue does the best advertising, period.)

      But it sure did work around Scott’s original question. For that, choosing which charities to donate to would be a more useful topic.

  39. Held In Escrow says:

    The interesting thing about how we weigh issues is that it explains why public opinion doesn’t have a huge effect on public policy. Mainly because people do vote based on their heavily weighted criteria and tribal affiliations rather than the smaller issues.

    There was a study a bit back that many Blue Tribes were trumpeting to the heavens about how Corporations have a million times more influence on public policy than how people actually feel. But if the parties have segregated themselves so that they can capture the big issue votes from both the Pros and the Cons, well, it doesn’t really matter what their stance is on the small issues.

    Let’s take a highly polarizing issue that’s also heavily weighted in the public conscience and call it Skub. Let’s start with 60% of people wanting to regulate Skub more and 40% wanting to regulate it less. Now, Skub, assuming that 51% is enough for a change, is going to continue to be more regulated until the regulation gets to the point that 50% of people think it’s perfectly well regulated. Thus you end up with half the country being pro-Skub and half the nation being anti-Skub, and it’ll be shown in the political parties.

    Now let’s take a much less weighted issue, such as subsidized private islands for corporations. Maybe 99% of the country is against this. But a politician isn’t going to lose any votes in a two party election for being for subsidized private islands as everyone is voting based on his Skub affiliation, so he can take the corporation’s campaign contributions and vote for the subsides. Thus the corporations get to build Jurassic Park and Skub remains stalemated.

    Granted, this is where primaries come in (and what caused the Tea Party insurgency), but for the most part your parties are going to be very hesitant to primary their own candidates.

    Also, having worked as a lobbyist for some time and knowing plenty of people on the Hill, the Senate might be an unrepresentative clusterfuck… but it’s tree(3) times better than the black hole of ignorance that comprises the House.

    • Evan Þ says:

      “having worked as a lobbyist for some time and knowing plenty of people on the Hill, the Senate might be an unrepresentative clusterfuck… but it’s tree(3) times better than the black hole of ignorance that comprises the House.”

      I’m interested – could you elaborate? What causes that ignorance, and why’s it stronger in the House?

      • I’m interested – could you elaborate? What causes that ignorance, and why’s it stronger in the House?

        I don’t completely agree with the original contention, but yes, certainly members of the U.S. Senate on the average are smarter, more knowledgeable, more articulate, etc., etc., than members of the U.S. House.

        (1) Senators are elected from whole states, which of course are almost always much larger than individual congressional districts (for electing members of the House), and hence candidates for Senate get a whole lot more scrutiny from the media and public.

        In a Senate candidate, deficits of knowledge or skill are more likely to be noticed AND affect the election outcome.

        (2) Senators are elected for six-year terms, which allow them to focus more on governing. Members of the House are elected for two-year terms, which force them to focus much more on getting re-elected.

        (3) Congressional districts are changed every ten years to maintain equality of population. Districts do not get to settle in to becoming self-aware communities, the way every state is.

        For example, if your state elects a senator who is an embarrassment, everybody knows it and feels partially implicated. But if some district elects a representative who is an embarrassment, not even the voters in that district are held responsible. Very few people know the boundaries of their own congressional district, let alone others!

        (4) Congressional districts are drawn by people who are intensely interested in the outcome of elections. There is a strong incentive to create “safe seats” in which the dominant party need not exert much effort to win and needn’t be too concerned about the candidate’s qualities.

        By contrast, no U.S. state, no matter how “red” or “blue”, is completely “safe” for either party. Hence, in statewide elections, both parties want to nominate strong candidates that will appeal to independents as well as the party’s base.

        (5) Every field has a shortage of really good people, and politics is no exception. It’s a lot easier to fill 100 senate seats with the most top-notch available politicians than it is to fill 435 congressional seats.

        Moreover, ever since the Watergate scandal of the 1970s, which discredited politics as a career in the eyes of many, as well as changes in the American workplace which strongly discourage workers from developing time-consuming outside interests, there has been a shortage of well-qualified candidates for elected positions, in both parties, at all levels, in every state.

        Gerrymandering is not the only reason that most congressional seats are “safe”: in most cases, the less-advantaged party is unable to field a strong candidate. That’s because the people who ARE good candidates know they are in short supply, and have a strong bargaining position vis-a-vis party leaders. They tend to be unwilling to throw themselves into races against incumbents or in less-than-friendly territory, where they will probably lose.

        But politics is not ALWAYS predictable, and sometimes district-by-district outcomes are not what the media and political professionals expect. In those cases, you end up with one party’s “sacrificial lamb” (a candidate thought to have no chance, who gets little vetting) actually winning.

        And — guess what — those un-scrutinized people are even less likely than the safe-district party regulars to become outstanding, knowledgeable, articulate members of Congress.

        • Anonymous says:

          There is some tension between (2) and (4).

          • ddreytes says:

            I’m not sure there is.

            I think there’s a distinction between being an electable candidate (in a competitive election) and spending a lot of time on getting elected. (2) is a factor that determines how politicians spend their time; (4) is a factor that determines who is selected to be a politician. And it’s not a thing where the two trade off for each other – the always-an-election-year, even in a safe House seat, is a lot different from being a Senator.

          • There is some tension between (2) and (4).

            Of course there is, and I didn’t even mention primaries. General elections are not the only time you’re at risk for losing your seat.

            There is an inherent tension between the interest of incumbent politicians, who want the safest possible district to get re-elected in, and their political party, which wants to win the largest possible number of districts.

            If the person drawing the lines is maximally partisan, and completely disregards the interests of her own party’s incumbents, her strategy will be to create a few ultra-safe districts for the other party, so that her party can win all the others with 55%.

            This doesn’t usually happen quite so starkly, since incumbent members of Congress usually have enough clout to prevent being put into a district that might endanger their re-election in a bad year for their party.

            And any time both parties compromise on a plan, or a court orders some kind of “consensus” plan, it ends up protecting both parties’ incumbents with safe seats.

            The other problem is geography, and the growing tendency for liberals and conservatives to cluster themselves in like-minded communities.

            Democrats often win a majority of the total votes for U.S. House candidates, while Republicans win a majority of the seats, not just because of partisan districting. A lot of Democrats are heavily concentrated in urban districts, whereas Republicans are more efficiently spread out across many suburban and rural districts.

