Open Thread 4: The Quick And The Thread

1. Big thanks to Bakkot and Alice for adding the script last month that highlights new comments in green and makes this place much more readable.

2. I’ve closed comments on any posts older than one month in order to cut down the spam problem. If anyone has objections you can voice them here, but they better include some other way of dealing with spam. I already use Akismet.

3. There’s been some discussion of improving the comment sections of very controversial posts (eg on feminism) by closing comments there, then making a comment section on a separate thread. The hope is that all the random people linked there by Reddit and Instapundit and whoever get confused and go away, but other people who specifically read this blog will find it and be able to talk about it. I’ll probably try that next time I’ve got something controversial to say.

4. Comment of the month is this description of algorithms and the halting problem.

5. Ozy and I will be in the Bay Area for a few days starting September 19. Is anyone able to lend us a room to crash in for some of that time? We will take you out to dinner or something for your trouble. (we are in town for a wedding and probably won’t stay too long, but if there is some big community social event going on around then we will try to attend) [solved! thanks to everyone who volunteered space!]

PS: NO RACE OR GENDER ON THE OPEN THREAD THAT NEVER HELPS

This entry was posted in Uncategorized and tagged . Bookmark the permalink.

290 Responses to Open Thread 4: The Quick And The Thread

  1. Matthew says:

    This seems like a pun Scott might appreciate.

    Also, I’ve just discovered the perfect card game for the SSC set.

  2. noahluck says:

    When reading Nick Bostrom’s “Superintelligence” in which he described Coherent Extrapolated Volition, I couldn’t help but think it was missing the insights about diverse communities from the Archipelago blog post here on SSC. Whereas CEV tends to be described in terms of humanity‘s volition, an “Archipelagian” version might start off “For all humans and all communities of humans…” and then describe doing CEV. Thus it could explicitly condone the idea that humanity-as-a-whole’s volitions might not cohere, but that smaller communities could still be formed and nurtured according to their own principles. (Those with anti-speciesist concerns could go farther and say “For all salient groups of living things, weighted by their moral worth, …”.)

    Plus, for Moloch-like reasons, corporations and nation-states might be more likely to collaborate on an AI that also sought their interests as well as the interests of the people of which they are made.

  3. rationalnoodles says:

    Idea for a blog: Slate Star Codex comments summaries.

  4. Paul Torek says:

    You (Scott) should receive a book from Amazon today, sent by me. If not interested, please pass it on (knowing you, that won’t be hard).

  5. Viliam Búr says:

    I was thinking recently whether some unpleasant aspects of internet feminism / social justice warriorism aren’t just discussion forum rules-lawyering on a much larger scale.

    To explain: When a discussion forum is actively moderated, some people demand to know exact rules about what is allowed and what is not, to prevent an abuse of power by moderators, such as banning someone for doing something which was not really nice, but which probably wouldn’t lead to a ban if done by someone on the opposing side. The rules of what is allowed and what is not should be predictable in advance and the same for all sides.

    However, as soon as the exact rules are given, some people will treat it as a game: how close can I get to the lines to be still on the good side. And unless the rules are draconic, they can get very far, but now with impunity, because hey, they are following the rules, even if they are obviously assholes who would risk receiving a ban on a site with inexact rules. For example: “People are not allowed to insult each other in comments.” “Hey, that’s very ambiguous! I don’t know what exactly would you interpret as an insult. I don’t want to risk violating your rules unknowingly. Could you please give me a list?” “Here is the list: never call anyone X, Y, or Z.” “Great! You forgot to add Q on the list! Now I am going to call Q everyone who disagrees with me.” The original intention was to have a discussion without people insulting each other, but with exact rules it became a discussion with some people safely insulting other people, because now there is the letter of law they can follow while ignoring the spirit.

    I suspect something similar happened with internet feminism and similar stuff. Let’s assume charitably that at the beginning there were people trying to make the world a better place, where everyone can be equal, etc. But then the rules lawyers came and started demanding exact rules. “Let’s not speak negative things about people merely because of their gender, okay? Such generalization are never good.” “Well, then you technically shouldn’t speak about Patriarchy, because, you know, it’s about men and it has negative connotations, so it’s against your rules.” “Uhm, let’s make an exception there: if some gender has institutional power, then such criticism is fair.” “So, you are saying that if some gender has institutional power, the rule about not speaking negatively doesn’t apply to them?” “Yeah, I guess.” “Great! Mwahahaha! Die, male scum! I bathe in your tears! Hey, I’m just following the rules of niceness we have agreed upon.” “Wait, I didn’t mean it like that.” “Too late, sucker, I like this definition and I am gonna use it!” With the difference that this is not happening on one specific discussion forum with one specific moderator who could break the pattern and stop the rule-lawyering, but across many websites with too many people having too much fun. Individual websites can have reasonable moderators, but the internet as a whole doesn’t. Therefore we can’t have a nice global discourse.

    • That may have some structural similarity to what happened, but as I understand it (from reading a good bit of Racefail– clearly not enough to get the whole story), the actual sequence is that people had been making polite, reasonable arguments for a long time about their interests and emotions being ignored, and those arguments were also ignored.

      At some point, there was a consensus to get nasty, partly because people were pretty angry by then, and partly because they wanted to see whether that would get them heard.

      In the earlier stages, there wasn’t much explanation attached to the anger (Educate Yourself!). Eventually, explanations were developed and a sort of consensus about rules were developed, but that’s a late phase.

      Also, people were volunteering to be allies, and eventually people worked out protocols for what allies were supposed to do.

      In some cases, allies were more willing to dump anger and destroy social connections than the people the allies were defending were. (I don’t have specifics on this, but I’ve seen a complaint or two.)

      The high anger level did get more attention and success than a politeness-only approach did, but there’s also been a lot of blow-back.

  6. Ialdabaoth says:

    So, I just discovered that I’ve been approved as lead developer on an educational gaming platform.

    I’m anticipating one thematic fight on my hands (which is about a Forbidden Topic), but once that’s settled I’d like to recruit people with art, writing, or coding skills that would like to contribute.

    Yes, there’s a budget to pay you. 🙂

    • Matthew says:

      I’m useless for coding or realistic art, but I’m a decent cartoonist and an excellent* writer. I can point you to samples of either upon request.

      *By excellent, I mean not quite as good as Scott, but not totally out of his league either.

    • Hainish says:

      How…do I get in touch with you?

    • I’m potentially interested in this, as I’m quite interested in educational technology that doesn’t suck and am not otherwise employed at the moment. I’m a coder (and also a decent writer but likely not with the specific kind of experience you’re looking for).

    • Ialdabaoth says:

      🙂 For now, you can find me on facebook (it’s not that hard).

      Keep in mind that quality of the game comes first, so I’m gonna need to see some work samples before I dole out any cash.

      Also, you’ll need to be eligible to sign a 1099 work contract, and there may be other weirdness due to the money coming from a federal grant through a community college.

      • Hainish says:

        But, maybe you’re looking for coders, rather than content writers? Anyway, you were not at all hard to find on fb.

        • Ialdabaoth says:

          I’m looking for coders right now; I’ll be looking for content writers once the framework is laid down.

      • Ialdabaoth says:

        Update: Everything has turned to fail and ruin.

        I talked with my manager yesterday to ask if there was anything I could do about my performance dropping at work due to seasonal affective disorder.

        He took me to HR, who has an Employee Assistance Program.

        They decided to fire me because I admitted my productivity had dropped, and I could not commit to 100% certainty that it would be back up to acceptable levels by Monday (at least partially because I couldn’t get a straight answer on what “acceptable levels” even meant).

    • jaimeastorga2000 says:

      Are you looking for fiction writers, or nonfiction? Here’s a short story I wrote which was well-received by the lesswrong community, and here is the post in which I rewrote various normal facets of the human condition as horrifying angsty undead curses.

    • Viliam Búr says:

      I am interested in games and education, so I would love to know more about the project, and offer some ideas, even without money involved. Actually, my job contract prevents me from making money aside. Could contribute some little code for free, though.

  7. Hainish says:

    To whom it may concern,

    ETS (the makers of the GRE) is now making their accumulated research available free through Wiley Online:

    http://www.ets.org/newsroom/news_releases/ets_massive_research_report_series

    [Unlike the comment directly above, this is not spam, and I am really a person.]

  8. coffeespoons says:

    Could anyone recommend any other spaces to go to for slatestarcodex style discussion on gender, politics, social justice etc? I’m on tumblr, but I’m so far not amazingly keen on it as a discussion space (though I like the rationalist tumblrs I follow)!

    • Nornagest says:

      I don’t think there’s anywhere good, but some places are less bad than others. SSC is near the top of the list. Tumblr (as in the Tumblr hive gestalt, not specific Tumblrs) is near the bottom. Most non-Tumblr blogs, and most gender subreddits, will fall somewhere in the middle-to-lower region depending on taste.

      For feminish stuff, Ozy’s about as good as it gets from my perspective. (Hi, Ozy!) They have a Tumblr, which I don’t want to go looking for at work, and a good fraction of their older work at e.g. Good Men Project is still online if you look for it. Clarisse Thorn is also better than average. For manospherish stuff, Hugh Ristik is about the least bad to my taste, but he doesn’t seem to be writing much anymore, and the collaborative blog he founded has been largely taken over by bitter ideologues.

      Another approach would be to mine the sidebar here.

  9. ADifferentAnonymous says:

    Sorry to bring up neoreaction on the Open Thread, but this is a meta-level thing that I expect can be discussed productively for a bit before upsetting object-level examples derail the conversation.
    Some of the comments on <a href="http://lesswrong.com/lw/kxb/nrx_vs_prog_assumptions_locating_the_sources_of/"http://lesswrong.com/lw/kxb/nrx_vs_prog_assumptions_locating_the_sources_of/ made me wonder, how good is history as an optimization process? What evidence do we have one way or the other? Neoreactionaries love to say that traditions and taboos are in place for a good reason, but is the process by which they arise something we would expect to converge on good ideas?

    • Paul Torek says:

      What is it that’s supposed to be evolving through the action of historical processes? Cultures? Nations? Memeplexes? Memes? All of these are presumably affected, but none of them seems exactly parallel to either genes or organisms in biological evolution. Well, maybe memes.

      The next question is, given that a meme is highly successful, how much of a recommendation in its favor is that to us?

      • ADifferentAnonymous says:

        The question applies to anything that Neoreactionaries claim is optimized over history. Memes, memeplexes, cultures, and political structures all fit. There doesn’t have to be a close parallel to biological evolution, there just has to be optimization.

        Political structure is probably the least derailingly upsetting example. If, over history, we saw societies end up with various forms of government seemingly at random, and everything that wasn’t monarchy quickly collapsed and was replaced with another random government, and meanwhile monarchies lasted almost forever, with the result that after a short time almost every society was a monarchy, it’d be fair to say that form of government was strongly optimized. At the opposite extreme, if every civilization until the modern era was always a monarchy, then we could say that form of government wasn’t optimized at all. These are two simplified extremes to illustrate the spectrum I’m wondering our place on.

        Some possible selection mechanisms: survival of the stable; collective memory observing and remembering the effects of things (a handful of cities try non-monarchy, they all fail, and the surrounding people acquire pro-monarchy folk beliefs); conquest by memes/structures/cultures better suited to conquering.

        (It’s also debatable–and I do mean debatable both ways–how much things optimized for, say, ability to conquer are likely to be good ideas today).

        • Fazathra says:

          The problem is that neoreactionaries also claim that different peoples are most suited to different styles of government, and that this depends on some melange of culture and innate dispositions so that there is no one form of government that every society is being optimised towards. Thus, we need to look longitudinally at particular societies and see whether, in some stable environment with no outside pressures, they naturally tend towards some stable political state, which may be different among different populations.

          I’m not quite sure how neoreactionaries reconcile this with their belief in monarchy, because here their own beliefs also run right into HBD, as empirically WASPs are one of the populations most unsuited to monarchy that there is

    • jaimeastorga2000 says:

      Neoreactionaries love to say that traditions and taboos are in place for a good reason, but is the process by which they arise something we would expect to converge on good ideas?

      You need to taboo “good idea” (or, alternatively, specify the metric by which the goodness is measured). Neoreactionary arguments usually point to traditional values which help a civilization exist in the first place (by motivating people to abandon forager lifestyles), help a civilization survive against external threats (by building and maintaining military power), or which help a civilization be productive and efficient (by incentivizing people to create wealth), all of which are highly interrelated.

      • ADifferentAnonymous says:

        It’s harder than I expected to disentangle the questions of how well optimized something is and which optimization criteria serve our values. Everything that exists is in some sense (probably) optimized for existing, but not all forms of that are equal.

        I’m implicitly quite confident in modern society’s abilities to create structures that didn’t exist or weren’t common before, and much less confident in its ability to make structures productive and efficient that previously were not.

        So if cultural values were set by the whim of the elites, and elites in various cultures tended to have the same whims, the fact that those values are appealing to historical elites wouldn’t be much of an argument in their favor, even though those values are optimized for existing. If nations rose and fell largely due to their cultural values and some cultural values were consistent winners, that would be a much stronger argument.

    • I wouldn’t expect optimization so much as the elimination of sufficiently bad cultural elements– and also the elimination of some good elements that were unlucky.

      Evolution doesn’t optimize.

      • ADifferentAnonymous says:

        I’m mostly thinking in terms of the LW usage, where evolution is typically referred to as an optimization process (call it meliorization or something if you prefer).

        Another framing is, if an element occurred in history with a certain frequency, what does that tell us about it? In particular, can we conclude anything about an element that as far as we can tell didn’t occur? If all history does is eliminate sufficiently bad elements and also some good elements, then we can’t conclude much about a rare/unobserved element.

        …unless the space of possible elements is so overwhelmingly loaded with very bad ones that our prior for a proposed element is that it’s almost certainly very bad and only seeing that it has actually worked is strong enough evidence to conclude that it’s not so bad. Maybe that’s the neoreactionary idea?

        (‘very bad’ above mostly referring to things like ‘leads to societal collapse’ that almost everyone can agree are very bad)

  10. CAE_Jones says:

    The talk of monetary incentives in Fusion research in the links thread reminded me of something that turned out to be different enough on reflection that I decided to post it here.

    A common complaint about companies that sell products for the disabled (I’ll focus on the visually impaired because I am the most familiar) is that the prices are unreasonably high for the quality of the products and the purchasing power of the demographic that will benefit. Freedom Scientific is the go-to example: the cost of a license for the Jaws for Windows screen reader costs ~$800-1000. The cheapest braille devices display less than 20 characters, and cost ~$1500 (those are the cheapest; something with 80 cells pushes $8000). Notetakers run on outdated technology and cost multiple thousands of dollars. There is no way the majority of peple who use this technology can afford it, so the government soaks up the cost.

    Meanwhile, there is a free, donation-funded, open-source screen reader in NVDA, courtesy a couple Australian programmers who finally got sick of this crap, and its capabilities are rapidly improving to the point that it is a viable alternative for Jaws. iOS devices come with accessibility out of the box (Microsoft Narrator does not count as well as Voiceover, but people have wondered if those government subsidies of Jaws went to funding Narrator development, would the government save loads of money in the long run?). Tactile output doesn’t get the same treatment, because it’s more like developing new monitors and most likely needs the money/coordination that went into getting us from black-and-white to Retina. (Or an eager enough market: see Senseg and ReduxST)

    There are more examples! In 2010, my dad spent $700 on a Trecker Breeze, a talking GPS device whose inferiority to smart phone apps and contemporary dashboard-mounted GPS is considerable (I’m told later versions of the device have improved, but if I’m spending $700, I’ll spend it on a considerably-more-powerful smart phone, thanks). Do you know about those remote alert devices that people in theory use to find their keys or whatever should they misplace them? There’s something that is effectively that, but marketed to blind people (I don’t remember the price differences off the top of my head). And let us not forget the afore-mentioned notetakers, just because tablets are cheaper and considerably more powerful and are missing only braille i/o.

    Unlike the Fusion example, there are market arguments in favor of this situation: Jaws is still the most robust screen reader for business applications like Microsoft Office, programmers and support staff are worth a lot of money, and the target audience is tiny. It seemingly took the US Government passing new equal access laws to motivate Apple to develop the accessibility that they currently have.
    But one might be forgiven for wondering why, if the government is looking for nickle-and-dime budget cuts, they haven’t taken 10% of the money they spend on Jaws licenses, offer it to NV-Access as incentive to resolve features where Jaws wins out, and save the remaining 90% for, like, healthcare or something. (Maybe they don’t actually buy that many Jaws licenses or notetakers in a year, but that disincentivizes users to update their technology.)

  11. Alexander Stanislaw says:

    Regarding Utilitarianism. Ordinal utility (having a rank order of preferences but no numerical size) supposedly doesn’t allow you to aggregate multiple people’s utility functions due to Arrow’s Impossibility Theorem. I thought that you could get around the problem by using cardinal utility (measuring utility with real numbers). You don’t need to aggregate rankings, you can just add up utils and compare that way – and since vNM gives you a way of converting ordinal utility to cardinal utility there is no problem . I no longer think this works.

    Cardinal utility allows you to specify that Person A prefers X twice as much as Y. But it does not really give you a way of specifying that person A prefers X twice as much as as person B does. For any one person, you can multiply their utility function by a constant and it is still a valid utility function that preserves all of their preferences. You can’t compare different people’s utility functions. You can normalize each person’s utility function, but then you still have to justify the values you’ve chosen (because an infinite class of utility functions can adequately describe one person’s preferences). vNM doesn’t get you there.

    • Douglas Knight says:

      That’s all correct, except that AIT is completely irrelevant.

      • Creutzer says:

        Why’s that? AIT is exactly about how to aggregate ordinal preferences and tells you that you can’t do it in a way you’d intuitively find satisfactory.

      • Alexander Stanislaw says:

        Thanks, I admit I don’t really understand why arrow is supposed to pose a problem for aggregating ordinal utility. I just have a vague sense that doing so is messy and there is no obvious systematic way of doing it.

    • RCF says:

      “For any one person, you can multiply their utility function by a constant and it is still a valid utility function that preserves all of their preferences.”

      Or, more generally, any orientation-preserving affine transformation (i.e. multiplication by positive numbers and addition).

    • dublin says:

      Yeah it turns out if the order of preferences is the only thing you care about, you have ordinal utility not cardinal utility, even if you assign the preferences numbers. Of course nothing says that’s all you can care about.

    • noahluck says:

      Interpersonal utility comparisons don’t have to be a problem. A fair scheme that agents behind a Rawlsian veil of ignorance might agree to is “relative utilitarianism”. In that, each agent’s least-valued possibility is set to 0 and each agent’s most-valued possibility is set to 1, and all intermediate values are interpolated appropriately. If an agent has a bounded utility function, then it’s easy to linearly rescale it to the range [0,1]. If an agent has an unbounded utility function, you can transform it to the range [0,1] with, say, a sigmoid function. The ordinal rankings are unaffected by such a transformation, but there is severe distortion of the relative differences in utility over most of the range, so the veil-of-ignorance solution would probably be to let any agent with an unbounded utility function select its own function to transform its utility to the range [0,1]. (To forestall voting strategy effects that can lead to suboptimal outcomes, they would need to choose those transforms without access to polling data on others’ values.)

  12. Andy says:

    I have talked about my game design project on previous Open Threads. In short, it’s called Elemental Brigades, and it’s a party-based strategy game where each player takes command of a small army, and the players work together in playing through a campaign. Picture Warhammer meets Dungeons and Dragons.
    To encourage specialization among the players (and because this grew out of Pokemon), players have to choose an Element – the four classical elements of Earth, Air, Fire, Water, plus Plant and Death – and build an army out of its playlist so the players have to rely on each others in different situations.
    Here’s my fluff from the rulebook about the different elements:

    Fire – Fire magic is among the simplest and most straightforward magic to use, but difficult to master. Minor Fire sorcerers can often be found among armies, where their destructive magics are highly valued, but Fire also has its purifying and healing sides. Fire sorcerers tend to be hot-tempered and courageous, passionate and energetic, but tactless, reckless, and intolerant. Fire Elementals tend to be fierce but fragile creatures that burn all in their path. The fire-worshipping Centaur nomads of the inland deserts sometimes run in support of Fire sorcerers.

    Air – Air magic is the magic of the wind and weather, the magic of flow and change. Air sorcerers tend to be mercurial and emotional, quick-witted but scatter-brained, quick to forgive and forget. Air magics on the battlefield take more mana to perform, but are impressive in their effect: devastating lightning bolts that fly across the battlefield, gusts of wind that guide clouds of arrows to their targets, or pick friendly units up and dropping them elsewhere. Air elementals usually move fast and strike from a distance, but suffer when they are forced to stand their ground.

    Water – Water magic is the magic of rushing rivers and killing cold. Like Air, it is the magic of flow, but Water flows in patterns, stagnates and freezes and finds its way into every crack and crevice. Water sorcerers tend to be intellectual and reserved, though they often come off as cold, detached, and insincere. Water magic on the battlefield allows units to enter shallow and deep water, to freeze the enemy in their tracks, and to pull the ground from beneath their feet. Water elementals are often amphibious and adaptable, and excel in melee combat, but lack in ranged hitting power. The amphibious Fishmen who live in many lakes and shallow seas sometimes leave their underwater villages to protect and serve Water sorcerers.

    Earth – Earth magic is stolid and steadfast, immobile and durable. Earth magic is the magic of defense, toughening soldiers’ skins and bogging enemies down in moats of quicksand. At its greatest level, Earth magic can reshape the battlefield, devastating enemy armies with earthquakes and avalanches. Earth sorcerers can seem slow-witted, but they are thoughtful, deliberate, and as stubborn as a mountain. Earth Elementals are tough but slow, with rock-hard skins. When roused to anger, they can attack with the uncaring power of a landslide. The Trolls who live under the mountains worship the Stone and often march out to serve Earth sorcerers.

    Plant – Plant magic is vital and hearty, the magic of growth and healing. While most would think “the green road” is wholly unsuited to the carnage of the battlefield, Plant magic is also the magic of nature red in tooth and claw, or the parasitic vine that kills a tree. While many Plant sorcerers are kindly, nurturing healers, the Green Lords who rise to command armies are ruthless, methodical, and uncompromising in defense of their objectives. Plant magic can heal allies, but it can also summon vines and roots to slow the enemy, shower foes with venomous insects or thorns, or see through the eyes of birds and mammals. Plant Elementals are resilient and powerful, with sharp thorns that return damage to their attackers.

    Death – Death magic is the opposite of Plant magic, the magic of decay and stagnation. While most think of the “Silent Aspect” as the province of power-mad sorcerers with visions of world conquest, Death magic has a crucial role in ensuring the orderly flow of life-force between all the realms. Death sorcerers are quiet and cautious, subtly undermining their foes before destroying them in open battle. Death magic is not merely slaying enemies with a gesture, it can also be subtle and tricky, hiding units from notice, calling spirits to frighten an enemy, or sending deadly plagues through an enemy camp. In addition to the undead hordes that are the most visible manifestation of Death magic, Death sorcerers can call the shadowy Elementals of the Death Realm to do their bidding.

    So an informal poll for those strategy gamers among the SSC commentariat: From the above descriptions, which Element would you try playing first? I’d do Plant, and it’s the first army list I put together, because it’s defensive in a very devious way. Instead of drawing a line in the sand, I picture a Plant army retreating slowly, deeper into the forest, and making the enemy pay for every foot of ground. This is closely followed by Death because the notion of maintaining balance in a progressive way, and a mindless, fearless army can be used in so many tactically interesting ways other than a straightforward skeleton-wave assault.
    So what would you play?

    • Emile says:

      Neat, I was looking froward to hearing more about that!

      I’d probably go for either death or air magic, because they seem to be the ones that rely the most on cleverness and strategy. I would have gone for the machine-themed element but there isn’t one (there usually is!).

      How advanced is the game? (rulebook? d’you have a playable prototype?) I played a few games that sound close to this (Summoner Wars, Magic, D&D4, Warhammer, Mordheim, Space Crusade (another collaborative squad tactics!), that Lord of The Rings cooperative card game (it’s interesting ’cause it doesn’t require a DM, is very challenging, and allows for some interesting deck building and specialization, players have to pick an element and compensate each other’s weaknesses), various computer turn-based strategy games), and would be interested in kicking around design ideas.

      How would individual battles work? Players together united against a DM’s army?

      • Andy says:

        How would individual battles work? Players together united against a DM’s army?

        Yep. The DM lays out a set of objectives (“Defeat the out-of-control necromancer,” “Return the exiled prince to his throne,”) And the players work out the military strategy, as well as diplomacy, necessary to winning. There’s two turn types: Strategic, where armies march around, scout, deal with diplomatic stuff, etc, and Tactical, where armies maneuver and clash.
        In terms of actual mechanics, I’m running it on a hex grid, which may make campaign design a little more difficult but makes the movement/assault mechanics much simpler. Especially because I want slightly different mechanics for frontal assaults vs. flank attacks.
        The rulebook is still in progress, I’m turning my vague outline into prose when I have spare time. EB is unlikely to be playable before the end of the year because there are so many other projects (like graduating from college and a long fanfiction project) in front of it on my plate. At that time, it will probably be tactical combat only, and then I can get to building the rules for the strategic component.
        The technology is generally at the late-Roman/early-Byzantine level, but Air does have transport and combat airships. I might figure out a golem or armored division subtheme for Earth if I can make it different enough from Earth’s existing heavy infantry emphasis.
        I like to think all of them require strategy, but the delicacy of most Air units requires them to stay well out of reach, and so you have to plan a few turns ahead. And Death has some truly clever ways to screw with an enemy. Death and Water are probably the most rogue-like, in thinking that rules and coloring inside the lines are for other people. Death has “Higher” rules embedded in the lore (don’t mess with the balance!), but illusions, driving your opponents’ troops to attack each other, or sending gangrene through their wounded are all fair game.