        • Anthony says:

          By contrast, no U.S. state, no matter how “red” or “blue”, is completely “safe” for either party. Hence, in statewide elections, both parties want to nominate strong candidates that will appeal to independents as well as the party’s base.

          While that’s not quite true, in states where the Senate seat is safe for one party, a truly embarrassing Senator will get primaried, because the risk of a divisive primary handing the seat to the other party is very low.

          • As to my comment that no U.S. state, no matter how “red” or “blue”, is completely “safe” for either party:

            While that’s not quite true

            Oh? In recent years, deep-blue Massachusetts and Illinois have elected Republican senators. Meanwhile, deep-red Nebraska and Alaska have elected Democratic senators.

            If it could happen in those states, it could happen in any state. Of course, it’s not an outcome to bet on.

            a truly embarrassing Senator will get primaried, because the risk of a divisive primary handing the seat to the other party is very low.

            Rather, a truly embarrassing Senator might get a primary opponent, regardless of the state’s political coloration, because the risk of losing the seat altogether will be elevated.

            In any case, the people of the state are the ones to judge whether a Senator is “embarrassing” or not, and by that standard, it doesn’t happen often.

        • SanguineVizier says:

          By contrast, no U.S. state, no matter how “red” or “blue”, is completely “safe” for either party.

          I disagree with this claim. Putting on my Popper hat, what, to you, would constitute evidence against this statement?

          • (See also my response to Anthony, above.)

            If there was a state where one of the major parties was so weak that none of its candidates had come within 5 points of winning a statewide election in the last 20 years, sure, I would concede that state is safe for the dominant party.

            Is there such a state?

            And no, the District of Columbia is not a state.

          • Alex says:

            Not sure about the exact numbers here, but Washington State has not had a Republican Senator since 2001, and a Governor since 1985 (although the 2004 election was pretty close). This place is pretty safe for statewide races if you write a D after your name.

          • SanguineVizier says:

            Larry Kestenbaum,

            To the best of my knowledge, no state meets your requirements for being safe. The best I had, and the one that meets my requirement for being safe, is Utah. It falls just shy, as there has been exactly one Democrat hold statewide office (attorney general, an office held by Republicans since the 2000 election) within the past 20 years there, and none of the losing Democrats for state offices have come within 5 points within that span (so says a quick, but not exhaustive, Wikipedia check). With the sole exception of attorney general, a Democrat has not won a statewide election since 1976. In any case, I predict that after the 2020 elections, Utah will be a safe Republican state by your standard.

            I consider Utah a safe Republican state as I would be willing to give very generous odds in betting on statewide elections in Utah, knowing nothing but the candidates’ party affiliations.

          • Yes, Utah comes very close to qualifying even by my admittedly stringent criteria. Also Texas, where a Democratic nominee came within a couple percent of winning the lieutenant governor’s position in 1998, but nothing since.

            But those are plainly a couple of extreme cases. Look at Oklahoma, where no county has ever voted for Barack Obama, but which easily re-elected a Democratic governor in 2006. South Carolina and Mississippi have elected Democrats to statewide office since 2000, and Maryland, Massachusetts, and Hawaii have elected Republicans. New York City has elected Republican mayors. And the 2004 Washington State gubernatorial recount shows that Republicans can be competitive there.

            Part of this is that every state is a kind of community which (to varying extents) has an ongoing conversation about candidates in statewide races. That makes it possible to occasionally depart from just mechanically supporting the majority party’s nominees for every office.

          • SanguineVizier says:

            Certainly Utah and Texas are extreme cases. Any state would have to be well out of the mainstream to even be in the discussion for being a safe state for a particular party. It is interesting that the two closest cases of safe states are among the most culturally distinct states.

            It seems that when a state becomes particularly red or blue, two things might happen. The state party might affiliate with a different tribe than would be expected from the national party (red tribe Democrats, blue tribe Republicans). In that way, a party can continue to stay relevant in statewide races. I would venture that a Democrat who polls well in Texas is more red tribe than a Republican who polls well in Massachusetts. This is probably the typical case, and keeps the number of safe states quite low, if it is even non-zero.

            The other case, which is what I actually foresee in Utah, is that one party becomes so dominant that everyone with real political ambition in the state joins that party as a matter of expediency. This second path need not be exclusive of the other. I would wager that the average Democrat in Utah is quite conservative on the national scale, but they just cannot seem to be competitive, and I would be surprised if the truly ambitious keep bashing their heads against that particular wall.

  40. Estimates for the social cost of carbon (the damage done by CO2-induced warming divided by the amount of carbon doing it) are not that high.

    • James Picone says:

      I’d be happy with governments imposing a ~$40 US/tonne of CO2e Pigovian tax on emissions (Which is the central estimate of the social cost of carbon in that website, although they claim it’s probably higher than that because of factors they don’t include), through whatever mechanism captures most of it (So probably a tax on petrol, a tax on coal, a tax on natural gas, and some mechanism for land use changes/agriculture).

      I’m not aware of any government in the world aiming to establish laws quite that strong. British Columbia’s tax is approximately $30/tonne, according to Wikipedia.

      A government gets my vote for Doing Basically The Right Thing Here if they impose such a tax, make it revenue neutral, eliminate any fossil fuel subsidies, and tie the tax rate to estimated social cost of carbon at the time via some mechanism (SCC goes up over time).

      • James Picone says:

        More interesting stuff – there was recently an IMF white paper calculating fossil fuel subsidies worldwide. Stoat had a look at it here and points out that the $10 million/minute figure people were quoting from it includes untaxed fossil fuel externalities and is not exactly what ‘subsidy’ usually means.

        Using the figure from that paper we get ~5500 billion USD/year ‘post-tax subsidies’ for 2013, which are mostly untaxed externalities. Wiki says 35,270,000 kilotonnes of CO2 were emitted in 2013. That works out to 155 USD/tonne, which doesn’t match the social cost of carbon from above…

        Except as Stoat noted (pointing to another blog), only about a quarter of those externalities are climate-change related, most of them are local and are related to coal power being incredibly friggin’ dirty. Divide by four, you get $38.75/tonne, pretty damn close to the SCC calculated by the folks above.

        I wonder whether the IMF paper used that work above as an input or whether it’s just that the numbers are the same.