        • Emile says:

          Have you playtested bits of it? Do you have a satisfying combat mechanic? Do you mind being bombarded by these sort of “adviceish” questions? (is there a good term for those? “have you playtested it?” has undertones of “you should playtest it” and may come off as rude)

          • Andy says:

            I have run individual combat rounds between units, but not full battles. I’m not entirely satisfied with the combat mechanic, but I took the most concrete I could get (individual units make attacks, roll to-hit, armor-save, to-wound, and to-kill rolls on many D6s a la Warmaster) and if that’s too cumbersome I’ll have to figure out another mechanic that’s easier. Actual combat, especially close combat, I see as being somewhat abstract anyway, given that each unit is 1,000 foot soldiers or an equivalent number of monsters/critters/etc, and turns haven’t been given a concrete time period yet. Maybe an hour, giving 24 Tactical turns for every 1-day Strategic Turn.
            I’d welcome any advice. Especially since I’ve never actually played a DnD-type game, and my experience with Warhammer and 40K were all when I was younger, and the primary inspiration for EB was an old computer game called Fantasy General, I often feel way out of my depth on this. But this is perhaps the best way to learn – dive into the deep end and figure out how to swim.

          • Emile says:

            OK, I’ll spam you with ideas tomorrow then, it’s past bedtime 🙂

          • Emile says:

            But before going to bed, how I would approach designing a game like that:

            Make a prototype where units are cardboard cups turned upside down with a symbol drawn on the bottom (a stickman with a sword, a horse’s head, an eye, a wizard hat etc.), and have a die up there representing the unit’s life (that works if you have cups with a little border around the bottom, so stuff inside stays inside) (this is something I did working on a Heroes of Might & Magic game that was eventually canned by the way). (If you want to be fancy you can have a color paperclip attached to the edge of the rim representing the faction, so you can easily switch factions)

            And draw a terrain big enough so that the cups fit in a hex.

            Brainstorm interesting unit concepts and make cup-units for the most interesting ones, and play!

            Example concepts: each turn, neighboring allies heal one life; glass cannon (fragile shooter); damages all neighboring units, ally or enemy; suicide bomber; guy that gets defensive bonus on forests; melee guy immune to magic; neighboring units get +1 to fire-based attacks; guy that spawns another unit (or several) when killed, etc.

            Anyway, I should really got to bed, but this kind of stuff should be prototypable in a couple of hours, and might be enough to give a feel of what units are exciting and what are boring; which ones make good combos and which ones fulfill the same role in different ways (meaning they should probably end up in different factions).

          • Emile says:

            “roll to-hit, armor-save, to-wound, and to-kill rolls on many D6s a la Warmaster” sounds a bit complicated, have you tried making it As Simple As Possible, and then only adding complexity if it improves the game in a way you like? (e.g. more strategic depth, more unit diversity, better balance)

            For example, how about no rolls? Unit have life (a die) and attacks that can either inflict a fixed amount of damage (usually 1 or 2), with some variations per unit (eg. “1 dmg to two units touching each other,within 3 hexes” / “1 dmg to all neighbouring units”; “3 dmg to target unit within 3 hexes, 1 to self”; “2 dmg to an unit exactly 2 hexes away”).

            Or how about very few rolls, eg. each unit has a power (say usually 2 and 5), equal to the amount of dice they roll each fight: 1 or 2 are defenses, 5 or 6 are attacks, and you hurt your target if you got more hits than he got defenses. And some units get special powers on 3 or 4 (e.g. heal, or give mana to your sorcerer; or reroll an opponent’s die) and may get special attacks for specific dice patterns (e.g. three of same value: teleport out of combat up to fixe hexes away in any direction); vicious units do double damage on 6, armored units get two defenses on 1, etc. And some powers add rolls to allies or remove rolls from enemies. And the number of rolls you get could also double as your life.

            (I personally don’t think background story, or technology level, or time units need to come into play until the core mechanics are worked out and got the heck playtested out of ’em; interesting fluff is easy to add afterwards)

          • Ialdabaoth says:

            I’d second Emile’s idea; in fact, I’ve become a big fan of “custom dice” lately.

            I’d suggest that the dice have the following symbols:
            1 – (shield symbol)
            2 – (shield symbol)
            3 – (lightning bolt symbol)
            4 – (star symbol)
            5 – (sword symbol)
            6 – (sword symbol)

            If your unit rolls more swords than the other unit rolls shields, you win.

            Each unit can have one or two special abilities – one which corresponds to the star, and one which corresponds to the lightning bolt.

            For example, an extra-fighty unit might have a special ability that just looks like “{Star symbol} {arrow} {two sword symbols}”, which would read as “Treat each die that rolls a Star as if it was TWO swords”. An extra-defensive unit might have “{Lightning symbol} {arrow} {two shield symbols}”. Magic units would assign special effects to the star, the lightning, or both; maybe the lightning symbol is “give an adjacent unit one shield symbol chit that can be spent at any time during a dice roll” and the star symbol is “give an adjacent unit one sword symbol chit that can be spent at any time during a dice roll.”

            Does all that make sense?

          • Matthew says:

            (Apologizes for wandering off on a tangent)

            Combining the locally-popular themes of congnitive illusions and gaming, I’m curious if anyone else has noticed that many people seem to have strong preference between using dice-based randomness/probability and cards-based randomness/probability in games. It’s an interesting psychological phenomenon, since there isn’t really much of a difference (it’s not zero difference, in that card-based results are not independent until you reshuffle, but that’s a small thing).

            My own preference has actually flipped, such that I now find dice something of a turn-off in a game. I think this started with my being impressed at the cleverness of the Fate Deck in the Fantasy Flight version of Warrior Knights.

          • Ialdabaoth says:

            Heh. You REALLY ought to try out my MechBrawl game. Rule 1 of the “design goals” doc says “NO DICE”. It’s a tactical wargame with 100% card-based randomness.

          • Matthew says:

            If it were possible to play games with hidden info over skype, I’d already have been offering to game with you.

          • Ialdabaoth says:

            Sounds like I need to set up a VASSAL module, then.

          • Matthew says:

            this is something I did working on a Heroes of Might & Magic game that was eventually canned by the way

            There actually is a board game based on Heroes of Might & Magic.

          • Mark Rosewater, the head designer of Magic: The Gathering, has written a good article on the role of randomness in gaming and how it’s perceived by players. (There’s a specific section on how their market research has consistently shown that competitive Magic players don’t like coin flipping and dice rolling mechanics, even though they’re fine with the inherent randomness of a card game.)

            It uses some Magic-specific terminology but I think most of it is understandable even if you’re not familiar with Magic. I highly recommend giving it a read.

          • Ialdabaoth says:

            Yup. I tend to concur. Shuffle-randomness at least gives you a sense of control, in that you’re getting a randomized slice of your deck, which you then have near-perfect control over (YOU get to decide which of those cards to play, and in which order). The remaining unknowns aren’t due to randomness, but are due to hidden information (you don’t know what cards your opponent has drawn, either.)

            This somehow *FEELS* like much more control than initiating a perfectly random event, which immediately determines which branch the game proceeds down.

            Put more succinctly: randomizing my OPTIONS feels good, randomizing my OUTCOMES feels bad.

          • Emile says:

            I second the Rosewater recommendation, I read most of his columns.

            On cards vs. dice: I agree cards are usually more interesting (randomizing your options rather than your outcomes), though random outcomes are also a nice way of reducing analysis paralysis.

            There’s an interesting turn-based strategy game on iPad with dominion-style deck-building: http://empire-game.net/ (and little other randomness)

            Some games with the kind of custom dice Ialdabaoth talks about:

            * King Of Tokyo, which has plenty of powers depending on roll combinations or giving rerolls

            * Space Crusade, which has attack dice; you have white dice and red dice for heavy weapons, where you have more chances of doing damage with a red die (see this image )

            * Quarriors, where your creatures and spells are dice, and each creature has a card that explains what the faces do; several creatures can be represented by the same dice, but the faces will be interpreted slightly differently (eg. there’s a little star in a corner on some faces, for some creatures it means something)

            But I’m also thinking that the custom fight mechanic I described may not work *that well* for Andy’s game, because for a stategy game you’ll have to deal with several creatures attacking one, ranged attacks, etc. So a “one way attack” might make more sense (eg. I attack you so I roll, and see how muc damage that deals; the defender doesn’t roll but may have powers like “armor” (blocks one attack), “counter attack” (exceptionally can roll), etc.

          • Andy says:

            Thanks everyone for responding! There’s going to be a certain amount of automatic complexity – a real battle is very complex, and I wanted to model different kinds of attacks – Roman-esque infantry (Like the Earth-aspect Troll Infantry regiments) being very difficult to damage with arrows due to their shields, but other archers or cavalry, or infantry without shields, being much easier to knock out. I’m counting on small army sizes (The largest “Test Brigade” I have set up is 6 units) to keep things moving, but I wanted to be able to (roughly) recreate battles like Shiloh or Gettysburg with fantasy monsters. But I’ll look at what I can do to simplify the melee attack sequence.
            I do know how unit life is tracked – Each player will have a couple “unit cards,” in front of them, each about the size of half a sheet of letter paper. These cards will have unit stats, experience, etc, and a row of boxes to represent the Hits of a unit. A regiment of 1000 foot soldiers would have about 20 Hits, and each attack might Wound or Kill one Hit. Wounded Hits might be healed and return to action, Dead Hits can’t – one design limit is that even Death magic can’t bring people back intact. As a unit loses troops over the course of a battle, its effectiveness drops vs. fresh troops. Wounded Hit boxes would get one slash (in pencil!), Dead hits would get an X, allowing Wounded Hits to be ‘downgraded’ by certain magical attacks, (my favorite Death spell so far: Gangrene) or possibly when a unit is forced to retreat, but that might be too complicated.

        • Matthew says:

          @TAB

          Excellent article. My only disagreement is with the claim that Magic players are representative of the hobby overall. I happen to agree with the cards > dice preference, but I’ve definitely seen people on BGG with the opposite intuition.

    • Ialdabaoth says:

      I’d definitely try Death first.

      Incidentally, I started out with a nearly identical elemental system, before collapsing into five elements each with two polarizations. The final result being:

      Earth – {Petrify/Shatter}
      Water – {Freeze/Dissolve}
      Wood – {Poison/Heal}
      Air – {Obscure/Blast}
      Fire – {Illuminate/Burn}

      • Andy says:

        Yeah, the first concept had a Light Side and Dark Side of each element, a Divine element, and a Blood element, but all that got cut as I’ve been paring away at the world. I might still introduce Dark Side units and spells – units and spells that can only be used by those who are very clearly Evil. But I don’t have good options for all 6 and I want to have the basic mechanics done before I come up with any more units.
        I do like yours though, it just doesn’t have Death, which was a core part of the concept. 🙂

    • Well-meant advice: consider really paring this back; balancing that many units is a lot of work, and it’s very easy to envision something that will take years and years to finish.

      If you’d offer just Plant and Death and specific scenarios (no strategic layer, which frees you from needing to properly cost units), that’s still a very ambitious first project.

      (To answer your question: I’d look at Earth or Fire first, but I don’t play multi-player games in the first place so I wouldn’t be your target market anyway.)

    • Emile says:

      (Splitting a subthread)

      You seem to have two different kinds of specialization going on:

      * Between players, where some take on the roles of healers, some tanks, some DPS, etc.

      * Between units a player controls

      .. so this creates some tension whereas you want a player to have an interesting diversity of units, but on the other hand, you want him to still have a specialized role in the team, which might encourage one playe rto take “only healing units”, another “only shooting units”, another “only melee units”, etc.

      Ways to approach this!

      * Forget about D&D4ish player role specialisation, players just have armies. In fact, players could also just bash each other and forget about the whole “DM” thing, so you have a much simpler tactical combat game.

      * Each player only controls one unit instead of an army, a bit like D&D 4.

      * Put the focus on “player” actions instead of “unit” actions, for example the player can do five actions, dispatched any way between his units (possibly with some constraints; slow units can’t do more than one move action, etc.), or the player has action cards/spells that greatly affect his units; this rewards both player specialization (the healer will often spend most of his actions on healing) and unit specialization within a faction (the healer will still want a bit of diversity in his troups (in some situations shooting or melee will turn out more useful than healing).

      * All of a player’s units must stay grouped, e.g. touch each other, so in effect kind of act like one macro-unit.

      * A player will need both “action” units (attack, aggro/defense, heal) and units that are only needed “internally” in his army for logistics (drawing cards, getting mana, giving extra actions to his units, summoning new units); so in terms of battle effectiveness a player has a specialized role but internally still needs to think about how his army is built. This can also add flavour by having each faction/element have it’s logistics system.

  13. RCF says:

    “but if there is some big community social event”

    There’s a house party the 20th: https://groups.google.com/forum/#!topic/bayarealesswrong/ZTaXCni_NjM (hopefully it’s okay that I’m cross-posting this)

    Also, there’s a large kink event that weekend.

  14. RCF says:

    “Big thanks to Bakkot and Alice for adding the script last month that highlights new comments in green and makes this place much more readable.”

    Are you sure that “green” is the correct label?

  15. potatoe says:

    I have heard that it is impossible to like electroswing and dubstep at the same time. Is this true? My anecdotal evidence is in favour of “no”. Which genre is more rational? Is there some sort of optimization process on beeps and boops to get the most electro music of all?

    • Olivia says:

      I don’t know. I enjoy listening to electroswing. I also kind of enjoy some dubstep if it was long enough since I last heard a dubstep song.

      But in my experience, if I listen to ten electroswing songs in a row, for each song I start to pick up finer and finer distinctions so that I feel that the songs are more and more different from each other. Whereas if I listen to ten dubstep songs in a row, for each song I get a stronger and stronger feeling that they are really the same song with just a few licks borrowed from whatever they are ostensibly ‘versions’ of.

  16. Joe says:

    Hey Scott I thought it was cool that you included some Catholic sites on the rationality map. Have you or Ozy read any interesting theology since Feser’s “TLS”?

  17. Sniffnoy says:

    Another thing from the meetup that I think is worth bringing up here — well, for other people to talk about; I don’t think I can really intelligently comment on the matter, for the most part!

    Spearman’s law of diminishing returns vs. “Why the tails come apart”

    The latter was posted on LessWrong back in August, pointing out that even if two things are positively correlated, the things that max out the one are not going to be the things that max out the other. (See also: Berkson’s paradox.)

    But in some cases where “the tails come apart”, this simple statistical effect might not be all that’s going on. In the case of intelligence, the tails seem to come apart much more strongly at the high IQ end than at the low IQ end, a sort of asymmetry that should not occur if we were just seeing the effect that Thrasymachus writes about.

    So at the least, we should possibly be careful about attributing this sort of thing to pure statistics. But it also raises the question if this sort of thing happens elsewhere. (I seem to recall people on LW suggesting the same thing for athletic abilities?) Is there something more general going on here? (These are not questions I can answer at all!)

    • Adele_L says:

      >But in some cases where “the tails come apart”, this simple statistical effect might not be all that’s going on. In the case of intelligence, the tails seem to come apart much more strongly at the high IQ end than at the low IQ end, a sort of asymmetry that should not occur if we were just seeing the effect that Thrasymachus writes about.

      The key here is: “The fact that a correlation is less than 1 implies that other things matter to an outcome of interest.”

      Let’s imagine a complex machine which manufactures soccer balls. At the factory, each machine is tested along several metrics, overall speed, stitching quality, precision, air pressure, etc.. and given an overall score (low scoring machines are still sold, but for a discount). Later, the machines have been producing soccer balls for five years, and their overall quality is assessed.

      I expect that if you plotted the factory scores vs. the overall quality, you would see the tails come apart more at the high quality end than at the low quality end. This is because all the components of the factory test are pretty important – if it does poorly on these, it’s just going to suck at making balls. Conversely, they do a pretty good job at catching all the major failure modes, and thus the tail will be pretty tight (If there *were* some really bad machines which did alright on the test, whatever these got wrong would be added to the test). But at the high quality end, there are other, more subtle things that matter to quality than what the factory test measures. So the tails come apart more here.

      • Sniffnoy says:

        Hm — but if one segregates into high end and low end of overall quality, and looks at the correlations between the individual components of the factory score, shouldn’t one expect to see tighter correlations at the *high* end and looser ones at the *low* end, reversing the effect? Because if each component is pretty important and without it you’ll get a crappy machine, then there will be many reasons something could score low, but to score high in overall quality one would need high scores in each component.

        • Adele_L says:

          You’ll get a decent ball if you have a high score, but it turns out that what makes a soccer ball go from very good to great is whether or not the thread for the seams has been dyed with a few days longer than normal. Also the optimal pressure depends on the altitude of where the ball is used.

    • covaithe says:

      Thank you! I was sure I had read something recently that was relevant to that conversation, but I couldn’t remember where. It was of course “why the tails come apart”.

  18. Brett Bowman says:

    I’ve got a couple of air mattresses and some extra space just off the T line in the Dogpatch if you still need a place to stay.

  19. Matthew says:

    Musings on the ship of Theseus.

    If you replace the mainsail with a larger cloth on the ship of Theseus, it seems to me it’s still the Ship of Theseus, with a bigger sail.

    If you build wheels on the Ship of Theseus to make it amphibious, the line gets a bit blurrier — is it still a ship?

    If you drag the Ship of Theseus up on the beach, secure it, and refit the interior to serve as housing, I think I’d say it’s a barracks made from the Ship of Theseus, but it isn’t the ship any more.

    I have related thoughts on continuity of identity and value drift, but I’ll save those for a bit and see first if anyone finds the first part of this comment at all interesting.

    • Whether or not something is a “ship” is a property of the map, not the territory. Words mean whatever people think they mean, and they don’t map precisely onto properties of things in external reality.

      I think one of the Sequences talked about this but I can’t remember which posts were relevant.

      Identity, as a special case, can potentially yield interesting philosophical questions, but the Ship of Theseus in general doesn’t.

      • 27chaos says:

        I think I disagree with your assertion that identity is sometimes philosophically interesting. I don’t think it’s ever interesting. Unless you can give an example?

        I suppose if you consider reference class problems for empirical predictions as philosophical and identity based, that would be an exception. Other than that?

          • 27chaos says:

            It would be incorrect to expect to win the lottery, even after making that precommitment. No idea why anyone would think otherwise, I genuinely don’t understand.

            Also, anticipating W and ~W simultaneously is not a sign that something is wrong in that scenario. It is just the sort of thing that happens if you’re making simulations of yourself. So biting horn 2 would be okay with me, though again I don’t think it is necessary.

            Also horn 3 doesn’t even resolve the trilemma. Change the question to “what will Eliezer+1’s memories contain” and the problem still exists.

            Regarding the other link, it seems clear to me that identity exists in degrees. I don’t see that as problematic.

        • I think identity as it relates to consciousness is the source of interesting philosophical questions. In particular, I believe that there is something to be explained that does not depend solely on the laws of physics. I’m aware that this is not the default position in the Less Wrong-influenced rationalist community, and am not surprised to encounter disagreement. I also haven’t yet developed a good response to the most common objection to this position (that consciousness affects the physical universe because conscious beings know that they are conscious).

      • Paul Torek says:

        Words mean whatever people think they mean, and they don’t map precisely onto properties of things in external reality.

        Such a strong map/territory dichotomy seems unworkable. The map is part of the territory, in many cases. That’s why there’s a big red dot saying You Are Here. Excellent book on the subject.

        P.S. Has anyone else noticed that blockquote cite=”blah” does nothing other than what plain old blockquote does?

        • Anonymous says:

          cite=”blah” is defined to do nothing by the standards. It is only for computers to read. I do not know if any computers actually bother to read it.

          Added: Here’s the standard. Actually, it is defined to be a URL, not, say, the name of the person being quoted. Which makes sense if a computer is going to make sense of it.

          • Emile says:

            Sure, but the help underneath the textbox has “<blockquote cite=””> under “You may use these HTML tags and attributes” (maybe there’s an implied “… but they won’d do anything, hahaha!”).

            The cite attribute does end up in the final html, so it is indeed up to the browser do decide whether to display it or not (and as you say it’s not in the standard); it would still be nice if the help section didn’t include that bit.

            I wonder what the q tag does…

            This is in a q tag

            …oookay, it just adds quote marks, cite=”blah” is ignored too.

        • Sorry, I don’t quite understand your objection. Of course our mental models (“maps”) are in our brains, which are in physical reality. My point was that deciding whether an amphibious Ship of Theseus is a “ship” or not says nothing about the actual vehicle itself; it’s just about how we use the word “ship”. Does this address your objection or am I still missing something?

          And yes, as Anonymous said, the cite attribute of blockquote doesn’t generally have any visible effects in browsers. It’s one of those things that somebody who was developing HTML thought would be a good idea that people would use, and it turned out not to be.

          • Paul Torek says:

            My objection was kind of off-base. You’re questioning the “mapping function” (to put a name on it) rather than the reality of the map.

            But still, the map:territory correspondence often goes deeper than it appears. System 2 doesn’t always pick up on what System 1 knows. The decision whether to precisify “ship” in one way versus another might be very arbitrary or very nearly forced – it’s hard to tell at first glance. This is part of why judges make the big bucks.

        • Vertebrat says:

          I have used some browser (I forget which, but it’s not Firefox) that added an entry in the context menu to follow the cite link. Since there was no way for the user to know when this option existed, it was useless. ISTM it would be enough to simply make the entire blockquote function as a link (showing the cite and a suitable cursor on mouseover, but not otherwise visually distinguished).

  20. 27chaos says:

    How can we avoid overfitting while building accurate models of poorly understood complex systems? What tests can be done to verify a seemingly excellent model is not the product of overfitting?

    • Princess_Stargirl says:

      The only method imo is to actually use the model to make predictions.

      interesting note:

      Sometimes you can even make predictions of events that have already happened, at least if you only trying to avoid fooling yourself. David Friedman talks about how his first economics article was on predicting the shape of nations. Someone (I think stiglitz) rejected his first version as it made no prediction. He later made a more precise theory and correctly predicted the general features fo the map at various points in time. Of course this only vindicates friedman’s theory if we trust him that he finished his theory before he looked at the data (otherwise he could surely find a way to fit the data).

      • RCF says:

        A tactic that works on similar principles is to start taking subsets of your training data, and seeing how stable your predictions are. You can also, instead of taking subsets, introduce perturbations into the training data. If you start finding that your predictions vary wildly depending on what training data you include, that’s a big warning sign of overfitting. And of course, a complexity penalty (regularization) is a popular tactic.

        • 27chaos says:

          Why would overfit models be more vulnerable to perturbations? Simply as a consequence of increased complexity? Sounds like a neat trick, certainly.

          • Douglas Knight says:

            Isn’t vulnerability to perturbation the very definition of overfit? What do you think it means?

          • 27chaos says:

            Overfit means that it doesn’t predict anything because the model is built wrong. The model looks correct when evaluating past data, because the model is built around the data.

            What do you mean by “perturbation”? I thought we were discussing slight changes to the data. I don’t see why those would break a model. Did you mean slight changes to the model itself?

  21. Kiboh says:

    I’d like to make sure anyone still interested in Scott’s post on ‘Investment and Inefficient Charity’ knows that Donor Advised Funds are a thing.

    The idea is that you donate your money to a third-party nonprofit like Vanguard Charitable, they invest it in your stead, and they transfer it to the charity of your choice when you decide the time is right. You get the tax relief – and hopefully the fuzzies – at the moment the money (irreversibly) leaves your bank account, but still keep the two main benefits of Hanson’s save-up-and-donate-on-your-deathbed strategy: compound interest, and the opportunity to leave big decisions to your smarter and better-informed future self.

    • Paul Torek says:

      Working on this, but the forms are somewhat annoying, especially if your spouse is busy and hates paperwork.

      • Did you decide to set up a donor-advised fund because you feel that your money will do more good later, because you expect interest to compound faster than positive effects of an early gift, because you just don’t know yet, ..? I’d be interested in your reasoning.

        • Paul Torek says:

          Not compound interest, because I expect that on that score, an early gift would outperform my investments. Rather, uncertainty about where I can do the most good, combined with pretty high confidence that changes to my future-self’s understanding of “the most good” would be regarded as benign by present-self.

  22. no one special says:

    If I could complain about only one behavior that I see here on Slate Star Codex, it would be lobbying to ban users*. I strongly recommend that we come up with something to counter this, and instead use “report and move on” for people who have gone past what we want to allow.

    Unfortunately, I have no strong suggestion to offer for what to do about people who do lobby. My weak suggests would be to reply with “report and move on”, or to report the lobbying and hope it gets deleted.

    *This is based on the assumption that lobbying to ban someone is primarily (Hanson!) a case of displaying your virtue to the community by slamming on someone who isn’t maintaining the community norms.

    • blacktrance says:

      I think lobbying to ban users is a good thing. It shows strong dissatisfaction with especially bad posters.

      • Zathille says:

        It certainly could, but it could also be used as a tool to squelch unpopular views. As long as the ones making the case for people being banned are explicit about their reasons, and thus, forced to substantiate their case, I have little problem with it. Temporary bans coupled with making explicitat the reason behind a ban has a chance to even turn bad posters into better ones, from my experience.

    • Multiheaded says:

      As someone who’s both been lobbied against and am lobbying to ban someone (namely, Steve Johnson), I’m not sure how to feel about this.

      P.S. fuck grammar

    • RCF says:

      One purpose of posting a comment in addition to pressing the report button is so that other people know that the poster has been reported. It reassures people who might wonder whether they are the only ones taking offense. It also prompts SA to, if he does not ban the poster, to offer a clarification as to why.

      • Wait, the “This comment has been reported” message is visible to people other than the reporter? I’ve only ever seen it on spam that I reported myself.

        • Anonymous says:

          No, RCF means that lobbying for bans is a way of telling people that you’ve reported a comment.

          • Ah, understood, thank you.

            …And now I’m thinking of what it would be like if that message was visible to everyone. This would probably be a bad idea for all kinds of reasons but it would have the positive effects RCF describes.

  23. houseboatonstyx says:

    1. Big thanks to Bakkot and Alice for adding the script last month that highlights new comments in green and makes this place much more readable.

    Adding a text string that can be searched for, would save wear on the Page Down button.