  41. Bryan Willman says:

    The post is also missing some of the realities of politics. If simple “majority rules” is best, and arbitrary lines are bad, why don’t China and India rule the rest of the world? Furthermore, why don’t the poorest 51% of the world’s population get to vote to impose a 100% wealth tax on the other 49%?
    Whether it’s *fair* that Delaware gets 2 seats in the senate has to be offset by whether it’s *fair* that large populations in CA, IL, NY, should get to set policy in Wyoming.

    As for elections and wars – be vary wary of presuming the better counterfactual would actually happen. That is, Mr. X opposes wars, Mr. Y is aggressive. Mr. X. wins the election. You may still end up in a *worse war* than before. Example – Mr. Obama opposed the prison at Guantanamo. About half way into his second term in office it’s stil there, with no signs of speedy resolution. I don’t think he’s changed his position, I think reality intruded.

  42. aretae says:

    But I’m pretty sure those only kick in at the extremes. Take it too far, and you’re just saying the life of a Delawarean is worth twenty-something New Yorkers.

    The claim is that politically, we are not sitting in an epistemically privileged position, and that majority rule has a tendency to run rampant over the rights of the minority.

    In the absence of both epistemic privilege AND credible protection for the rights of the (state) minority, you put in checks on majoritarianism.

    Pure House is great, if you’re going to be _right_ most of the time. If you’re expecting error, then putting in a voice that will oppose the majority will frequently is a real win.

  43. Alex Trouble says:

    When the country was founded, the states were supposed to be the primary government entities, not the federal government. The point of a Senate–which has been lost now that we directly elect Senators–was to give the state legislatures power in the federal government. The feds weren’t supposed to be able to rule over the states like they do now; they were just supposed to be able to keep peace between them, mobilize an army, and a few other things. In that context, it makes sense for Delaware and NY to have the same number of Senators.

  44. David says:

    Hi Scott,

    This is my first comment on your blog (although I’ve been reading your blog for about 2 years) so I’ll start by saying that I really, really, really, really like it. 🙂

    But sometimes you write things that make me remember that smart people often believe the craziest things. I find this whole assigning of values to the lives of chicken and comparing them to the “assigned” values of humans to be complete BS. Really. If I were to meet your friend Buck I would just offer him a deal (not really, but I would offer a hypothetical deal at least), I would save the lives of a thousand chicken but in exchange he would have to take his own live. Do you think he would accept? If he truly believes that the life of a human is less valuable than the lives of 1000 chicken (or a bigger number if that is what his math is telling him), it should be a no-brainer. Or he simply believes that those thousand chicken lives are more important than the life of, OTHER humans?
    The same thing with the fat man trolley thing. Don’t ask yourself if you have a moral duty to kill a fat man by pushing him on the rails to save the life of 5 other people, ask yourself if you have a moral duty to jump YOURSELF on the rails and die in order to save those 5 people. After you have answered that question you can think about what to do with the fat guy.
    I’m sorry but religion has figured out the solution to these sorts of “problems” a long time ago, it’s called The Golden Rule. Don’t treat others in ways you would not want to be treated yourself. Follow this rule and you can keep your sanity.

    And you know what’s a reasonable way to think about the “death of chickens problem”? Sure chicken have a non-zero value and simply going around killing chicken randomly and for no reason is a bad thing. But a chicken that dies in order to feed a hungry person serves a good purpose. A chicken that dies of old age serves none.

    P.S. Please excuse my English, I’m not a native speaker.

    • Buck says:

      Hey there. If we talk about suffering instead of death, because it’s easier to compare between species with different lifespans etc, then if we manage to standardize some amount of pain and fear between humans and chickens–say, the experience of having something as sensitive as a beak chopped off–then for sure I’d accept that deal. (Unless I thought that I was going to lose enough productivity from the experience that I can do enough good to make up the difference.) Even if you weigh suffering linearly by brain size, 1000x is still a no-brainer.

    • Zykrom says:

      Other-animals vs self-humans isn’t a fair contest. You’d have to make it either a choice between his chicken-self, which he doesn’t have, and his human-self, or between other chickens and other humans.

      • Randy M says:

        You mean he might not have a chicken self, right? You wouldn’t want to assume he isn’t a henkin, I hope.

    • Anonymous says:

      ask yourself if you have a moral duty to jump YOURSELF on the rails and die in order to save those 5 people.

      Suppose we apply the “veil of ignorance” idea here. You are in a group of six people and you are told that five of you are going to be placed in front of the trolley. The sixth person (chosen at random) will be given the option of sacrificing themselves to save the other five. Would you sign a binding contract with the group that requires the sixth person to sacrifice themselves to save the rest?

      I’m sorry but religion has figured out the solution to these sorts of “problems” a long time ago, it’s called The Golden Rule. Don’t treat others in ways you would not want to be treated yourself.

      The golden rule seems to strongly support jumping in front of trolleys to save others! I would be extremely grateful if someone jumped in front of a trolley to save me.

      • disciplinaryarbitrage says:

        We could also apply a similar logic to the chickens vs. people question to remove the self-vs-other complication: would you rather take your own life now (and just be dead), or live out the rest of your natural life, but then be reincarnated 1,000 times as a factory-farmed chicken?

        (Sure, this requires some heavy-duty magical handwaving w/r/t the notion of continuity of experience between differently-embodied consciousnesses, but that’s the beauty of thought experiments?)

        • Jiro says:

          This conflates questions of “I wouldn’t want to be a chicken because chickens suffer” and “I wouldn’t want to be a chicken because the life of a chicken is intrinsically low in value”. I wouldn’t want to be reincarnated as even a billion free range chickens, or oak trees or rocks.

  45. Gbdub says:

    What’s so great about “one person, one vote” anyway? Obviously it’s easier to get buy-in with that level of superficial equality, but what’s the evidence that it’s the most “fair” or leads to the beat outcomes? Maybe there are some issues where 1 Delawarean really should be worth 20 New Yorkers.

    Is every person’s vote being worth the same fair? A person with a lot of kids obviously has a strong incentive to care about public schools – shouldn’t their opinion on schools matter more than a childless person’s? Should someone who pays no income tax have an equal say in taxation as someone in a high tax bracket? And so on and so on- there are many issues with minimal effect on most people and a strong impact on others, but everyone gets only one vote.

    Even with “one person, one vote” it’s not like everyone’s vote is valued equally anyway – seats are distributed based on total population, but minors (and some others) can’t vote. So people in districts with a lot of minors or ineligible voters have outside influence. Plus many people live in districts that are “safe” for one party, or have a strong incumbent. Their vote is much less influential than that of someone in a contested district.