    • Anonymous says:

      click on the [-] in the upper right corner of the screen and navigate that way.

      • Bakkot says:

        (“Navigate that way” means click on the listed comments which are revealed by clicking [-]. It’s not as obvious as I’d hoped but I don’t have a more intuitive solution which is also as unobtrusive as the current way.)

    • RCF says:

      If there were a way to keep the list of new comments from resetting when I post a comment, that would also be good.

      • Right-click on reply and open a new tab– that way you won’t affect the tab you’re reading in when you reply.

        Sometimes it’s useful to copy the time in the “comments since” window so that you can paste it back in.

      • no one special says:

        You can edit the time in the box to change when comments are highlighted since. I just note the time before I submit, and set it back once the page has reloaded.

  24. A number of users, including myself, have expressed interest in a way to identify particularly high-quality comments. At the same time, a number of users have identified problems with explicit voting systems.

    I’ve been thinking about how one might design a system that satisfies all these criteria. Here’s the best I’ve come up with; I’ll call it “Comment Radar”.

    Each comment has a button that means “I think more people should see this comment”. When somebody clicks on it, we’ll call that a “ping” (not to be confused with a pingback). What makes this different from an upvote is that there’s no public ping count, reordering of comments based on pings, or anything like that.

    On the sidebar, there’s a “Comment Radar” panel. In it is the text (truncated for length if necessary) of a comment which has received a high number of pings, along with a link to that comment for its original context. Once you click on the link, or click a button in the panel to dismiss the comment, that comment won’t appear in the panel for you again.

    Would something like this be a good idea?

    • Kaminiwa says:

      So… upvote only, no downvote… you can’t see the exact count, and a magic algorithm decides which one I should see… I’m going to point out a lot of people get upset with Facebook doing more-or-less exactly the same thing.

      I’d also say it breaks because of the context problem: If I am reading a thread about X, I want to read good, high-quality commentary on it. Having something pop up on radar 2 weeks later isn’t helpful because I don’t remember the rest of the discussion as clearly. Having something pop up on the radar about Y, which I don’t want to read comments about in the first place, just gets worse.

      I could see it being an awesome solution for identifying little gems that might get linked as “top comment” at the start of an open thread, or for blogs that are more mono-topic, but it doesn’t seem super useful here.

      • I wouldn’t want something to pop up two weeks later. Probably it should only apply to threads that still have people actively posting comments.

        • Roxolan says:

          Even so. The only time I read SSC comments, it’s right after reading the article above them. I don’t think I’m alone in this.

          (Admittedly, the system you suggest doesn’t do me any harm. It just won’t improve my experience the way some kind of upvote-and-sort system would.)

          • To clarify, this proposal wasn’t really intended to solve the general problem of how to read comment threads. On many SSC threads I read all the comments, at least for a while. But often a thread gets away from me; to give an extreme example, there’s probably at least one brilliant comment in the “Radicalizing the Romanceless” thread that I’m never going to see because I’m not about to read through 1200+ comments, most of which are people responding to each other in debates that I don’t want to read because they would make me sad.

            That’s the problem I was going after: if I’m no longer reading every comment in a particular thread (or never was), how do I know if I’ve missed something exceptional? (The comment I nearly missed that started me thinking about this problem wasn’t part of an argument at all; it was a brilliant pun from the “Fifty Swifties” thread.)

            This was simply a random idea for how to do that without running the risk of enforcing ideological conformity, as voting systems tend to do according to many commenters here. Sort of like “Comment of the Month”, except without just one every month and without Scott having to do all the work.

    • Or, for simplicity, clicking the “this is good” button could put the comment back at the top of “recent comments’.

      • I think you’d want to distinguish good comments from recent comments. More importantly, just having one person click the button is a pretty weak signal, and I’d want a stronger one before bringing something to people’s attention.

  25. Alex Godofsky says:

    Is there any way you could add a “hide thread” button like reddit’s [-] button? The comment threads here get [i]really[/i] long and it is very painful to follow them without being able to hide the higher-up threads I’ve already read.

    • Alex Godofsky says:

      Ugh, too used to BBCode.

    • Kaminiwa says:

      I will happily second this.

    • Alejandro says:

      It exists already – I have Hide/Show buttons on each comment that collapse and expand the whole subthread, as of a few days ago. Maybe they don’t work on your browser?

      • Douglas Knight says:

        Perhaps the key missing detail is where they are: next to the reply button.

      • Alex Godofsky says:

        Bizarrely, I had to enable cookies for it to show up.

        • Bakkot says:

          Eh… strictly speaking, you probably had to enable local storage, not cookies. Local storage is used for the comment highlight, and if it fails so will the whole script, which includes the injection of the “hide” button.

          • Alex Godofsky says:

            I’m sure you’re right, but I had to click the cookie icon in Chrome and select the option “Allow slatestarcodex.com to set cookies”, which from an end-user POV reads as “I had to enable cookies”.

        • RCF says:

          Perhaps the cookies are tracking which comments you’ve hidden?

          Also, there is an edit option for one hour after posting.

  26. suntzuanime says:

    I had something I wanted to remember to post in the next open thread, but I forgot what it was.

  27. Paul Torek says:

    In the meetup I said something wrong about the CIA’s “vaccination” program to get blood samples from Al Qaeda leaders. The vaccines weren’t fake – it seems to be a real vaccine, but only one out of three standard doses were administered. Source: The Guardian

  28. BenSix says:

    Czesław Miłosz’s The Captive Mind is recommended as a provocative analysis of the submission of intellects to ideologies and spirits to totalitarianism.

    • Multiheaded says:

      Reading now. Fascinating, I haven’t quite seen anything like this written in Russian by Soviet dissenters.

      EDIT: but also, of course, veers into incredible racism/chauvinism. Sure, we are the evil savage Slavs who are so backward and beneath the noble enlightened Western Slavs! This cheapens the author’s whole thing about Stalinism.

      • BenSix says:

        If it’s any consolation, he could be as patronising towards Poles who embraced national traditions as Russians. He was certainly an Anglophonic Conservative’s ideal Slav but it doesn’t make the book much less interesting.

      • social justice warlock says:

        Trotsky trots the same thing out in “Revolution Betrayed” – of course Stalin was a thug, it was in his barbaric Asiatic blood, and furthermore,

  29. Quixote says:

    I started reading quantum computing based on the review post. I like it a lot and recommend it to anyone Mathy

  30. A few days ago, a Facebook friend posted a link to an NPR piece about fast food workers campaigning for a $15-per-hour wage, emphasizing the quote “Why should middle-class taxpayers subsidize corporate profits?” I linked them to Scott’s post which dealt with this question, and a protracted argument ensued.

    While my Facebook friend was not opposed to a more expansive welfare state, they strongly believed that anyone who employs a person full time should be made to bear responsibility for paying that person a living wage, and that having the government assume this responsibility instead is tantamount to a corporate subsidy. One of the questions we argued over was, what happens when employers are forced by the government to increase wages above the current market equilibrium? (Classical economic theory suggests they reduce the size of the workforce, but neither of us suggested that as the outcome, presumably under the assumption that demand for labor is relatively inelastic.) I argued that they pass the cost on to their customers, so the minimum wage ultimately amounts to a regressive tax. My friend argued that their prices are already as high as market conditions allow, so the cost has to come out of corporate profit margins and so amounts to a progressive tax.

    (I’m steelmanning slightly as my friend didn’t use this much economics terminology.)

    While I still think an expanded welfare state is a better idea than minimum wage, I’m beginning to suspect that the case for this proposition may not be as airtight as I had previously thought. Does anyone have any arguments I should consider?

    • Jack says:

      I found this question come up in a previous pro/anti minimum wage conversation and was disturbed to realise I really didn’t know.

      My current best guess is:

      * A much much much better welfare state (eg. guaranteed basic income) would be better than a minimum wage. Eg. it might be a better reflection of the natural price of things if charity work, internships, etc could be unpaid, but that no-one had to suffer in order to take them up.
      * In order to get that, you might need better corporate taxation.
      * It’s possible an intermediate level of welfare might inadvertently act as a corporate subsidy allowing companies to pay even less than they could if they had a complete monopoly. We have experienced this problem in real life 🙁
      * It’s likely that at the moment, most jobs which pay minimum wage _could_ afford to pay more, and that would reflect a fair market price better, redressing the disparity that minimum wage workers have too little bargaining power. But some really can’t, and it’s hard to say when those companies need time to grow (and implicit government subsidies are good) or when they’re worthwhile for other reasons (eg. charity, etc) and when they think they’re doing something worthwhile, but are actually not (which is sad and I don’t know how to deal with it).

      But I’m really not sure if that’s correct, I’ve occasionally seen people with more economic knowledge than me talk on the issue, but haven’t been able to separate the facts from their biases.

      • It’s possible an intermediate level of welfare might inadvertently act as a corporate subsidy allowing companies to pay even less than they could if they had a complete monopoly.

        See, this is the part I have a hard time believing. If the government stopped paying welfare benefits to people whose full-time jobs weren’t paying enough for them to live on, employers wouldn’t raise wages to make up the difference; such a policy would only hurt workers’ well-being, not corporate profit margins. Maybe you could come up with some kind of supply-side explanation for how this isn’t in employers’ interests over the scale of the entire economy, but I’m more interested in looking at more immediate effects.

        • Zorgon says:

          The UK has both a (rather pitiful) minimum wage and a low-earner subsidy called “Worker’s Tax Credit” which supplements low earnings to a basic standard of living. We also have a means-tested, if sometimes punitive welfare system for the unemployed.

          In practice, this is a massive subsidy to companies that employ people on zero-hour contracts and part-time labour in order to avoid what little employee protection still exists in the UK. Wages are permanently depressed at the bottom end because there’s no need to compete for workers when a) youth unemployment has been horrifically high for decades and b) tax credits automatically supplement whatever pittance you pay.

          Thing is, this is pretty much fine for the companies and the workers; although it does mean the workers have virtually no employment protections, they still have more than most US low-end workers and they have the NHS and State Pensions (for now, at least). The taxpayer is the primary loser here.

          What gets me about this is that it’s almost a Universal Basic Income already. We’re a tiny step away from it. But we’re still stuck in our post-Victorian “labour is virtue” mindset, so instead of acknowledging that we have no real industrial base and that a large portion of our population will never be able to do useful work, we implement US-style Workfare schemes that force unemployed people to do work that was previously done by low-paid workers in return for their welfare payments.

          • So I looked this program up on Wikipedia, and apparently if you’re a beneficiary of this program your wages are effectively taxed at a marginal rate of 73%? Like, have I failed to properly understand this? Because if not it seems like that might be maybe kind of sort of part of the problem.

          • Anonymous says:

            High effective marginal tax rates are endemic in welfare programs. The worst is: in America, earn too much money and lose medicaid.

          • Zorgon says:

            No, the name is misleading and the Tax Credits scheme has nothing to do with taxation other than that it’s run by Her Maj’s Revenue and Customs (HMRC) rather than the Department for Work and Pensions.

            It doesn’t garnish your wages, your Tax Credits entitlement is instead reduced according to increasing marginal levels at higher rates of income. It’s designed that way because the taxman is best equipped to perform means testing the same way it calculates taxation bands.

            It really is just a low-earner’s wage supplement.

          • Anonymous says:

            Zorgon,
            Taymon is correctly reading the wikipedia entry, which I assume is correct. The issue is not whether it is called a tax, but what is the effective tax rate. If I am on this program (at the right income level) and I work an extra ten minutes this year to get paid an extra £1, how much more money do I have in my pocket? 27p. Maybe I won’t bother.

          • Zorgon says:

            Ah, I see what you mean. Yes, I’ve personally fallen foul of that particular problem before now.

            To make matters worse, the UK also has a means tested Housing Benefit, which is very commonly used by low-waged workers. So it can frequently be the case that a small increase in wages isn’t worth it.

            On one occasion I actually asked my boss to hold back my planned £1500pa pay rise until he could make it £3000pa, as between moving me above the base Income Tax bracket, killing my Tax Credits, Housing Benefit and Council Tax Relief and restricting my wife’s means-tested disability benefits, the former would lose me significantly more than £1500pa.

        • ADifferentAnonymous says:

          Seconding your skepticism that first-world wages are currently set by the amount employees need in order to survive. And +1 to reframing it as “if we cut benefits will companies raise wages” rather than “if we add benefits will companies lower wages.”

    • Andy B says:

      It’s true by hypothesis that fast food prices are already as high as current market conditions allow, but a minimum wage changes the market conditions, so your friend’s argument is empty. My suspicion is that the demand for fast food is rather less price-elastic than he would wish, because the speed and convenience of fast food mean that there’s no proximate substitute for the industry as a whole. There’s a big gap between getting a Big Mac from a drive-through and going inside to order a Chipotle burrito.

      The tax incidence is moot, however, because in reality, a $15 wage would only accelerate this: http://singularityhub.com/2014/08/10/burger-robot-poised-to-disrupt-fast-food-industry/ .

      As a general point, I will hazard to state the Econ 101 argument that market prices are information and when you control prices the system loses the information and Bad Things Happen. In the case of the minimum wage, the system is screaming at the top of its lungs that we have far more unskilled/low-skilled workers than we have useful work for. Sticking our fingers in our ears by raising the minimum wage is a way of ignoring that information, and if we continue to ignore that information we won’t actually be fixing the problem.

      • The interesting thing is that (so far as I know) there’s no clear evidence that minimum wage laws increase unemployment– I expect that if the microeconomic argument is true, there’d be at least some evidence for it.

        • suntzuanime says:

          I’ve heard it floated that employers mostly just cut spending on benefits and other things that make employees happy (e.g. in a factory they might decide not to buy floor mats to ease the pain of the workers being on their feet all day) in response to minimum wage hikes, so that the actual cost of labor remains the same, and the jobs pay more but become shittier.

          Obviously there is a limit to how much money you can save by making a fast-food job shittier, and it might not cover all the way up to $15. But it’s an explanation for why marginal increases in the minimum wage don’t seem to cause unemployment.

          • Matthew says:

            In that specific example, the higher minimum wage is obviously a superior policy. Employees who value the mat more than the money will just buy their own and bring it with them to work.

            Tax subsidies for health insurance are criticized on similar grounds; just give workers more wages and let only the ones who need to spend it on health insurance/medical care do so.

      • See, I hear a whole bunch of things like this where pro-free market economists say that according to the laws of economics it has to work this way, and cite a whole bunch of work that backs this up. Then liberal economists say that no, in the real world it actually works that way, and cite a whole bunch of work that backs that up. And I have no idea who to believe because I’m not an expert. (Scott’s written about this phenomenon on several occasions too.)

        I have no expectation of being able to figure out which theory of macroeconomics is correct. But I do expect that it should be somewhat orthogonal to the more mundane micro-level question, “If the minimum wage for fast food workers went up to $15 per hour, would McDonald’s raise prices to compensate?” That’s what (this part of) the argument was about, and I think it ought to be possible to form an educated guess.

        • Daniel Speyer says:

          There are countries in Europe with much higher minimum wages. McDonalds functions there, with slightly higher prices. I don’t know if there’s a greater degree of automation.

          • Anonymous says:

            This is the foil of laissez-faire evangelists: there are real world examples of the sky *not* falling when the government intervenes.

          • This is weak evidence at best because economic comparisons to different countries are always somewhat apples-and-oranges, but it still helps.

            I’d be curious to know if they operate on a lower profit margin there.

          • The Anonymouse says:

            How much is “slightly”? Any commenter in Europe care to give some current prices?

            I frequently eat fast food. I primarily do so because of the low cost and high convenience (in that order: I would drive an extra bit for my preferred restaurant, but a twenty cent increase on a dollar menu item will cause me not to order that item). When I lived in Sweden, the first time I walked into a McDonald’s there, I looked at the price of a Big Mac, thought “wow, I could buy a whole pizza for that,” and never went back.

          • I had totally forgotten that the Big Mac Index existed! This gives me an idea for a (rather sketchy given the nature of PPP) quantitative analysis: if you divided a country’s Big Mac Index by some other, more respectable measure of its purchasing power parity, the resulting number would be a relative indicator of fast food prices. Then you could graph that against things like minimum wage. I might do this if I can find time.

        • The Anonymouse says:

          Speaking as a dollar menunaire, it seems that the current rising cost of beef is bumping burgers, but not chicken sandwiches, out of various dollar menus. No, I haven’t done a rigorous study, but it seems that if the fluctuations of the price of their beef inputs is enough to raise the cost of an item ~25%, doubling the cost of labor inputs will most certainly raise prices.

          Of course, it’s better populism to just handwave and say that the corporate overlords will just eat the loss, but that has never been my observed reality. 🙂

  31. lmm says:

    Did you watch that movie recently too then? That’s an interesting piece of synchronicity.

    >3. There’s been some discussion of improving the comment sections of very controversial posts (eg on feminism) by closing comments there, then making a comment section on a separate thread. The hope is that all the random people linked there by Reddit and Instapundit and whoever get confused and go away, but other people who specifically read this blog will find it and be able to talk about it. I’ll probably try that next time I’ve got something controversial to say.

    This reminds me of the various things 4chan tried to keep its quality high. Perhaps you should try blacking out the comments with CSS.

  32. Shmi Nux says:

    Do you know of any examples of pleiotropy which are just as funny as http://www.smbc-comics.com/index.php?id=3476#comic but real? Or semi-real.

  33. gwern says:

    Praise for SSC: http://www.themoneyillusion.com/?p=27448

    I can tell he’s a great blogger by just reading a few of them.

  34. social justice warlock says:

    I notice that I am noticeably meaner here than I am elsewhere (either online or in person.) I am neither sure why this is so (mere proximity to people I strongly disagree with isn’t it) nor whether it is healthy for me to continue posting here.

    • Multiheaded says:

      Senpai! It might be because of the very mild gaslighting that our host is (unintentionally, I’m sure) doing to us; he claims to have high standards for the kind of people we disagree with, but then picks through the worst of their shit to find something worthwhile + doesn’t ban the worst of them like he promised to, and our side gets nothing but the law. It’s aggravating.

      • Nornagest says:

        Speaking as a third party, I don’t see much in the way of law over either side.

        • CaptainBooshi says:

          Speaking as someone who is on the social justice side, but doesn’t comment very much, I concur that I don’t see really see the law being laid down much on either side. Scott’s consistent misinterpretations may be aggravating (and I often find them so), but it’s just unfair and not true to say that our side gets nothing but the law.

          • Paul Torek says:

            I am getting into an annoying habit of agreeing with everything CaptainBooshi says. But Scott’s intermittent law-laying strikes me as a pretty good balance so far. I think in this context, it only takes a few Examples Being Made to get most folks reasonably in-line.

        • Anonymous says:

          Same here!

        • social justice warlock says:

          Scott has stated both that he intends to lay down the law on NRx and that he’s keeping Multi based on affirmative action, so he’s being more procedurally fair in practice than he is in principle.

      • memeticengineer says:

        If the leftist side got nothing but the law, you would be permabanned by now, given how often you post things that have no obvious purpose but to provoke. FWIW I’m glad you’re not. And I do think there’s some NRx folk who said ban-worthy things in recent threads, who did not get a ban. But Scott isn’t actually being harsher against your side.

      • Zorgon says:

        Can we all just agree to quit with the accusations of gaslighting against people who aren’t actually engaging in campaigns of psychological torture intended to deconstruct a personality?

        The word is now almost meaningless. Disagree with someone about the facts? Gaslighting. Disagree with someone about the interpretation of the facts? Gaslighting. Apply a comment policy in a way someone doesn’t like? Apparently now “very mild gaslighting”.

        I’ve seen the film “Gaslight”, incidentally. I occasionally wonder how many people who (ab)use the term actually have.

        • no one special says:

          I am +∞ on this proposal. I very much want to reserve gaslighting as a term for personal abuse. You cannot gaslight groups. It would be extremely difficult to gaslight someone in public. I’m close to saying that you cannot gaslight someone that you’re not in a relationship with, but I’m still waiting for the double-blind tests to come back on that one.

          Okay, this is one of my hot buttons. Sorry.

          • Zorgon says:

            I’d say it’s not so much that it requires a relationship, but more that it requires a trust dynamic that gives you special reason to believe the person who is manipulating you. You can feasibly have that outside of a romantic relationship. I’d argue, for example, that it’s possible to be gaslighted by your doctor or your teacher.

            (It’s one of my hot buttons too.)

          • no one special says:

            Privileged trust dynamic seems legit.

            Now how do we convince all the commentariat on twitter to restrict themselves to this limited form? 😉

    • The Anonymouse says:

      I think you hit the nail on the head. If you’re anything like me, the reason you feel less charitable here is precisely because we have a community that is not an ideological echo chamber. There really is a diversity of opinion argued.

      Which, not coincidentally, is the exact reason I love the SSC comment section so much. The incidental meanness you feel is, at worst, a mild side effect of the greater benefit.

      • social justice warlock says:

        I’ve spent plenty of time in places with ideological opponents – indeed, since I’m an extremist nutcase, this is true for most places, but this would be true if we left out liberals as well – without getting agitated. This extends to chatrooms with mostly NRx folks! It’s only here that I froth at the mouth, and that’s a mystery to me.

        • Multiheaded says:

          Yeah, I’m telling ya, one way or other it’s definitely an effect of the certain atmosphere Scott helps set. Scott, you’re wonderful, all of us leftists here love you but…

        • memeticengineer says:

          Maybe it’s because here, your idealogical opponents are more likely to have good arguments? Or at least ones that are interesting enough that they take more effort to refute?

          • Viliam Búr says:

            Also, the opponents are usually polite. Which should not matter logically, but it matters emotionally. It is easier to dismiss an argument when it was said impolitely.

            Also, comments are short, it is difficult to state a bullet-proof case in a short comment, so almost any comment can be countered somehow. So we always take some information from context. And the context here is that people who write comments are generally nice and reasonable.

        • Scott F says:

          Possibilities:

          – You see people you consider Dunbar’s number-real taking SSC seriously, and you don’t see this happening with the other ideological strongholds you read. (more strongly: you see them taking SSC seriously for non-tribal reasons)

          – This is the only true ideological battleground you fight on. Elsewhere you are playing the part of a warrior for your cause, knowing that your enemies are acting out their roles too, and that at the end of the play nobody will be dead. Here, you really are a warrior for your cause, because you’ve seen the corpses (metaphorically, people have actually changed their minds in response to arguments).

    • I, for one, would like you to continue posting but to moderate your tone a little; what you say is worth hearing. The name change doesn’t really help; maybe try posting as “cuddly SJ bunny”? 😉

      • Matthew says:

        Intuitively, this strikes me as backward. He should feel more at liberty to be nice in the substance of his comments if he’s already got the aggressiveness baked into the username. I’d expect more hostility from “cuddly SJ bunny”.

  35. Harald K says:

    Can we have no IQ in the open thread too? Oh wait, too late.

  36. Princess_Stargirl says:

    I don’t mean to make myself feel bad but I currently feel as though I am just ignoring the stated comment policy. I feel ok doing this as I am still following community norms and what I “maybe” think is the spirit of the policy (but how can I really tell the spirit?). But here is how I honestly parse the comment policy:

    True: I really have no way of knowing if what I say is true or not. Most of what I talk about on here isn’t pure math. Even if I post about something concrete like the speed of various sorting algorithims I can’t be sure that new work hasn’t come out recently that changes the picture. Stories from my own life could easily be partially (or entirely false) as my memories are unreliable. I am not sure what standar of truth is implied by the policy. Unlikely anything I post will pass this gate.

    Nescessary: I cannot imagine much of what I post on here is possibly nescessary. Many blogs similar to your own are sucessful with no comments. And they have less insightful posts than most of what Scott posts. So I assume that none of my comments are likely to be needed. Never passing this gate as I parse it.

    Kind: Seems easy to pass this gate.

    • Paul Goodman says:

      I think you can get credit for true as long as you aren’t actively lying or bullshitting.

    • Rob Miles says:

      I think you’re being too harsh. Note that comments don’t require pre-approval, they are just subject to removal. Thus a comment doesn’t have to be absolutely *True*, so much as it has to be not-so-false-that-it-is-a-problem, and it doesn’t have to be absolutely *Necessary* so much as it has to be not-so-unnecessary-that-it-is-a-problem, and so on. The rules are stated strictly, to reduce quibbling, but they are enforced loosely, and with discretion.

      I’d say you’re fine as long as you make a genuine effort towards truth, necessity, and kindness, and if you decide to to drop one, you spend extra effort on the other two.

    • lmm says:

      I notice that I’m sailing closer to the wind as I watch the comment policy go unenforced. My posts are becoming meaner and more gratuitous, as that starts to look like the way to win arguments. I don’t want this to happen, but it’s happening all the same.

    • Jack says:

      I find being kind surprisingly difficult, especially when I’m talking to someone who seems to be talking bollocks: it’s easy to lambast them, and sometimes necessary, without asking _why_ they’re saying that.

      And yeah, I’m not sure. I think most comments here are _ok_, but I think the stated comments policy sets a high bar to aspire to, that most of the time most of us don’t really meet. Which may be doing a good enough job, I think if your comments are better than most, there’s no need to worry on your own account, but I don’t know if the community as a whole could be doing better.

      I wonder if, if one could have check boxes for “which two of T/N/K do you think your comment is” would make people think before posting. It would be a really interesting experiment. But as things like stack overflow show, it’s possible but very difficult to create civilised discourse, it may just prevent people from commenting at all, or just give something else for people to criticise each other for.

      • Rob Miles says:

        I wonder what the effect would be of a Slashdottish peer moderation system with True/Necessary/Kind in place of Insightful/Funny etc. (Note I’m not actually proposing this)

      • I don’t know if I’m reliably kind, but I’m pretty reliably not nasty, and the underlying motivation is a gut-level belief that people get stupider and meaner when they’ve been insulted.

        • Ialdabaoth says:

          Sometimes you WANT your opponent to get stupider and meaner, though. Wounded Gazelle Gambit is an amazingly effective strategy.