    • onyomi says:

      Another problem with the concept of “one person, one vote” is that it creates a bad incentive.

      Consider Ayn Rand’s story of the collectivized factory in Atlas Shrugged: the predictable result of people receiving money in proportion to their “need” was that people were incentivized to have very large families, but to work as little as possible.

      In a democracy, for example, if family a has two children and educates them very carefully to be good citizens and informed voters, and family b has six children, but expends little effort, then family b more than cancels out family a’s votes. What’s more, if there are social welfare programs and family b does not instill a work ethic in their children, then family b’s six children amount to a claim on the work of family a’s responsible children. And guess who’s going to vote for the social welfare programs?

      “From each according to his ability, to each according to his need” results in everyone mysteriously becoming very needy and incompetent.

    • Anthony says:

      I’ve advocated for “one dollar, one vote”, where the dollars measured are your tax payments net of tax receipts. If your get more from the government than you pay, you don’t get to vote (it’s a conflict of interest).

      The primary effect of this would be to make the government more responsive to the upper-middle-class and less so to the really rich. It would probably also end up increasing taxes on the really rich, but not much, while decreasing taxes on the UMC, but not much.

      • Jiro says:

        I would oppose this unless the people who don’t get to vote also aren’t affected by laws made by politicians who get voted for.

        If you get more from the government than you pay, you can still get arrested.

        (Or you can do it in reverse, and say that anyone who can vote for someone less likely to arrest them has a conflict of interest. If you can prove you have practical immunity from being arrested, you get to vote.)

      • onyomi says:

        I agree that there are huge conflicts of interest for everyone from the CEO of a company that gets government contracts, all the way down to your local postman.

        The problem is, there is an idealistic quality to the idea of “one person, one vote” which many of the proponents of democracy would be loathe to give up, even in cases of obvious conflict of interest, as with, for example, public sector unions donating money to political campaigns of politicians who make sure they keep getting the funding to donate to them.

        But if you’re going to go so far as to suggest “one dollar, one vote,” which is never, ever going to happen in a democracy, why not just go all the way to anarcho-capitalism and say “everybody gets the government they pay for–quite literally”?

      • Anonymous says:

        (it’s a conflict of interest)

        I don’t think the idea of a “conflict of interest” makes sense in voting. If we only want disinterested parties to vote, you should only be allowed to vote in country X if you are NOT a citizen of X. Anyway, people paying positive taxes have exactly the same “conflict of interest” as people paying negative taxes.

        • onyomi says:

          There’s a difference between “conflict of interest” and just “interest.” Even people who are not citizens of the US are affected, albeit usually indirectly, by what the US government does. They have an “interest” in what the US government does.

          Most Americans have a conception of the government which says it is there “to promote the general welfare.” It is a conflict of interest if you are disincentivized to vote for “the general welfare” by a personal interest of your own. Of course, this applies to a very large percentage of the population today, maybe even most people, to one degree or another, but that’s only because the government is involved in so many aspects of our lives.

          • Jiro says:

            It is a conflict of interest if you are disincentivized to vote for “the general welfare” by a personal interest of your own.

            That would equally apply to “I don’t want to be taxed, so I might object to taxes that hurt me even if they benefit society” and “I don’t want to be arrested, so I might object to laws that hurt me even if they benefit society”.

            Also, it would mean that a gay person who votes for gay marriage has a conflict of interest.

        • Irrelevant says:

          I don’t think the idea of a “conflict of interest” makes sense in voting. If we only want disinterested parties to vote…

          No, we only want interested parties to vote. The standard term here is “skin in the game.”

        • Anonymous says:

          @oynomi, Irrelevant:

          OK, but for example everyone has an incentive to vote themselves a lower tax rate and more services at the expense of the general welfare. Everyone has a “conflict of interest” with the general welfare.

          Perhaps I’m just arguing terminology; I think Anthony should have just said that people paying negative taxes don’t deserve to vote.

      • “tax payments net of tax receipts”

        Define “tax receipts.”

        Do dividends from stock in a company protected by a tariff count? A company that has government contracts? The government paying you $100,000 for land you could have sold to someone else for $99,999? The government paying you $900,000 for land you could have sold for $1,000,000, with the threat of eminent domain preceding if you don’t agree?

        • Anthony says:

          ok – I haven’t worked out all the details, but:

          More like 1 vote per thousand dollars net paid in taxes by the *person*. Calculate over 2-year congressional election cycle. Round up. If you don’t want to completely disfranchise people, you can give everyone a minimum of one vote, and extra votes based on taxes paid.

          Paid: all documentable taxes paid directly by you – income taxes, property taxes, sales taxes. Taxes hidden in the price of something you pay for don’t count. (So unless your landlord breaks out the property taxes on your rent, and changes your rent every time the taxes change, you don’t get to count them. You also don’t get to count gasoline taxes or liquor taxes, unless you run the business selling them as an individual.) For married couples, taxes paid jointly (income tax, property tax on jointly owned property, sales tax) are divided in proportion to the taxable income received by each partner. If that’s hard to quantify (ie, they run a business together), or the couple chooses, split 50/50 instead. (Or maybe just split 50/50 anyway.)

          Received: All income from the government paid in cash to you. Welfare checks, government employee (including military) income, government pensions, social security. Probably not food stamps. Section 8 payments are income to the landlord, not the tenant. Medicare is income to the doctor, not you. Subsidized services don’t count – college, your kids’ public school. If you don’t see a check, it’s not income.

          I’d also make it that people working for (or receiving dividends from) corporations with government contracts should have part of their income counted as government income – U.S. gross receipts from government divided by total U.S. gross receipts. So if 40% of your employer’s gross receipts come from government work, 40% of your paycheck or dividend check is government income.

          Money from forced transactions probably shouldn’t count, but from voluntary transactions should. Sell your property to the city for $400,000, that’s government income. They take it and give you $400,000, it isn’t. (But the taxes you pay on that $400,000 don’t count on the other side.)

          It’s mostly a pipe dream, but if we actually got to that point, little details can be fixed in legislation.

      • Tracy W says:

        But tax money is only one measure of government involvement in our lives. To take a vivid example, if a government conscripts you and sends you off to war, you’re providing services to the government even if the net flow of money is from the government to you. To say that a conscript shouldn’t vote sounds wrong.

        And the government has other ways of interfering (rightly or wrongly) in others’ lives by means of regulation. Eg, banning religious practices, mandating various topics to be taught in schools, regulating pollutant run-offs from private land, or state-owned land, judging custody disputes on divorce, etc. Not as dramatic as conscripting someone, but still an interference in things that can be very important to people. I think that one of the advantages of one-vote, one-person is that it gives everyone a stake in those questions too (and even that’s rough on someone whose preferences systematically differ from the majority).

      • Adam says:

        Just agreeing with Tracy here, but one dollar, one vote can only be considered a good idea if the only thing a government does (or overwhelmingly the most important thing it does) is take money from some people and give it to others. It’s not at all obvious that’s the case. The government does a lot more than allocate resources. It decides what is and isn’t criminal, how severely to punish what is criminal, where to draw boundaries, who to fight wars against, what to feed schoolchildren.

        You’re also permanently disenfranchising anyone institutionalized or imprisoned, on permanent disability, retired and receiving Social Security and Medicare, possibly government employees and military depending on how you define “tax receipts,” all full-time students, anyone who is unemployed. It’s not obvious why these are people whose interests the government should ignore.

    • What’s so great about “one person, one vote” anyway? Obviously it’s easier to get buy-in with that level of superficial equality, but what’s the evidence that it’s the most “fair” or leads to the beat outcomes?

      Many states structured themselves in imitation of the federal system, allocating, say, one state senator per county regardless of population. The effect of this was to put an entire state under the control of a relatively small number of rural voters.

      The Connecticut House of Representatives was one of the most extreme examples. The state is divided into more than 150 towns of roughly equal area. Every town elected either one or two reps. Out of 294 members of the House, the urban areas had a relative handful (e.g., 2 for Hartford and 2 for New Haven), but small rural towns with 12% of the state’s population held a majority of the seats. Indeed, the rural towns all together held more than 80% of the seats.

      In some states, such as Michigan, the boundaries of the state senate districts were written into the state constitution, so that it would take a constitutional amendment to change them.

      Even in states where periodic adjustments of districts was theoretically required, legislatures would simply ignore that, because rural politicians weren’t interested in giving up seats to urban areas.

      Moreover, all over the U.S., many city governments had fixed ward boundaries (for election of city council members) which took no account of changes in population, so that an alderman from a newer part of town represented sometimes 20 or 50 times as many people as an alderman from a more established area.

      More remarkable yet, the U.S. Congress simply ignored the 1920 Census, so that the allocation of congressional seats to states was based on 1910 populations until March, 1933.

      And even then, many states had congressional districts which were widely divergent in population: cities were shortchanged, so that rural areas could have more seats.

      As the U.S. population gradually became more urban, state legislatures (particularly) became less and less representative, and there was essentially nothing that anyone could do about it politically. Arguably our democratic system was decaying into one of rule by a small rural minority.

      Courts were reluctant to wade into this “political thicket” until 1962, when the U.S. Supreme Court decided Baker v. Carr, and ultimately forced every state to base representation on population.

      By the 1960s, only the judicial branch could solve this problem, and the only neutral standard (Schelling point) it could reasonably use was equal representation.

      And those decisions didn’t just solve the problem that existed in 1962. Ongoing equal representation has become an enduring principle, and periodic reapportionment happens whether the incumbents like it or not.

      There are other problems in representative democracy besides the slide into malapportionment, but at least this one (outside the U.S. Senate) has been pretty well solved.

      That’s what so great about “one person, one vote”.

      • A further thought, from Oliver Wendell Holmes, Jr.:

        The life of the law has not been logic; it has been experience… The law embodies the story of a nation’s development through many centuries, and it cannot be dealt with as if it contained only the axioms and corollaries of a book of mathematics.

        This applies to the political system as well.

        And this country’s experience is that the power to determine the basis of political representation, if left to self-interested politicians, will be abused.

        And that’s why we have rigid, arbitrary rules like decennial redistricting and one-person/one-vote.

    • satanistgoblin says:

      “Maybe there are some issues where 1 Delawarean really should be worth 20 New Yorkers.”
      Or maybe 1 New Yorker should be worth 20 Delawareans? Is it not a priori just as likely? Then we risk having 400:1 mismatch instead of 20:1.

  46. TomA says:

    It’s human nature to apply different weighting factors to different circumstances. Most of the time, this distinction occurs subconsciously. It’s only in formalized analysis that this sort of differentiation typically becomes obvious. Over-analysis is a byproduct of affluence and excess leisure time.

  47. Rafal says:

    Moral feelings are an insufficient guide to moral feelings. It’s not just that they cannot guide action alone, they can hardly even be used for meta-ethical adjustments. Morality must be grounded in reality, and in reality everything has a price. The problem with democratic politics, whether of the proportional or the gerrymandered type, is that it hampers pricing, and thus silliness leads to dashed dreams and crushed bodies. Luckily, in the privacy of our own minds we can explicitly ask ourselves the unhallowed questions about prices of lives and suffering, that cannot be easily debated in the chambers of state.

    It helps to develop the habit of consciously asking yourself pricing questions, like “How many adorable little puppies would you vivisect to give your adorable daughter a 10% chance of being cured of an incurable cancer?”

    After some years of engaging in this mental exercise my answers are fast and calm: “As many as it takes, subject to availability and time constraints”, to the above question.

    Scope insensitivity is a failure which can be rectified by the cold hearted application of reason to the raw substance of moral feelings, mercilessly channeling them into coherent, and therefore, legitimately actionable forms.

    • Anonymous says:

      “How many adorable little puppies would you vivisect to give your adorable daughter a 10% chance of being cured of an incurable cancer?”

      After some years of engaging in this mental exercise my answers are fast and calm: “As many as it takes, subject to availability and time constraints”

      I’m fine with you saying things like this as long as you never have the opportunity to act on it. For example if you’d be willing to kill people to save your daughter, I’m going to make sure you’re never allowed to do so.

      • Rafal says:

        It doesn’t seem that you are succeeding at saying something relevant to my post. Also, you are vaguely accusatory, you insinuate wickedness on my part, and maybe you even promise violence.

        Whatever, the puppies must die.

    • TomA says:

      In a cosmological sense, the universe cares not what we sentient beings think of it. In a similar vein, for nearly all of the past billion years that life has existed on this planet, evolution was not altered by the morality of our species. Memetics is a game changer and where this will lead is anyone’s guess.