    • Scott Alexander says:

      You are way overestimating the bar. The policy is there to get rid of incredibly stupid and mean comments. Fewer than 1% of the comments here fail to pass the bar.

      By the way, this is going to be a personal question and I’m sorry, but I’ve got to ask – are you really a girl? I somehow developed a heuristic that anyone online whose name signals femaleness too hard is being ironic, and I want to see if it’s true.

  37. @JohnWBH says:

    A voting system might discourage bad comments and sort the better ones to the top,

    • Jaskologist says:

      I’ve never seen a voting system that didn’t primarily enforce ideological conformity.

    • Shmi Nux says:

      Echoing those who say that the good comments drown in the sea of marginal one, maybe there is a way to at least have a flag “Scott finds this comment interesting”? This seems mild enough to not be game-able, while drawing attention to the comments which OP likes but does not reply to for one reason or another.

      • Anonymous says:

        Seconding this idea.

      • suntzuanime says:

        Letting only one person vote on comments definitely sounds like a better idea than letting two or more people vote on comments, but I’m not sure it clears the high bar set by letting zero or fewer people vote on comments.

      • 27chaos says:

        Not sure if I like it better or worse than the current system, but I hate the idea of putting upvotes in these comments. Definitely endorsed if a compromise is demanded.

      • Multiheaded says:

        Problem: people might just grace-whore with very good links, and this will turn the comment section into a race to dump one’s best loosely related links. Solution: one Scott-tag for good links and another for good original comments?

        Source: this is how I would grace-whore.

      • Roxolan says:

        On posts with over a hundred comments, I find myself already doing this; Ctrl+F to find posts Scott replied to, and skipping basically everything else.

      • Emile says:

        I’m also in favour of “one man, one vote” as long as the man is Scott.

      • “Scott finds this interesting” is probably a bit too fraught; I would worry about “but is it fair to skip over this boring thread, someone might have posted something good” or “I really don’t want to (en-/dis)courage this poster, but it’s really (not) very good”.

        Note that you can “solve” the problem by just replying with a pat string (or username) to any comments you really like, e.g. “Shmi Nux thinks this comment is one of the best of this thread.” (Of course, you’d really want someone else to do that work, which gets us back to square one…)

    • Anonymous says:

      I am against any sort of voting system unless it also forces anonymity. V

    • Princess_Stargirl says:

      Voting also will push dissenting views to the bottom. In “less rational” community this effect often means that even the best “non conforming” comments will have low score. But the affect is strong even on less wrong. There are explicit examples of people (Allicorn) posting almost the exact same content as someone else but getting much lower scores because they were taking a position counter to most of the participants. On LW popularity doe snot dominate content but the effect is there.

      • Joe from London says:

        Could you link to an example? I am curious to see how this played out.

      • RCF says:

        I’m not sure whether I understand what you’re saying. You mean that someone says X, and also expresses agreement with the orthodoxy, and Alicorn also says X but adds a disagreement with the orthodoxy?

    • Multiheaded says:

      Plugging my grand proposal from the previous open thread:

      The way I see it is like this: before voting, everyone has to register and pick two or three, no more, tribal components for their identity (and migration is limited in some clever way) out of some list hashed out in a grand big thread. For example, “LessWrong + Islam”, or “Transhumanism + Rad-Fem”, or “Islam + Deep Green + Neo-Confucian Technocracy”. And then everyone, no matter their thede, can tag their posts as open for upvotes from any combination of tribes – and people who are registered as part of those tribes (can’t have bots, must limit alts) can contribute to those specific karma meters. And in addition, there’s the outgroup karma, which can be given to any posts not tagged with any of your tribes, to show appreciation for a different way while signalling something different from in-group approval.

      So you can see how this is intended to turn clustering from a flaw into a mechanism. This probably has all sorts of horrible bugs and exploits. But you seem to think about such things a lot. Any comments?

      Nydwracu’s reaction:

      Moldbug proposed a wiki along those lines. (But maybe that’s what you’re referring to…?)

      That might be worthwhile for some (much larger) sites, but probably not here — it would be best to keep divisions in the background, to avoid the breakdown of the overarching identity that fosters civility and inter-group exchange of ideas and so on here.

      Reifying divisions reifies divisions, and thedes like the currently-nameless-and-only-approximated-by-labels-like-’rationalist’ LW/SSC-sphere are rare enough that it would be best not to fuel their fracturing.

      • Nornagest says:

        I still think one of the most valuable features of LW‘s implementation of voting is that it discourages circlejerks: downvotes are aversive enough that people generally have an incentive to avoid them, and it’s an ideologically diverse enough forum that content-free cheering for one side will generally get you downvoted by someone (exceptions: atheism, anti-deathism, a few other rationalist shibboleths), so everyone has an incentive to add content and avoid being rude. The norms reinforcing this are, unfortunately, weakening; but they’re still strong enough that I think the basic point holds.

        Your proposal exposes interesting information about how members of various tribes respond to a post, but that comes at the cost of completely destroying any resistance to circlejerkery. Actually, I predict that a number of people would use that setup to optimize for negative reception from some quarters, because hey, fuck those guys, right?

        I consider this harmful.

      • Douglas Knight says:

        That is too complicated to get tacked onto a blog.

        It would be interesting to do cluster anaylsis on the voting data from, say, LW to see if you can find tribes. I think that is something to try long before building a site enforcing tribes.

    • Harald K says:

      It does not. The problem with internet voting systems is that something like 5% of the voters make 50% of the votes – and that’s not even counting voting from sockpuppet accounts.

      This is not a new observation. In the early days of online social media, they experimented with systems to make distributed moderation/voting more egalitarian, and to get rid of sockpuppets. Slashdot accidentally reinvented sortition as a means to make moderation work, and Advogato came up with its “attack resistant trust metric” to effectively resist sockpuppets. These systems really worked, although the ambitions for them were low (Advogato only set itself as a goal to identify open source contributors, slashdot only used moderation in comments and made very little use of the scores for sorting etc.)

      Sites like reddit knew that. They threw it out. A low threshold for participation is far more important to an angel-funded social media startup, than lack of sockpuppets (in fact, the reddit founders used sockpuppets themselves in the beginning to make the site appear more populated than it was). And the catharsis of an upvote/downvote button (“Yes!” or “take that downvote, jerk!”) was far more important to attract and retain people than quality of discussion.

  38. James Miller says:

    I know this is probably beyond your technical ability, but the best way to fight spam would be with micropayments. It would be great if you charged, say, a penny per comment. Might there be a way to do this with a cyber currency?

    • Chris says:

      Perhaps some variation on hashcash captures what you’re thinking of here? (TL;DR: require a Bitcoin-esque proof-of-work per comment.)

      Glancing at the source code, it’s not clear that the plugin actually implements a “hard” proof-of-work problem, but it seems to have some level of anecdotal support — presumably it’s not widely adopted enough for spammers to seriously look into breaking it.

      • This problem is harder than it seems. Javascript is not a good language for implementing fast cryptographic primitives [1]. As a consequence, if your commenters are willing to wait e.g. 1 second before a comment is actually posted, a dedicated spammer – who’s not using Javascript in a browser, but e.g. native code – will be slowed down by ~.001s [2] (this is how you defeat wp-hashcash, by the way).

        There is also the economic argument. When this was proposed for e-mail, some smart people eventually raised concerns that legitimate users of the system can have systems that are so slow that any system that actually stops spam also stops these senders from working normally. (I.e. grandma has a fifteen-year old computer and spends most of her day sending mail; a spammer has a big botnet.)

        Finally, hashcash was proposed at a time when legitimate users’ CPU cycles were essentially free. This is not true on modern PCs, and very much not true on a smartphone – do you really want to be “that website that drains your battery”?

        (I’ve looked into this area a bit, but I just don’t see a way to make proof-of-work work well on smartphones.)

        [1] The new WebCrypto standard may provide fast cryptography for newer browsers, but your system is only as strong as the fallback you provide for browsers that don’t support it.
        [2] Exact value for illustrative purposes only; I don’t have my notes at hand. But pure-Javascript crypto is really a lot slower than native crypto.

        • Douglas Knight says:

          The fallback is probably a Captcha, which may well be more secure, just more of a hassle to legitimate users.

          • Good point; I should have clarified that CAPTCHAs really are not a solution, either. E.g. this paper finds a cost of <$1 per thousand CAPTCHAs. Note that computers are getting really good at solving (increasingly bizarre) CAPTCHAs, too…

          • Douglas Knight says:

            A tenth of a cent sounds to me more expensive than a second of CPU time, not even taking into account that CPU time is much easier to steal.

          • Douglas: yes, you’re right. Note, however, that CAPTCHAs are getting less and less effective as computer vision algorithms get better and better and that CAPTCHAs are much more annoying than one second of CPU time. I looked into finding something better than CAPTCHAs. 😉

    • rsaarelm says:

      Per-transaction payments are a pain, especially with the present web that doesn’t really have either the culture or the seamless technical support for them. What does work on some big sites is requiring a one-time fee to create an account.

      It’s still probably a nonstarter on a site this small, but requiring registered accounts for commenting and making account creation cost, say, 50 cents might be actually technically doable (online shops for one-off transactions are probably pretty turnkey compared to micropayment architectures), and might also be culturally more palatable.

    • Meekras says:

      At last, you’ve found a way to support the economies of collaborative geofictional internet playtime!

    • Echoing Scott, NO BITCOIN THAT NEVER HELPS.

      (You really don’t want to wait 10 minutes before posting a comment, for instance, and you probably don’t want to turn a fairly standard WordPress website into some monster that requires an extensive array of supporting software.)

  39. Alexander Stanislaw says:

    2 Captcha is the definitive way of eliminating spam. It is a bit of a hassle for commenters who aren’t spammers. Any commercial commenting system will include it though. But this forces people to have to register before commenting, which could reduce the number of commenters.

    4 The fact that I missed the halting problem comment makes me much more sympathetic to an upvote system of some sort (I do really dislike the effect that has on the social structure though).

    • peterdjones says:

      Can I vote against? CAPTCHA=PITA, IMO.

    • See also how Making Light handles spam– in addition to the usual filtering, they have a link for the most recent thousand comments. Not only does this make it easier to keep up with the long, unthreaded comment threads, it’s easy to crowd source spam elimination.

      It would take people helping Scott to do administration, but I’m sure there are competent volunteers available.

    • no one special says:

      In theory, you could captcha posting, making it a hoop on each post, but eliminating the account requirement. In practice, Scott is going to use a plugin, so the theory doesn’t matter.

    • Douglas Knight says:

      We should try an “I am not a spammer” checkbox. Or a 2+2=? Might not work, but worth trying.

    • a person says:

      Having a captcha would be 100 times more annoying than the occasional spam comments this blog gets.

    • RCF says:

      How difficult would it be to come up with a whitelist of posters?

      • Zorgon says:

        Whitelists amount to a barrier to initial entry, which means lower potential for new quality commenters.

        • RCF says:

          Yes, but given that SA is seriously considering hiding the comment thread so that only people familiar with the blog can find it, he’s clearly willing to consider barriers to entry, and arguably this is a barrier to entry more highly correlated with quality.

          • Matthew says:

            How would someone new provide evidence of quality?

          • RCF says:

            This is an alternative to hiding some comments threads: allow only people on the whitelist to post in the threads that are expected to be problematic, and let everyone post in the other threads (and be whitelisted if they have good posts).

          • Ialdabaoth says:

            I’d be down with that, but I’d be skeptical of my ability to make the whitelist.

  40. Jaskologist says:

    3. Comments were already closed on the Nice Guy post by the time Instapundit linked to it, so you can’t blame him. Not sure the same can be said of reddit, though.

    • Ialdabaoth says:

      Also, if they are unable to comment here, they will just snipe at us on facebook and reddit. This has been happening all weekend regarding the Great SJ Threadwar of 2014, for example.

      • Kaminiwa says:

        As a reader, I have nearly zero chance of running in to sniping that is done on Facebook/Reddit, and thus don’t care about it. Conversely, I often read the comments, although the exact probability varies based on how likely sniping seems from the last couple times I checked them.

        So… I’m pretty okay with this tradeoff 🙂

  41. Andy B says:

    Pursuant to yesterday’s conversation regarding a peak in the utility of IQ: upon reflection, it seems plausible that the answer has actually changed in the past two decades or so, because there are at least two major new avenues by which people who are very smart but don’t play well with others can achieve extreme wealth and success: software (especially via startups) and investment banking.

    • Erebus says:

      The highest echelons of investment banking and the ranks of wealthy software developers are chock-full ‘o people who are essentially salesmen. You need to ‘play well with others’ to achieve success at that level.

      …And this is especially true in banking. Investment banking is a group activity — and if you don’t play politics and don’t play well with others, you’re not even going to have an opportunity to do well. All of the entry-level positions are mind-numbingly menial, and you’ll be chewed-up and spit-out within a couple years, before you’re granted an opportunity to make serious money. There may be a handful of reclusive and misanthropic investment bankers who have done well for themselves, but I believe that they’re an exception to the rule.

      There are better opportunities for ultra-high-IQ misanthropes in software. The world is badly in need of more and better data-encryption solutions.

      • Jaskologist says:

        “People skills” will always beat out raw IQ in the end, since people skills enable you to do whatever high-IQ people can do simply by having (many of) them do it for you. But that doesn’t mean that intelligence won’t get you further now than it used to.

        • 27chaos says:

          Without high IQ, you won’t know what goals you should ask others to pursue or which people you should befriend. Strategy is necessary, so IQ is too.

      • Andy B says:

        Nothing you wrote is false but my priors are unmoved: we are talking past each other.

        The conversation concerned (roughly) people in the 130-160 IQ range, not the off-the-scale outliers. Scott mentioned the hypothesis that people can only comfortably interact with people whose IQs are within 30 points of their own. By “plays well with others” I meant something less like “misanthropic” and more like “can handle dealing with mundanes”. The question was whether there are more/easier ways than in the past for such people to attain high welfare and/or big bucks.

        I maintain that the software industry provides unparalleled opportunities for people who can’t comfortably work within the system and/or can’t comfortably interact with mundanes. One or two coders can produce a prototype with roughly zero outside investment (just a computer with an internet connection). A company requires few employees and no specialized facilities. The proliferation of small firms means that coders don’t have to impress HR in order to be hired. Etc, etc, etc. Many small factors add up to a sea change. Exhibit A: Bill Gates.

        The traditional aspect of investment banking is as you say, and that aspect persists. Quantitative analysis, however, is more or less new and does not involve interaction with mundanes. Exhibit A: my friend the quant.

        I hope that clarifies my argument.

        • Princess_Stargirl says:

          Are there any fields where 160 IQ is actually commonish? 160 is 1 in 31,560 (15 Sd).

          Also my IQ consistently tests at approx 146 (lower than predicted by my SATs sadly). I am certainly comfortable interacting with people way below 116 IQ. I get along perfectly well with people above around 85 IQ imo. Maybe I just have “low bro tastes.” I enjoy watching reality Tv, talking about clothes, gossipping, disney movies etc. Seems like this means me + any reasonaable girl have stuff to chat about!

          I don’t actually if I have ever seriously interacted with anyone above 160 IQ. It seems plasuible to me I have not. But people around 160 IQ seem to get along with me and not find me too dull to work with on intelectual topics. (my best friend is approx 160 Iq). People at around 170 and up are 1 in 500,000+ and so very rare to work with.

        • BJ Terry says:

          A quant isn’t called an “investment banker” which may be why your initial post was confusing (I was an investment banker). A quant can work at an investment bank, but would be called a quantitative analyst, quantitative trader, or something. Traditional investment banking requires “interacting with the mundanes” as you put it even at the junior level, and at the senior level an excellent investment banker is mostly judged based on their ability to generate deal flow, i.e. sales.

        • RCF says:

          “My friend the quant”, without any further identifying details, isn’t much of an exhibit. And Bill Gates is an exhibit for what? The lack of a need for impressing corporate people? Microsoft started out being a part of MITS, and after that it was highly dependent on IBM.

      • Quixote says:

        This is a commonly said. What’s your evidence for it? You would expect people in software to say it regardless of it’s truth value.

        • Erebus says:

          There are still no good alternatives to TrueCrypt. When the developers pulled the plug a few months ago, they left a lot of people desperate for alternatives, and that need has not yet been addressed. Not satisfactorily, anyhow.

          There are also still no good end-to-end encryption solutions for smartphone messaging. And end-to-end encrypted simulacrum of “What’s App” would make you a millionaire overnight, even if you’re an utterly misanthropic recluse with zero people skills. The damn thing would sell itself.

          • Douglas Knight says:

            Maybe there is a tiny market that would pay for encryption.

            How are you going to sell a clone of truecrypt? If you make it open source, it will be pirated. If you don’t, why trust it over bitlocker?

            Access to a centralized server is a way to force people to pay for encrypted messaging. But this is a winner-take-all social network. It is a hassle to message different people different ways. How popular do you expect google and yahoo’s browser plugins for end-to-end encrypted email to be? In fact, there is an end-to-end encrypted chat app built into Apple’s SMS app. iphones have tremendous privacy advantages over everything else and no one cares. Don’t you trust Apple to know what will or will not sell? That they choose not to talk about this is pretty damning.

          • Erebus says:

            I don’t believe that the Apple apps are truly end-to-end. In fact, I don’t believe that any current mobile solution is truly end-to-end. Though I know that people _are_ working on the problem. See: https://dev.guardianproject.info/projects/gibberbot/wiki/Always_Secure_Messaging?title=Always_Secure_Messaging

            BitLocker is Microsoft’s locked-down and single-platform encryption solution. It’s not an ideal replacement for something like TrueCrypt, I’m afraid, as TrueCrypt was cross-platform, ostensibly open and more secure, etc…

          • Douglas Knight says:

            Apple may not be a True Scotsman, but it is far more secure than anything else out there AND NO ONE CARES. Why do you think anyone will care when you get a True Scotsman?

          • RCF says:

            AFAICT, the issue is entirely one of coordination. Writing an end-to-end encryption app would be absurdly simple, as long as everyone is using your app. The issue is that not everyone’s going to be using your app, so it’s going to have to interface with other apps, and that’s where it gets complicated.

          • Douglas Knight says:

            RCF, WhatsApp failed repeated, so I wouldn’t call it trivial. Moreover, beyond the technical aspects there are huge user interface problems. To get full security, you need to verify the public key across another channel. Typical users are not going to do this and are going to be scared by the existence of the option. Instead Apple runs a directory server and you have to trust them that they are not running a man in the middle attack. They should at least record the public key locally so that you can detect if they start running such an attack on you after you sign up. But there are a lot of details of what to do if the user gets a new device and it is difficult to avoid false positive alerts.

          • For what it’s worth, Google has released an interesting proposal for a key distribution protocol for the end-to-end encryption extension that they’re working on (and that you can already download the code for). It looks potentially promising, in terms of both security and usability by regular people. (Security is not my area of expertise within computing so take this with a grain of salt.)

          • Douglas Knight says:

            I trust google to get the technical details correct. Yes, when things work correctly, this doesn’t require input from typical users, but the problem is when things go wrong. How often are people going to change keys and will this generate false positive?

          • Key revocation follows the same principle as the existing, fully-working-and-secure-but-too-complicated-for-humans-who-aren’t-crypto-nerds OpenPGP system. If you lose your key, you make a new one and then tell the keyservers that your old key is no longer valid. Google’s new protocol contains a way for the keyservers to make sure anyone who wants to communicate with you finds out that you’ve got a new key. Naturally you won’t be able to decrypt anything that was encrypted to your old key.

            All in all it doesn’t seem likely to result in significantly more usability challenges than users currently face when dealing with this kind of situation. Not sure what you mean by false positives though.

          • RCF says:

            @Douglas Knight
            “RCF, WhatsApp failed repeated, so I wouldn’t call it trivial.”

            By “failed”, do you mean the program did not display the intended functionality, or do you mean people did not adopt it?

            “Moreover, beyond the technical aspects there are huge user interface problems. To get full security, you need to verify the public key across another channel.”

            How are you sending SMS messages in the first place? You must have exchanged phone numbers somehow, and obviously that involved some channel other than SMS. So why not exchange public keys over that channel as well? Let’s say you and I meet on the street. You open the SecureSMS app on your phone, then hand me your phone. I open the SecureSMS app on my phone. I type my phone number in SSMS!YourPhone, press enter. There’s a cryptographic handshake. If I’m paranoid about someone at the phone company (or someone who’s hacked/spoofed the phone company) doing MITM, there are various verification methods:

            1. I have SSMS!YourPhone display my public key. But let’s assume I don’t have my public key memorized, so we need another method.

            2. I type a test phrase into SSMS!MyPhone, encrypt it, send it to SSMS!YourPhone, and SSM!YourPhone displays the plaintext.

            3. Both SSMS!MyPhone and SSMS!YourPhone display my public key, and I check to see if they’re the same.

            4. Both SSMS!MyPhone and SSMS!YourPhone flash the characters of my public key one by one, synchronized, and I check whether they’re the same.

            5. My public key is represented by an abstract design, and I check they’re the same.

            etc.

            With option 4, it could easily be 6 bit/sec, so 200 bit key could be checked in just over 30 sec. Yeah, that’s a slight inconvenience, but it’s less time than “You pull out your phone, navigate to your address book, have me slowly give you my phone number one digit at a time while you type it in, and then spell out my name so you can complete the entry”, which is how a lot of people exchange phone numbers.

          • Douglas Knight says:

            RCF, WhatsApp had lousy crypto, repeatedly.

            Taymon, if you lose your key and revoke it by demonstrating control of your email address but not control over your old key, that looks just like someone else has hijacked your account. Either you alert other users of a potential compromise (probably a false positive), or you have to trust your email provider not to turn over control of your account to the government to impersonate you in future messages. But this does make such a hijacking a one-way affair and makes deniability difficult.

          • From the proposal:

            The model envisioned in this document still relies on users being able to keep their account secure (against phishing for example) and their devices secure (against malware for example), and simply provides an easy-to-use key discovery mechanism on top of the user’s existing account.

            So account hijacking is out of scope for this proposal; anyone who has control of your email account is assumed to be you, and if that person revokes a key then this is assumed to be because your key was lost or compromised rather than because your email account was compromised.

            You can’t prevent your email provider from issuing an unauthorized revocation of your old key and replacing it with a new one which they turn over to the government. You can, however, find out if they have done this. Hopefully people working on the UI end of this will include a way to make it obvious to users if this has happened, since most won’t bother to look it up. But it is a risk.

          • Douglas Knight says:

            That’s a reasonable choice, to sacrifice real-time security for convenience, but still have auditing. And it’s definitely an improvement over Apple. But it’s only a bit better and a consistent Erebus would call it fake end-to-end encryption.

          • RCF says:

            “RCF, WhatsApp had lousy crypto, repeatedly.”

            From my cursory googling of WhatsApp, it appears that this was due to incompetence (if not mendacity) on the part of the people behind it, rather than any inherent difficulty in the task. Do you wish to argue otherwise?

          • Secure messaging has been solved pretty well; just use Moxie Marlinspike’s TextSecure. The most convenient way to get someone’s TextSecure key is just to scan a barcode from the screen of his/her/their smartphone.

            The crypto used to implement TextSecure is somewhat more involved than my fellow commenters seem to give it credit for: notably, TextSecure incorporates Trevor Perrin’s new Axolotl cryptographic ratchet to implement forward security. (Forward security ensures that e.g. someone who steals your friend’s phone can’t recover all messages you’ve ever sent to your friend.)

            (Disclaimer: I am a professional, but this recommendation is based on third-party recommendations and Moxie and Trevor Perrin’s other work/reputation.)

      • Joe from London says:

        Do you count trading as a subset of investment banking?

        • Anonymous says:

          “Trading” covers a range of activities that range from sales to trading with the impersonal market. One end is IB; the other is not. Trading with the impersonal market requires smarts and not social skills, but getting such a job requires more social skills than getting a quant job. It is a job with easily evaluated performance and thus unlimited salary, while quant jobs are usually structured more on a salary basis and may require more social skill for negotiating salary.

    • no one special says:

      Software startups are essentially investment banking. That is, they’re so strongly tied to venture capital that the “go huge”/acquihire model means working for a startup is like buying a lottery ticket. I’d say that the last thing people who don’t play well with others want is to be looking for a job again in a year because Facebook decided to buy your relationships by paying off the VCs who funded the startup you were working for.

      Non-startup software often requires deeply probing non-technical people to figure out what they actually want so that you can build it. I suppose there are a few programmers who can just go heads-down on some technical quest, but I’m not one of them, and you’re probably not either. Most of us will be building forms-over-data intranet or brochureware sites, which will require close work with the squishy humans.

    • covaithe says:

      Anecdotally, there is a failure pattern for overly smart software developers that I have observed a handful of times in my career. It goes like this.

      Nearly every software developer has to learn the hard way that there are upper limits to the complexity of the code that they personally can understand and maintain, and that when you get close to your limits (or preferably well before you get anywhere close to them) you need to take steps to limit complexity. Things like extracting methods, reducing conditional nesting, etc; basically all the things people mean when they talk about refactoring or good style.

      We pretty much all have to learn this the hard way by bumping up against our limits. But really excessively smart programmers sometimes have more expansive limits. “Lesser programmers,” they argue, “may not be able to understand my work, but *I* understand, so what’s the problem?” Then, when they inevitably do manage to write code more complex than they can understand, they’re totally screwed. They have dug themselves a far deeper hole than their less intelligent comrades could have managed, and it’s correspondingly harder to recover. And of course they can’t ask for help, since they’ve gone around being arrogant about their code for so long that it would be hard to ask for help, and even if they did nobody could understand it anyway.

      Most programmers hit a wall like this at some point in their career, learn a little humility, clean up their mess, and start to work on improving their complexity management skills. But I’ve seen people burn out on programming because they were so smart that by the time they hit their wall, they had made too big a mess to recover from and ended up just walking away.

      I wouldn’t exactly say that therefore overly high IQ is harmful to programming ability, but I do think that intelligence needs to be tempered with some knowledge of one’s limitations to be successful at software.