    • Adam says:

      The government does do this. Not at the legislative level as far as I know, but tables assigning dollar values to different categories and durations of human life, in order to make them directly comparable to other goods, are used in valuing capital projects and public investment.

  48. Furrfu says:

    By “election paper trails”, are we talking about not allowing voting machines that don’t produce verifiable paper trails? Because that might actually be a more important issue than the others — there’s a credible claim that non-voter-verifiable voting machines delivered Ohio and thus the Presidency to Bush in 2004. If Kerry had won the election, the US’s policies on global warming and probably the Iraq War would have been pretty different in 2005–2008. (Kerry initially supported the war, but then harshly criticized it as having been based on lies.)

    It seems to me that delivering US election results into the hands of Diebold and similar politically-connected voting-machine manufacturers — or whichever foreign intelligence agency is most adept at backdooring their machines — is unavoidably more important than any other issue of the form “What should be US government policy on X?”, since it would determine all of them.

    (Or maybe you just think that the Chinese Ministry of State Security, or the NSA, would do just as good a job of picking US presidents as the voters do?)

    • Irrelevant says:

      (Or maybe you just think that the Chinese Ministry of State Security, or the NSA, would do just as good a job of picking US presidents as the voters do?)

      Was that a rhetorical question, or…?

      • Furrfu says:

        No, that’s really who will choose the results of US elections under a non-verifiable electronic voting system — whoever wins the king-of-the-hill game of computer security on the voting machines. The smart money here is probably on NSA.

    • Samuel Skinner says:

      “If Kerry had won the election, the US’s policies on global warming and probably the Iraq War would have been pretty different in 2005–2008.”

      Why? The US was already in Iraq- why would the methods for countering the insurgency be any different?

      As for global warming while the policies might have been different, the changes probably wouldn’t. Its either efficient to invest in (in which case it doesn’t have an effect), it isn’t efficient (in which case it causes a temporary drop) or it changes things to be efficient (which I’m not aware of a currently proposed policy actually doing).

      “(Or maybe you just think that the Chinese Ministry of State Security, or the NSA, would do just as good a job of picking US presidents as the voters do?)”

      Better. While its possible the Chinese will go the virulent nationalism route, their government really seams to be focused on the “make the country and its people rich”; the best way to do that is to encourage trade with the US. Free trade is a net positive and outweighs pretty much any other policy when it comes to bootstrapping the Chinese up and getting them to deal with global warming (or at the very least getting them to stop burning so much coal).

      I’m not sure how competent the NSA is, but I’m not aware of any systematic political bias on their part (Obama “the transparent” has defended them so they can get along with any administration).

      • Irrelevant says:

        I’m not sure how competent the NSA is, but I’m not aware of any systematic political bias on their part.

        Pro-Navy, but there is no anti-Navy side to be on there.

    • Adam says:

      I don’t think election outcomes make much of an actual policy difference (at best, they moves changes into the current cycle that would have happened one or two cycles later). Encouraging trust in the voting system is probably the more important outcome of verifiable paper trails, since preventing civil wars and revolutions is one of those very heavily weighted priorities of a government.

      • Douglas Knight says:

        Yes, encouraging trust in the voting system is important, but do paper trails do it? Maybe they would encourage you to trust it, but you are not representative. And certainly advocating paper trails undermines the system.

      • Adam says:

        It doesn’t actually make much of a difference to me personally whether the voting system is trustworthy or not. I don’t vote and largely don’t care who gets elected or holds public office. It seems to matter to most people, though.

    • Jaskologist says:

      All of the 2000 election drama was around Florida, which had paper ballots and therefore a paper trail. This did not seem to help much. Instead, there were claims that butterfly ballots were too darn confusing, disputes around divining voter intent (the elderly among you will surely remember the term “hanging chad”) and arguments that only counties favoring Gore should get recounted.

      Not that I’m opposed in principle. Combine your paper trail bill with one requiring voter ID and you should have a nice, bipartisan, ensure-faith-in-the-election-process bill.

    • Look, I’m a liberal Democrat, been elected to public office seven times on the Democratic ticket. Moreover, I’ve been writing and speaking about issues in election administration for decades, I’ve been an attorney in election recounts, I’ve been an election official, and I’ve been a critic of no-paper-trail voting systems since well before the 2000 election.

      there’s a credible claim that non-voter-verifiable voting machines delivered Ohio and thus the Presidency to Bush in 2004.

      No, there’s not. The closer you look, the less you find. Bush carried Ohio by more than a hundred thousand votes. It isn’t even remotely plausible that Kerry actually won the state, and lost it through sinister manipulation of the election.

      All of the 2000 election drama was around Florida, which had paper ballots and therefore a paper trail. This did not seem to help much.

      You’re conflating two completely unrelated problems.

  49. blacktrance says:

    What’s wrong with saying that chickens have a positive value but the badness of the sum of all chicken torture is capped at a certain value? Torturing 1 chicken is much worse than zero chickens, but torturing 3^^^3 + 1 is only barely different from torturing 3^^^3. This may seem counterintuitive because we usually don’t expect the rightness of something to depend on how much it’s happening already, but it makes sense if we dispense with agent-neutral value and reframe it as formalizing how much you care about them. Is there anything inconsistent with declining marginal disutility to you from chicken torture?

    • Irrelevant says:

      What’s wrong with it is that Utilitarians –at least the breed we have around here– define Utility as having mathematical properties such that it follows superposition, which rules out that sort of fall-off.

      I consider that sort of bizarre and unnecessary, since “utility” isn’t a natural thing, and all you really need is for utility to be mathematically well-behaved within the small ranges of possibility that real decision-making incorporates, but I think utilitarianism only makes sense for organizations anyway, and individuals should be virtue ethicists. But then, I’m not a Friendly AI enthusiast.

    • Investigating what is morally right is different from investigating what you want. What you want may or may not be moral. I think you’re approach may be conflating two distinct things.

      • blacktrance says:

        Why be moral if it’s not ultimately reducible to what you want? (See also constructivism in general.)

        • If the right thing is the same as what we want, then aren’t we just answering “what is the right thing to do?” with “whatever you want” or “whatever achieves what you want”. Ethics or morality are misleading words for that notion. It’s more like everyone just asserting what they want.

        • blacktrance says:

          The substance of what achieves what you want is complicated and requires non-trivial investigation. If you ask “What is the right thing to do?”, “Whatever you want” is correct, but not helpful except as a starting point – it’s like going to the doctor and them telling you that your problems are caused by atoms. Also, this framing suggests that what you want is relatively simple, though it’s actually complicated at best and often internally inconsistent. “What you’d want if you were rational” is closer to the mark. The term “want” conflates whims and rationally considered and endorsed desires.

          I don’t think it’s misleading to call this “ethics” or “morality”, both because it has many of the attributes we expect from morality (answers questions delineated as “moral”, is motivating, claims about it are truth-apt, etc) and because it’s traditionally classified within morality in academic philosophy (egoism and contractarianism are both ethical theories). It’s also more than everyone asserting what they want, because it has an interpersonal aspect: the results of negotiations between different people’s desires.

    • Anthony says:

      It’s actually completely intuitive – diminishing marginal utility is what most people *really do*.

  50. I find it interesting and odd that I am one of only a handful that didn’t think this post was actually about bicameral systems of government.

    I also think all the attempts to create moral discount rates for chicken lives that are based on subjective experience lead to motivated reasoning and factual claims that are fairly blatantly wrong. Chickens and other animals obviously suffer – we know they suffer because they have most of the same apparatus for suffering as we do (nervous system, brain). Suffering drives us to avoid harmful things – that’s neccessary for all sophisticated animals not just us.

    A more sensible version of our discount rates is found in a more biological aproach to ethics. Members of our own species are our first priority. But as Earth’s cooperative structures expand it’s not surprising that we would start to bring other species into the fold, like we have mostly done for other humans. Other species are our genetic relatives and while they don’t matter to us the same way humans do, they still matter.

    A society that abhorrs suffering is a good one because its a society that values life. For those of us also interested in astrobiology, sensibly working towards a world that minimises suffering (ie. moving to vat-grown meat) also makes perfect sense in a universe where Earth is not the most advanced civilization – it is not impossible that our interaction with other civilizations could be influenced be our interaction with other species on Earth.

    (formal philosophical defence here)

    • Irrelevant says:

      I find it interesting and odd that I am one of only a handful that didn’t think this post was actually about bicameral systems of government.

      I thought he was commenting on the cleverness of our mental structure: listing every axis you’re interested in and taking a majority vote seems dumb as hell, but since (ability to effect issue) x (importance of issue) ~= (interest in issue), making a list of everything you’re interested in actually gives a decent first-approximation utility weighting.

    • Jaskologist says:

      A society that abhorrs suffering is a good one because its a society that values life.

      Not so. Suffering is wiped out just as, if not more, efficiently by wiping out life. In fact, this is the Buddhist solution.

      “You were not made for comfort, you were made for greatness” is an ethos that also values life, but does not get terribly hung up on suffering.

      • Adam says:

        Or at least better captures what seems like the fairly obvious intuition that not all suffering is bad. Otherwise, the rational behavior of any major league baseball player with a guaranteed contract is to self-induce a career-ending injury and go live on an island, avoiding all the pain of travel, practice, slow body degradation, and defeat and frustration. I endured an awful lot of suffering in the Army, and imposed quite a bit on my Soldiers, but it was for the purpose of making us better at our job and better at staying alive when others were trying to kill us, and I don’t think very many of us regret it.

        On the other hand, I can’t imagine ever willingly agreeing to live in a cage and have my arms and nose cut off, surrounded by hundreds of other people that had the same thing done to them.

        • Peter says:

          Suffering in the army: this isn’t at odds with the utilitarian view, well, maybe the negative utilitarian view but possibly not even that one. Similar to what I say upthread, while suffering may be an intrinsic bad, it may have good consequences; something which involves suffering – possibly which inescapably involves suffering in the real world – may well end up being a net good, that doesn’t stop the suffering itself being one of the things that makes a negative contribution to that net good.

          “Neque porro quisquam est qui dolorem ipsum quia dolor sit amet, consectetur, adipisci velit …” and all that.

        • Adam says:

          Well, I’m not trying to be at odds with a utilitarian view. I’m just saying that suffering is not the opposite of utility. Some types of suffering are quite gratifying, like summiting Everest.

          Although, I suppose in this vein, it is not always easy to tell the difference between utility maximization and pathological compulsion or addiction as a motivator of human action.

        • As I am not a utilitarian I can totally agree that sometimes suffering is worth it. (actually Peter makes a valid point that could be true even for utilitarians) Defending people from death is certainly noble enough to fit into that category!

          You’re also raising an interesting distinction. Somehow self-imposed suffering (eg. willing service to your country) seems quite different from suffering imposed by circumstance (eg. starvation) or other people (eg. torture). It’s hard to see a society without the first type achieving anything at all. I personally am cautious using the term “greatness” though, because its quite a flexible and easy-to-manipulate concept. Still, there can be no doubt that if we don’t suffer a little sometimes then we’re not fufilling our full potential as humans.

      • I’m happy to admit suffering can have instrumental value. I agree that making the elimination of suffering your only concern is a terrible idea. However imho systematic indifference to suffering is also very unwise.

        I take your point that notions of “greatness” can also have utility, but I’ve noticed its a pretty subjective term that can be easily manipulated for agendas that do not value life and instead confuse utility (eg. strength) with instrinsic value. So I think we ought to be careful to only use the term in a very considered and specific way.

        I find the idea that Buddhism wants to wipe out life to be absurd. But I’ll leave you to argue that with the Buddhists.

  51. Emile says:

    Intriguing hypothesis: how much we care about the well-being of various entities (chickens, the homeless, our family, etc.) is directly related to how much of our brain is dedicated to modeling them – in other terms, our mirror neurons vote for who they’re mirroring (this is not to be taken as neurologically accurate).

    This is a huge simplification, but I find it plausible that there’s a causal, “mechanical” link between caring for something and knowing about it (beyond the obvious “we can’t care about something we don’t even know exists). Even if there are millions more chickens than members of my close family, my brains has more neurons dedicated to modeling my family.

    On the other hand, I probably know much more about chickens than I do about gorillas, yet I care more about death and suffering of gorillas than I do about chickens.

    • Anthony says:

      How much you know about chickens versus gorillas isn’t the same as how much you identify with chickens versus gorillas.

      Gorillas are “almost human” in a way which chickens just aren’t, for almost anyone. This supports your hypothesis, possibly with some modification.

  52. Tom says:

    I have to disagree on the reasoning about the Iraq war. Let’s start with a counterfactual: If Dubya were right, if Saddam were making weapons of mass destruction or involved in 9/11, would it have been the right decision to go to war? In this hypothetical, I think both candidates would agree that the decision to go to war was correct. We’re not talking about a pacifist vs. a hawk here. The question is when a candidate will go to war, not if they’re willing to go to war.

    For a historical example, let’s take Neville Chamberlain and Winston Churchill. Chamberlain favored appeasement, Churchill favored a hard line. Yet, when Germany invaded Poland, Chamberlain was still in office and he was the one who declared war.

    So, I think it a severe blunder to say that, since the Iraq War killed a million people and McCain supported it, that McCain would be willing to kill one million people and Obama wouldn’t.

  53. Katherine says:

    The congress model can be reduced to the senate model simply by subdividing any issue that is n times as important as another issue into n subissues.

  54. Ellen says:

    Great article! But how politically scope insensitive are we, really? From my limited experience, not terribly.

    I remember sitting around a campfire with my mom and grandpa in, like, kindergarten and listening to them discuss politics. My grandpa was talking about all these issues, and my mom was admitting to not being well-informed or having strong opinions about any of them. She said, “As long as democrats approve of abortion, nothing will convince me to vote for them.” My grandpa sighed and told her she was trapped in a religious bubble, that there was a whole world out there she was ignorant of.

    But I remember thinking, “Wow! Mom is actually smart.” If democrats are murdering tons and tons of babies every year, and we only hope they go to heaven but God doesn’t actually say, who cares about money or guns or school or any of that other stuff? Even global warming was insignificant, since we believed the earth was going to end anyway. Maybe global warming would just be His way of destroying it. So based on abortion alone, I considered myself Republican until I deconverted from Christianity.

    As soon as I de-converted, I thought, “well, I guess I should probably vote democrat now.” There was no longer any guarantee that the earth would end anytime soon, and I’d quite rather it didn’t. All the other issues were fun to think about if I could find the time to inform myself, but I wasn’t terribly concerned about them.

    I loved the article, but maybe we’re less politically scope insensitive than you think, and maybe a lack of scope insensitivity (plus herd mentality, of course) explains the strong correlation between political party and religious belief.

    • Adam says:

      I think you’re largely correct. The checkboxes might exist, but I’m not convinced very many people actually use them, and in general, getting voters to believe “if you vote for my opponent, he’ll kill a million babies a year” is a more effective strategy than exhaustively enumerating the entirety of your platform and legislative agenda.

      • Ellen says:

        Yeah, exactly.

        People talking about religion often ask, “If it brings people comfort, and it’s not really hurting anything, who cares?”

        I thought like that for a while and, to some extent, still do.

        But it is hurting something, indirectly. Not the individual believers, but the earth and the future.

        • suntzuanime says:

          You don’t have to be religious to care about the lives of the unborn.

          • Ellen says:

            Nope, you don’t.

            But if you’re not religious and you still care about the unborn, that makes it significantly harder to choose a political party.

          • John Schilling says:

            If I’m a liberal secular humanist who considers abortion to be morally equivalent to murder, I don’t give a damn about the religious or ideological basis for the GOP’s pro-life stance any more than Winston Churchill cared about Joseph Stalin’s ideological differences from the Nazis.

            If I value unborn human lives as I value any other human child, I’ve been in crisis mode since the 1970s at least and the only single thing I can afford to even consider on election day is, which guy will, for whatever reason, join the fight against this unending 100 mH/yr atrocity.

            Well, unless I also value chicken lives as I do human children.

          • Irrelevant says:

            I believe Ellen was attempting to imply environmental concerns are for “the unborn.”

          • Ellen says:

            Personally, I probably value the “conceived unborn” more than the “not-yet-conceived unborn” but far less than the “born.”

          • Personally, I probably value the “conceived unborn” more than the “not-yet-conceived unborn” but far less than the “born.”

            I reject any formulation which requires that a mere fertilized egg be instantly endowed with full citizenship.

            The awkward reality is that human-ness develops gradually from fertilization to birth. If we must have a single bright line for legal purposes, neither fertilization nor birth are good places for it.

            (And maybe this comment thread isn’t the right place for this kind of discussion. If so, then please accept my apologies and delete this posting.)

    • Jiro says:

      It sounds like you were a person who took the logical implications of your religion seriously.

      This often leads to deconversion.

    • Matt M says:

      I agree with you. Not only do I think we’re NOT scope-insensitive, I think in many cases, popular culture actually promotes and encourages people to in fact BE scope-insensitive, but many people still aren’t.

      Consider the amount of grief and mockery directed towards “single issue voters.” While in the post, Scott makes a logical and reasoned argument for how some issues are much more important than others, consider that in reality, if we went around saying something like “I agree with all of McCain’s economic, social, and environmental policies, but I voted for Obama because foreign policy is more important than all of that” you would almost certainly be dismissed as a closed-minded idiot.

      And consider that the two issues where this most commonly happens ARE in fact wars and abortions, the ones where millions of lives literally ARE at stake.

      Popular culture and sentiment actually encourages people to treat all issues as approximately equal in the pursuit of proving how knowledgeable and/or “open minded” someone is.

  55. Deiseach says:

    I’m watching the first semi-final of the Eurovision right now.

    This is something that ideally should not be undertaken in a state of sobriety, but as it’s midweek and a working week – well, you know yourself.

    The Final is on Saturday night. And Australia will have an entry.

    Yes, now Australia is counted as part of Europe.

    I almost wish the U.S.A. had the equivalent of a postal vote or something; you lot couldn’t make this contest any crackier than it already is, and the best bit is always the voting anyhow.

  56. Eric says:

    > “Well, I agree with this guy about the Iraq war and global warming, but I agree with that guy about election paper trails and gays in the military, so it’s kind of a toss-up.”

    I’m confused, from the rest of your post it sounds like you agreed with Obama on all four. Or was that sentence not meant to be taken literally?

  57. fwhagdsd says:




  58. Pingback: Outside in - Involvements with reality » Blog Archive » Chaos Patch (#63)

  59. Uniqueness says:

    Isn’t your attitude to chickens entirely consistent with your attitude toward *yourself*?
    Your value goes down when there are close substitutes available. If there are trillions of Yvains, that’s not much better than having one Yvain. Why should the same not hold for chickens?

  60. 27chaos says:

    Here’s a similar problem resulting from the same underlying flawed thinking: