SSC DISCORD SERVER AT https://discordapp.com/invite/gpaTCxh ; SCHELLING POINT FOR DISCUSSION IS WED 10 PM EST

Vote On Values, Outsource Beliefs

I.

Today I learned about social impact bonds. They are a thing that exists. I would expect them to be in an adequate civilization like Raikoth or dath ilan. But they are a thing that exists on Earth.

The basic idea is: government could save a lot of money if some problem got fixed. For example, if people stopped committing crime, they could spend less money on prisons. So they make a deal with a corporation. The corporation agrees to spend a certain amount of money to prevent crime for five years. And if crime goes down and the government saves on prisons, the corporation gets half the savings (or a third, or whatever).

Zero taxpayer money gets risked. It is entirely up to the corporation to fund the problem-solving effort. If they fail, then it’s their own loss. If they succeed, then the government pays them money, but less than the government made, so the taxpayers still get a profit.

(The main exception I can think of is if by coincidence, crime was about to drop by 50% anyway right when the program started, and the government ends up giving half of its prison savings to the corporation for no reason. But presumably you hire a couple of mediocre economists and they are able to price out this risk. Also, a lot of the social impact bonds use a slightly different method of assessment, where they compare crime among the people the corporation has helped to crime among a control population to be sure it was the intervention that did it.)

The particular article I read about this today was How Goldman Sachs Can Get Paid To Keep People Out Of Jail. It was the name “Goldman Sachs” that got me excited. They’re an investment bank. Their job is predicting risk. I don’t know if they’re any good at it or not. But they’re the sort of organization that potentially could be. So we have people who understand risk trying to figure out what social policies will produce which results, with money riding on the decision.

This is looking impressively close to prediction markets. Futarchy says “vote on values, bet on beliefs”. Asking a corporation to invest money in crime-solving is a form of betting on belief – they are betting on what anti-crime programs will decrease crime most and win them the most reward. You still have the elected government deciding what bonds to place – voting on values – but you’re outsourcing your beliefs to the corporation involved and giving them an incentive to get it right.

Think of all the possibilities.

Right now we have a system where we don’t really help people in need, unless the need becomes desperate, in which case we would feel bad about not helping, so we do, but then the cost of helping has gone up by an order of magnitude. This is exactly the sort of stupid thing that a market should be able to profit from solving.

We could have a health insurance company giving free preventative care to the poor, and the government paying them out of decreased emergency room visits.

A psychiatry clinic giving therapy to at-risk patients, and the government paying them out of decreased involuntary commitments.

A university accepting students without tuition, and the government paying them out of the increased tax revenue when they take higher-paying jobs.

Planned Parenthood offering free IUDs for women who need them, and the government paying them the money it saves from not having to put the kids through school.

Trade schools offering free classes to people on welfare, and the government paying them back from not having to give them welfare checks once they get good jobs.

I’m not sure what it means that we’re not doing those sorts of things already. But if we can’t figure out a way to solve those problems without bringing in a corporation to profit off of our incompetence, I say bring in the corporations.

II.

I think many people are against government social programs for a lot of the same reason that The Last Psychiatrist is against maintenance of certification exams (a position I totally called). There’s too much temptation to use it as a signal that you are Doing Something while in fact funding programs like DARE which look virtuous, but do nothing or even actively make the problem worse.

If you lean this way – and I think I do – then it is not solely out of stupidity that we wait until problems have become dire before doing anything about them. Yes, it would be great to give free job training to people on welfare and save money when they come off welfare more quickly. But actual job training programs for welfare recipients are abysmal and have been denounced as a “charade” from both the left and the right. They may be a lost cause, but I would like to see someone who has an incentive to succeed try first before writing them off – or at least get the evidence that would be provided by no such person being willing to try.

III.

For a while I was confused by the old libertarian talking point that “greed is good”. I think I could phrase it a little better now. Greed isn’t good, per se. It is honest. You know where you stand with greed. You never wonder if greed has an ulterior motive, because it’s already the most ulterior motive there is. Greed feels no temptation to corruption, because the thing it would do if it were corrupt is precisely what it’s doing anyway. Greed is like the Harlot in one of Khayyam’s rubaiyat:

A Sheikh beheld a Harlot, and said he:
“You seem a slave to drink and lechery”
Replied the Harlot: “What I seem … I am!
Are you, O Sheikh, all that you seem to be?”

As I see it, capitalism isn’t about worshipping greed, but about figuring out how to make greed work for good ends. So far, it has mostly tried to apply greed to get us cheap and attractive consumer products. And the amount of cheap and attractive consumer products is, like, the one thing that everyone can unambiguously agree our civilization hasn’t dropped the ball on. If we all die tomorrow and aliens discover Earth ten thousand years from now, their anthropologists will publish books saying “They sure were screwed up, but man did they have a lot of cheap and attractive consumer products.”

And I think some of the most exciting proposals for the future involve finding ways to use this privileged incorruptible perfectly-incentivized status of greed to do other things. Prediction markets are promising because they use greed to fix epistemology. Neocameralism is promising because it uses greed to fix governance. And social impact bonds are promising because they use greed to fix social problems.

…which isn’t to say it’s going to be easy. Ozy’s first response is that Goldman Sachs should use their $10 million to give ten thousand people in the control group a $1000 bribe each to commit a small crime; this will be more than enough to demonstrate a vastly reduced probability of criminality by being in the intervention group and earn Goldman $20 million.

I told Ozy zir plan is unnecessarily complex. Look at the numbers. Two hundred potential criminals. And they need a 50% decrease in jail time to meet their target and earn $20 million.

So go to the potential criminals and tell them “I’ll give you $50,000 to not commit any crimes in the next few years. $25,000 now, in order to help you solve whatever problems turned you to criminality. And $25,000 at the end, after you’ve successfully avoided jail, as a reward.” If half of them stick to it, then boom, you get $20 million and you’ve made a $10 million profit. And incentivized the next generation of criminals, but you’ve already got your profit, that’s the next generation’s problem.

The fact that this would work probably says a lot about the inefficiency of prison compared to any other conceivable way of dealing with crime. And about the profits Goldman Sachs or anyone else willing to face the inefficiency head on could make.

I don’t know if it’s exactly a good idea to bring in the people who caused the financial crash to help the people who came up with the prison system. But since all we’ve got is incompetent institutions, maybe sticking different incompetent institutions in different roles might at least shake things up a little.

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

125 Responses to Vote On Values, Outsource Beliefs

  1. Andrew says:

    If you’re interested in things that work better than the prison system, Megan McArdle wrote about this system in Hawaii that seems like it’s stood up to randomized controlled trials.

    http://hopehawaii.net/

    • Scott Alexander says:

      “If you’re interested in things that work better than the prison system” sounds like a Fully General Introduction.

      • Deiseach says:

        There is a private company in the United Kingdom that provides these kinds of services for the government; I know its name mainly from regular references to it in “Private Eye” on grounds of incompetence, over-charging, inefficiency, and good old-fashioned fraud and corruption.

        The main problem seems to be: if the government gets out of, say, building and running prisons, then it’s up to the private outfit to do so. So who oversees them? If thirty prisoners a day (from all their prisons and detentions centres overall) are committing suicide, this certainly reduces potential crime and under the terms of the contract with the government they are owed the promised recompense, but is that the kind of service that’s good in the long run for society?

        Also, given that they’re subsidised with public money (otherwise they wouldn’t enter into agreement with the government), they can’t really fail: what they get from the government contract is their basic operating costs and if, on top of that as a private business, they make a profit by cutting to the bone staffing levels in their care homes and retirement communities, then that’s gravy for them and who cares if Granny is left sitting in her own urine?

        It emerged in November 2013, Serco, which won a contract for Suffolk Community Healthcare in 2012, had 72 vacancies after earlier cutting 137 posts. Problems identified by Ipswich and East Suffolk Clinical Commissioning Group include “staff capacity, skill mix, workload, succession planning and morale, training, communication, mobile working, care co-ordination centre processes, incidents and near miss incidents”.

        I’d love to believe free market capitalism can solve all ills, but as we were taught on my secretarial course back in the day by our business studies instructor: “The purpose of a business is to make profit. The business owner is not interested in giving you a job, he’s interested in can you make money for his company. They’re not a charity.” (This was in the context of an anecdote about an employer who fired one of his staff after he saw her outside of work, in wet weather, without her head covered. His attitude was “She’s not taking care of her health. If she gets sick, this will cost me money to replace her while she’s out ill. So I’ll fire her and hire someone who won’t cost me money”.)

        • Deiseach says:

          Oh, and I forgot: Atos, another out-sourced provider. UK Department of Work and Pensions has given a contract to them to carry out the assessments for people claiming disability payments, in order to whittle down the number of claimants who could take up paid employment and to reduce fraud.

          There have been stories of them holding assessments for wheelchair users in an upstairs office without lift access. You don’t turn up for your assessment (because you physically cannot get up the stairs)? You’re struck off claiming benefits!

          It all depends on what you want to achieve by these privatised systems: reduce public expenditure? That’s certainly possible. Make the best use of services and make sure those in real need are served? That’s a different question and you may not make savings (or, what is more important for stock market prices) profits on the contract if you are delivering a client-oriented service.

          And, as seen over here in Ireland and the U.K. anyway, there is much too cosy a relationship between these companies and governments where ex-civil servants get plum jobs as non-executive directors on the boards of these firms which then lobby the departments these new directors used to work at.

        • MugaSofer says:

          “The main problem seems to be: if the government gets out of, say, building and running prisons, then it’s up to the private outfit to do so. So who oversees them? If thirty prisoners a day (from all their prisons and detentions centres overall) are committing suicide, this certainly reduces potential crime and under the terms of the contract with the government they are owed the promised recompense, but is that the kind of service that’s good in the long run for society?”

          That’s a possibility, sure, but isn’t that why we have … laws?

          Also, that seems like an objection that applies to any activity not carried out by the government (and arguably every activity that is carried out by the government; see: libertarians.)

        • Scott Alexander says:

          I hope you’re not suggesting I approve of private prisons. I hate them with a violence.

          I believe corporations are very good at doing exactly what they’re incentivized to do and nothing else. If you pay corporations to make cheap prisons, they will do a great job making very cheap prisons – by making them unliveable and inhumane.

          (if you made this real capitalism – by allowing prisoners to choose which company’s prison they could serve their sentences in – the problem would disappear in a heartbeat. But some other people might have a problem with that)

          On the other hand, if you pay corporations to keep people out of jail, I expect people will effectively be kept out of jail. I’m not sure whether or not they will find some way to do this that involves terrible consequences to someone else, but I think it’s worth a shot.

        • Said Achmiz says:

          On the other hand, if you pay corporations to keep people out of jail, I expect people will effectively be kept out of jail. I’m not sure whether or not they will find some way to do this that involves terrible consequences to someone else, but I think it’s worth a shot.

          It’s only worth a shot if said shot is easily and quickly reversible, else this attitude looks pretty similar to “I’m not sure whether or not this not-proved-to-be-Friendly AI will find some way of optimizing for maximum human happiness that involves terrible consequences, but I think it’s worth a shot”.

        • James James says:

          “If thirty prisoners a day (from all their prisons and detentions centres overall) are committing suicide, this certainly reduces potential crime and under the terms of the contract with the government they are owed the promised recompense, but is that the kind of service that’s good in the long run for society?”

          Possibly.

  2. Daniel H says:

    The part about greed being honest sounds a lot like something you’ve touched on before: make the incentives align with the good results. This is the same idea as in the recent posts on fixing science: make the scientists want to get reproducible results, and want to actually do the replications.

  3. suntzuanime says:

    Greed isn’t good, per se. It is honest. You know where you stand with greed. You never wonder if greed has an ulterior motive, because it’s already the most ulterior motive there is. Greed feels no temptation to corruption, because the thing it would do if it were corrupt is precisely what it’s doing anyway.

    I don’t have any point to make, I just wanted to applaud.

    • Randy M says:

      Greed is also reliable, in that more people will be motivated by it more often than by altruism. At least if you define greed broadly enough to include self-interest.

      • Said Achmiz says:

        Greed and self-interest are only reliable if you assume that people aren’t ever irrational. Then it stops being reliable.

        • Randy M says:

          Is there a motivation that wouldn’t work in that sentence?
          Can you do me the favor of reading reliable there as relative to other motivations rather than 100% so, a reading consistent with the use of the phrase “more often”?

        • Said Achmiz says:

          “relative to other motivations” is exactly how I did read it.

          The difference between greed and, say, altruism, is that if you’re an altruism and your attempts at being altruistic turn out to have horrible consequences for people (not you, but other people), you might notice this, in which case you might be horrified and try to fix it (because you’re an altruist). If you’re greedy and your greed turns out to have horrible consequences for other people, and you notice this, so what? You’re motivated by greed, so those other people can go take a flying leap.

          So it’s not that I think greed is more vulnerable to distortions by irrationality. It may be, it may not be; I don’t know, though I have no good reason to believe so. The point, though, is that for greed to be a good alternative to altruism (or whatever else) as the engine of public good, it has to be substantially less vulnerable to such distortions, because it is not inherently self-correcting.

        • Greed is self-limiting in regards to damage to other people if those people or those allied to them have some way of shoving back.

        • Head Stomp says:

          The difference between greed and, say, altruism, is that if you’re greedy and your attempts at being greedy turn out to have horrible consequences for people (not you, but other people), you might notice this, in which case you might be horrified and try to fix it (because you’re concerned with losing market share). If you’re altruistic and your altruism turns out to have horrible consequences for other people, and you notice this, so what? You’re motivated by a morally righteous cause, so those other people can go take a flying leap in the name of the greater good.

        • Steve says:

          As a *motivator*, greed is more reliable than altruism. As a *constraint on people’s expected actions*, it’s less reliable than, say, the laws of physics.

  4. Anonymous says:

    The Health Impact Fund is a good example of this. The idea is to have billions of dollars to reward companies that make drugs to treat diseases in the 3rd world. There’s lots of diseases that cause a lot of suffering, but there’s no incentive to cure, because the people that get these diseases don’t have any money. The reward for developing the drug is supposed to be based on how many people it helps (measured in DALYs maybe? not sure of the specifics). It’s not exactly the same–there’s no taxpayer savings that are used to fund it, it has to be subsidized. But governments spend a lot of money on foreign aid anyway, might as well try to make it as helpful as possible.

  5. Neocameralism is promising because it uses greed to fix governance.

    Could you clarify what you mean by this? I’m not completely sure what the word “neocameralism” actually refers to, and I suspect different people use it to mean slightly different things—most of which bear too little resemblance to anything that could actually be implemented in real life to be describable as “promising”.

    • Scott Alexander says:

      My understanding of it is the setting up of small governments – maybe ruling areas no larger than a city or county – run by corporations. If you don’t like the way it’s run, you move to a competing city or county and they lose your tax money.

      In its purest form it’s impractical. In less pure forms, it’s things like charter cities that might work. I’m also interested in what will happen when life becomes mostly virtual, or when basic income/online work makes people much more mobile. Expect a post on this eventually.

      • Hainish says:

        I’d love to see this idea applied to k-12 education in the form of (geographically overlapping) charter school networks. In this case, though, the goal would be to eliminate the strict dependence of school access on geographic location (in that it would be possible to change networks without having to move).

      • peterdjones says:

        More Detroits!

    • Leonard says:

      Neocameralism is government by a modern sovereign joint-stock corporation. That “modern” points to technologies that are thought to be necessary to cut off certain failure modes that are otherwise problematic. In particular it relies on cryptographically controlled weapons locks to prevent coup by the security forces.

      Contrary to our esteemed host, neocameralism as such does not have to do with the size of government. Its adherents do tend to emphasize small states to make exit work for the system. This is a sort of back up failsafe, though, compared to the one which Mr. Alexander is getting at in this post. And in any case, the exit advantage of many small states applies to any system of government, not just neocameralism. It’s always easier to flee a state when the border is close.

      The subtle thing about neocameralism is that it aligns the financial interest of the owner(s) with the good of the realm. (This is also true of monarchy; the improvement is that in monarchy the monarch may or may not be greedy.) This is the greed thing working for you. Greed thus solves one of the fundamental problems of democracy: that democratic government is, and must be, incoherent. By incoherent I mean there is no consensus (nor anything even close) on what the ends of democratic government are.

      • Doug S. says:

        (Incidentally, shareholders in corporations today tend to be notoriously ineffective at making sure that corporations are being run for the benefit of shareholders and not for the benefit of management.)

        • Leonard says:

          I dunno. It seems rather hard to measure. How can you determine that some management team is overpaid?

          In any case, greedy management does not seem like a problem for neocameralism. Management’s greed, although it conflicts with the greed of the owners, does not render the organization incoherent. Both owners and management will agree on the ends, namely: profit. As such, the incentive for good government is unaffected. It’s a problem for the owners, not the subjects. I agree that there might be a problem inasmuch as management attempts to use the state for their own ends other than greed. That is an example of incoherence.

          It is worthwhile contrasting with our current situation.

        • Doug S. says:

          It’s not only about management salaries. Corporate board members have an interest in not being replaced, so they tend to set rules which make it very difficult for shareholders to replace board members.

          See also.

  6. lambdaphage says:

    Forgive me if these questions have been addressed somewhere already, but are SIBs also vulnerable in their own way to some standard public choice concerns? For instance:

    1) The government promises Goldman Sachs $1M USD for every 5% decrease in the incarceration rate. The next morning, at-risk people start dying in accidents, or leaving town in the middle of night without telling anyone. The two are totally unrelated.

    2) AliceCo, BobCo and CarolCo are bonded to reduce the crime rate through addiction counseling, basketball leages and childcare, respectively. AliceCo would find it profitable to provide the addiction counseling at the going rate of return, but suspects that crime won’t decrease by the necessary amount unless BobCo and CarolCo pull their weight as well. BobCo and CarolCo know this, and know that AliceCo know that they know… and sooner or later you’re in the Corporations-Tasked-To-Reduce-Prisoners’ Dilemma, where everyone would be better off if they cooperated, but each faces an individual temptation to defect. Except perhaps worse than before, since public agencies, for all their faults, are at least nominally obligated to render services, whereas a bonded private player could just eat their losses and bow out?

    3) The police have read Michael Sandel and Gneezy and Rustichini too. Whereas they were once willing to take a bullet to uphold the generalized public order, they now see their job primarily as one of implementing the bright ideas of hot-shot Ivy League 22-year-olds in return for cash bonuses. The next time the dispatcher calls, the cops quietly bury their noses in their copies of What Money Can’t Buy: The Moral Limits of Markets and pretend not to hear it.

    I would like to see SIBs work, but I don’t quite understand how they avoid some of the same principal-agent pitfalls of public agencies. To take an example familiar from my own city, both the police commissioner and state’s attorney have strong incentives to keep crime down. Well, actually they have strong incentives to keep the crime stats down, because that’s the only way to sense how bad crime is in large cities. But the police commissioner has some discretion to influence over what reports get classified as what, and the state’s attorney has even more discretion to decide who gets charged with what. What Happens Next Will Shock You, I’m sure.

    That problem probably gets easier when the people collecting the stats are not the same people who get fired if the stats don’t look good. But it does make the stats collectors seem pretty vulnerable to corruption, which was the very vice that SIBs were supposed to transmute into virtue in the first place.

    Or maybe there are obvious answers to these questions I just haven’t thought of.

    • Steve says:

      Those were really good questions, and I’ve wondered about #2 in particular, in relation to prediction markets. Do you know of anyone writing about them?

      • lambdaphage says:

        I think most people who study them would encourage trade in outcomes conditional on other events. I.e. “I promise to pay the bearer of this ticket $1 if crime goes down 5% and BobCo provides a basketball league, or 3% if it doesn’t” or so.

    • Scott Alexander says:

      1. Killing people is kind of illegal. If Goldman Sachs is willing to go that deep into crime to make money, it would probably be easier for them to just rob banks.

      2. Luckily, contract law!

      3. That only seems like a problem once social impact bonds start replacing policing, at which point it’s not a worry, because social impact bonds would have replaced policing.

      • lambdaphage says:

        1. Clearly killing the people you’re supposed to help is about as bad as an unintended consequence can get. Maybe I should have picked a less lurid example– I’m not worried about Goldman Sachs literally offing people, but what if they just paid people to move away? Other commenters have offered other scenarios, some quite subtle, and they’re not even getting paid millions of dollars to think of subtle ways to fulfill the letter of the contract. When you’re paying for outcomes rather than directing efforts yourself, it’s always worth thinking about how you plan to measure those outcomes.

        In general you should be more willing to pay for outcomes when you know exactly what you want, but have no idea how to get it. In social policy, though, specifying outcomes non-perversely can be hard.

        Imagine the mayor who sits down to draft a contract for a SIB with fifty clauses like “no, you can’t just pay people to move away and no, you can’t just hook them on heavy sedatives and no, you definitely can’t exploit Simpson’s paradox by encouraging criminals to take up more serious crimes where the offense rate per unit time is naturally lower god what is wrong with you…” At some point, the complexity of specificying the outcome might exceed the complexity of just managing a public agency directly.

        2. Sure, but the contract might result in AliceCo terminating services because they think BobCo is in breach. The most unnerving feature of SIBs for me is the prospect of a firm realizing that it’s behind and cutting its losses, leaving the city with no provision of that service. Well, you might say, the city can start taking bids from other firms again in that case, and perhaps they would do well to pay a premium that ensures that the firm has an interest to stick around under all reasonable circumstances. But then we’re no longer offering a bond that only matures when the Good Thing happens– we’re just hiring contractors and probably wanting to supervise them now, too.

        I dunno, maybe I need to sit down and try to model this, but I wonder if there’s a potential problem in markets for social goods like crime control where you can’t just wait for the market to clear, and the illiquidity turns the SIB into something that no longer looks much like a bond anymore? Hopefully someone with economic training will tell me what I meant to say here.

        3. Maybe we should. I’m just observing that the generally sound principle of giving people correct incentives to solve social problems also requires us not to unwittingly muddle the good incentives we didn’t realize they had in the first place. That’s not a reason never to go in for an SIB, just a reason to do experiments and try to think through things carefully.

        Again, I’m not opposed to SIBs in principle at all. I’m just curious about how they avoid existing pitfalls, when they’re most appropriate, &c.

        • Actually murdering prisoners is a bit lurid, but neglecting the weaker ones to death is undramatic.

        • Paul Torek says:

          In general you should be more willing to pay for outcomes when you know exactly what you want, but have no idea how to get it. In social policy, though, specifying outcomes non-perversely can be hard.

          Exactly. We could make a little more headway here if our legal system weren’t so, um, legalistic about contracts. But for now, anything contracted out like this is much more likely to deliver the letter, rather than the spirit, of the specified goals.

          Relatedly, Robin Hanson’s second premise on futarchy says

          It is not that hard to tell rich happy nations from poor miserable ones.

          Which is crazy talk. Specifying happiness is ridiculously hard.

  7. Said Achmiz says:

    Using greed to solve problems seems like using evolution to solve problems or using a not-quite-friendly AI to solve problems.

    We envision the outcome we want, where the incentives we’ve constructed turn out to leverage greed in such a way that the right people have a (completely greed-based) incentive to do just the right things. But incentive structures don’t care about our values any more than evolutions or arbitrary AIs. They’ll just mindlessly arrange people and their greed in whatever way, and maybe that way will be terrible and completely not what we want. You allude to this at the end of your post, with the crime and whatnot, but I think it’s a much bigger issue with the whole idea than you seem to imply. I think it’s the problem.

    Goodhart’s Law seems like another relevant take on almost the same issue. If you pay people for X’, which you think is a good proxy for X, what you’ll get is X’, and its link to X may or may not still be intact; the more you pay for X’, the more tenuous the link will become.

    • suntzuanime says:

      That failed utopia in the second link is a damn sight better than our current topia. Let’s not have the perfect be the enemy of the good. It’s a good reason not to worship greed, like many on the right do; it’s not a good reason not to experiment with using greed to solve problems.

      • Said Achmiz says:

        Ah, I see my choice of linked articles to illustrate the AI example was poor. Fair enough. Consider me to have instead linked to a discussion of how a poorly programmed AI will tile the universe with smiley faces.

        Experimenting on populations of real live humans without their consent is ok, but you have to…

        … no, wait. It’s not ok. What’s the ethical justification for this, again? “Things are already pretty bad, so let’s throw social policy ideas at a wall and see what sticks”?

        Or are we suggesting that the people consent to this?

        • suntzuanime says:

          There is an argument that it is not ethical to do things to people without their consent, but it is a pretty hardcore libertarian one and one that does not find much traction with our government, which you can tell because we have a government.

          If you aren’t going full anarchist, it seems like you’re trying to say that the agents of the government are ethically justified in locking people up and taking their money and all the various nasty things the government orders done to people, but that they aren’t ethically justified in experimenting with interventions to keep nasty things from happening to people. Is it the scientific method itself that is ethically objectionable? I’m not saying it’s not, just want to make sure we’re on the same page.

          As far as the particular intervention in question is concerned, my understanding is that the participants do “consent”, to whatever extent the notion of “consent” is even philosophically coherent.

        • ozymandias says:

          Presumably either people already consent to participate at Roca (ooooh, potential confounder) or they are coerced into it some way. Being coerced into participating in a program with a control group doesn’t seem strictly worse to me than being coerced into participating in a program *without* one.

        • peterdjones says:

          There’s consent and consent. A government can’t hold a referendum for everything they want, but wise government endeavour test the water about controversial ideas [*], since backlashes and revolts aren’t good news for them.

          [*] eg accidental-on-purpose leaks.

        • MugaSofer says:

          The universe will not absolve those people of the consequences because you protest the options were “too risky to try without consent”.

          Refusing to choose is like refusing to allow time to pass, as they say.

        • Said Achmiz says:

          MugaSofer, that sounds like a fully general argument for any and all interventions in people’s lives “for their own good”. Because the universe will not absolve you of the consequences to the people you would otherwise fail to save!

          “I don’t care whether ‘the universe’ will ‘absolve’ you; keep your hands off my ” seems like the appropriate general response.

        • suntzuanime says:

          “I don’t care where you think my hands should go, they’re going where I put them” is the response to that. You don’t even really have a right to be left alone by other people, much less the government.

        • ozymandias says:

          There are some very important pieces of paper outlining the ways that the government and/or other people are not allowed to put their hands on me. Regardless when people are manhandling me I certainly have a right to object.

        • suntzuanime says:

          You have the right to object; the government has the power to ignore you. The pieces of paper do make good banners to rally under, but ultimately it’s the government that has to read them, and you’d be amazed at what counts as interstate commerce when they want to get their hands on it.

        • Xycho says:

          The fact that “The government has the power to ignore you” is (to me at least) an almost insuperably strong argument for not having one at all. In much the same way, I don’t like the part of the SIB example in the original post where the government keeps half the savings; any procedure whereby the people who manufacture regulation acquire more leverage (which money is effectively a proxy for) is generally a bad thing. They already have more ability to direct people’s behaviour than is conscionable.

        • peterdjones says:

          Your friends and family also have the right to ignore you when you ask to be left alone to your depression, alcoholism or whatever.

          Governments are just how people organise themselves, as Moraelin writes:

          ” If only it were that simple. I think if you look at history without the selective glasses of how it fits Ayn Rand’s _novels_, you’ll find that there was a subtly different common thing there: whenever you lost your rights and liberties, it was actually simply because rich and powerful fucks wanted more riches and power… and nobody could stop them. Yes, taking over the government is a popular way to do it, but even there the trend isn’t necessarily “giving the government too much powers” but equally often “the government wasn’t strong enough to stop them.” You can see such examples as: – the 1075-1990 civil war in Lebanon: rich fucks and political parties simply hired their private armies and started a devastating civil war, making everyone’s lives miserable in the process. Yeah, I’m so sure that in the ensuing war where extortion, theft, robberies and collecting “customs” at random checkpoints are everyday occurences was soo much better than a government you can control through democratic means. Newsflash: once those warlords managed to take over a region by just having a larger militia than the government’s army can do anything about, they _were_ the de-facto local governments, and they didn’t give a flying fuck about your thinking they should have less rights. – the chaos in Somalia: ditto. Nobody was in a position to stop the warlords’ private armies from plunging the country into chaos – various coups where an impotent government just couldn’t do more than watch a bunch of mercenaries or CIA/KGB agents shoot their way into the palace and have it their way. Yeah, I’m sure the people now live so much better because they didn’t give the democratically-elected government even enough power to defend itself. – heck, even the russian revolution: communism didn’t happen because of someone giving too much power to the Tsar’s government (though it already had too much,) but because said government couldn’t stop Lenin – the fall of the Roman Republic: it didn’t happen because some erosion of rights by the Senate, but because Caesar marched with the legions on Rome and nobody could stop him. Then he got all those titles and rights at sword point, as a conqueror. And, oh, the third century crisis? Yeah, it was soooo great that the government couldn’t even elect an emperor after the praetorian guard murdered the previous one. I’m sooo sure that after a century of internal warfare and millions dead, they were still congratulating themselves for having an impotent Senate 😉 – some corporation simply getting powerful enough to do what it damn pleases, because nobody has the guts any more to say “no” to the largest employer and land owner in the country. Do a bit of googling on “United Fruit Company” and how the term “banana republic” came to be. Etc. It seems to me that past a point such cases can _only_ be stopped by a strong enough government. Yes, an uncontrolled government is dangerous too. Which is why you have to control it. If you still (think you) are a democracy and don’t want it snooping on you, fine, organize it so it can’t without a mandate. But “OMG, governments are evil, keep them away” attitudes are just bloody stupid. That’s how worse problems come to you further down the line. A government is nothing more than how a large number of people organize to live together in _some_ way, and mostly because anarchies are even worse and power vaccuums just beg to be filled by the biggest bastard. And arguing for a government which _is_ a power vaccuum, at most buys some time until some bastard gets powerful enough to replace it with himself.”

        • Xycho says:

          They have the ability, they do not have the right. In addition, they also have a duty NOT to do so. It isn’t morally supportable to ‘help’ someone who would rather you didn’t, even if you think they would thank you for it afterwards. Everyone has the right to throw themselves off their own bridges without interference.

        • Anonymous says:

          “MugaSofer, that sounds like a fully general argument for any and all interventions in people’s lives “for their own good”.”

          Well … yeah. I fail to see how “that would result in doing GOOD THINGS” is a counterargument.

          Obviously, you don’t actually have the resources or understanding to efficiently optimize *everything*. But in general, yes, do Good Things.

        • Said Achmiz says:

          Anonymous: Because intervening in someone’s life “for their own good” against their wishes is generally bad.

          If you don’t think so, then I suppose we have different terminal values.

        • suntzuanime says:

          So we are talking full anarchism here, then. Good luck with that, guys.

        • Xycho says:

          Not so much anarchism; utter chaos isn’t good for anybody. It’s simply a matter of priorities. We currently have a distasteful situation whereby the agency of individuals is massively contained and curtailed, even when they are not directly interacting with impersonally large numbers of other humans.

          Mostly this appears to be due to an irritating level of risk aversion. For example, I would rather live in a society where people occasionally drop dead of overdose than one in which possession and use of drugs carries legal penalties. I would much prefer to live somewhere with the gun laws of Texas than of England, since despite the slight increase in violent crime the former at least more closely holds to the principle that on their own property a person can own and do whatever the hell they like.

          Thus as above – I maintain that it is unethical to do anything to a person without their consent, unless they have previously done so to others who are unable to enforce consequences themselves, and equally, that it is unethical to prevent a person from doing something to, with or for themselves, regardless of how damaging.

          (The usual response to this is along the lines of ‘but damaged people are expensive’. This is true. I also do not support giving new lungs to lifelong smokers.)

          On the other hand, within a framework where ‘do unto non-explicitly-consenting others’ is a permissible action, I don’t see that experimentation, scientific or otherwise, would be unethical. You’ve already crossed the line of whether other people are inviolable; once you’re there you might as well carry it to the conclusion that they can be test subjects.

        • peterdjones says:

          @xycho

          The Texan homicide rate is 4 times the English one. Not slight.

        • peterdjones says:

          @Xycho

          If someone is brought into an .ER in coma and patched up,
          Did they have something to them without their consent?

        • Xycho says:

          If they’re most people, no. If they’re a Christian Scientist, or they have a DNR tag as a tattoo/in their wallet/whatever, yes. How would the ER doctors tell the difference? Well, they pretty much can’t unless it’s spelled out explicitly, so in the absence of any information erring on the side of the patient not dying is probably wise, given the most commonly expressed preference.

          On the other hand, if they’re conscious and refusing treatment despite having a hole through their chest, the right thing to do is let them die.

      • ozymandias says:

        comment in wrong place; please delete

      • Eli says:

        When dealing with optimization processes, the good is the enemy of the perfect — opposite way around from normal life. And capitalism is a very Unfriendly optimizer, given that its only actual utility function is “stored/liquid optimization power” (which the foolish humans persist in referring to as “capital” or “money”).

  8. One more way to game the system– Goldman-Sachs pushes for shorter prison sentences. This might be a good thing, but it’s not quite what was intended, since what you actually want is less crime as well as fewer people in prison.

    On the other hand, things frequently aren’t as bad as the possibility of misapplying incentives would suggest. People have *some* benevolence and/or desire to do things well.

    • Douglas Knight says:

      I believe that this particular case is structured as an experiment and GS is judged not on absolute results, but compared to the much larger control group (which is why Ozy suggests sabotaging the control group). Thus phenomena that affect both groups, like laws and the weather, are irrelevant. But if you want to do general policy and not experiments, this is a problem.

  9. gattsuru says:

    I’d worry more that you’ll find your signals quickly corrupted. Goldman Sachs could recommend policy changes or advocate for ‘custodial parole’ or something similarly stupid, Planned Parenthood could find the definition of child expenditures changing drastically every eight years, and even otherwise rational changes in Census data can completely bork any sort of health care analysis.

    (There’s a good argument that this already happens in large portions of the public interest sphere. Sometimes the problem comes in bizarre forms, too, where providers of a solution overly hype or exaggerate the Problem, as in anti-crime initiatives whose advocates quickly come up with very questionably high crime stats.)

    That’s not to say it’s a bad idea, but that’s another weak spot you need to be aware of, and one that economists have historically not been terribly good at correcting for.

    I’m not sure what it means that we’re not doing those sorts of things already. But if we can’t figure out a way to solve those problems without bringing in a corporation to profit off of our incompetence, I say bring in the corporations.

    At a deeper level, we kinda are doing these sorts of things already, in many cases. Medicaid cover preventative care and (in most states) mental health care and IUDs, we have federal student loans and income based repayment and pay-as-you-earn training programs. The trick is that we’re not really at the point where we’re going to consider these things valuable in themselves, so we can’t start forcing people to see therapists or use birth control or see a doctor twice a year or take a college program that’s got positive return-on-investment.

    The corporation angle’s mostly nice because you can actually say no to a program that doesn’t make money, where as public programs with negative effects can keep running for centuries. But the flip side is that you don’t get much in the way of protections or direct oversight. We have public outrage over paying folk to undergo surgical sterilization for a reason, and it’s not something you can really change public opinion on by running through a private corp.

    • Anonymous says:

      That link doesn’t really explain the reason, beyond “they, ironically, do not sterilize their medical tools well enough”. That one seems relatively easy to handle with regulation.

      • ozymandias says:

        In general, people are upset about paying people to undergo surgical sterilization because

        (a) they object to the targeting of certain kinds of people (people of color, drug addicts, poor people, disabled people)
        (b) the payment is viewed as pressuring people into getting sterilized, especially if people are not paid to adopt other forms of birth control or other forms of birth control are expensive
        (c) historically, such programs have not been super-good about such niceties as explaining to their patients that they’re getting sterilized, that sterilization has health risks, that birth control options other than sterilization exist, or that sterilization is irreversible
        (d) sterilization is irreversible, so if women decide they want children later on (for instance, if they are no longer poor or addicted to drugs) they can’t have them; therefore, it seems particularly dubious to nudge people into getting *that* form of long-acting set-it-and-forget-it birth control rather than say IUDs, which are highly reversible

        • suntzuanime says:

          Well, if you’re paying people to have it done, it wouldn’t make sense for it to be reversible. People could take your money and then reverse it.

        • ozymandias says:

          So don’t pay people more than $200 (the cost to have an IUD removed).

        • suntzuanime says:

          That doesn’t really resolve the issue, it just reduces it from “free money” to “payday loan”.

        • Randy M says:

          I have to fight against the part of my brain that says “If someone can be tricked into being sterilized, isn’t it serving a eugenic function rather well?”

        • ozymandias says:

          suntzuanime: But the whole profit model of payday loans is based on the fact that a substantial number of people *don’t* have $200 on hand to pay them off and they can charge absurd interest rates. It seems unlikely that there’s a substantial population of people who need two hundred dollars right now enough to get surgery, have no other way to get such money (payday loan? borrow from a friend?) but will have two hundred dollars two months from now.

          Randy M: Not necessarily. Think about, say, a doctor who sterilizes a mother during a C section without her consent, or a woman who believes her doctor that the surgery is reversible. Unless you’re proposing that “wants C-section” and “trusts doctors” are traits which should be eugenically removed from the population…?

        • suntzuanime says:

          Well, if the sterilization is never going to be reversed anyway, it seems pointless to complain it can’t be reversed. If it is going to be reversed, you’re spending money and not getting the public policy outcomes you were hoping for. Either way, the math does not come out in favor of the IUD.

        • Randy M says:

          Ozy: Yes, trick is the wrong word there. Also, I don’t think those are necesarily hereditable traits, but perhaps evidence of some that are.
          I was thinking more of people who would take a straight up offer. Reminds me of Jacob and Esau; if you’ll sell your birthright for a meal, did you really deserve it?

        • ozymandias says:

          suntzuanime: Well, no, with an IUD you have a significant reduction in *unplanned* and *unwanted* pregnancies as well as in abortions (both of which are fairly common for poor women IIRC) and you have ensured that women must be in a slightly more financially stable place before they can have babies. I don’t think it’s likely that many women will want it to be reversed, say, two months out, but there are probably quite a few women who would want it to be reversed five or ten years out.

        • suntzuanime says:

          The public policy goals stated in the article were to reduce overpopulation. If you don’t think that’s an acceptable policy goal, say so. Don’t call them hypocrites for not pursuing a goal you like better.

  10. Benquo says:

    The problem with Goldman Sachs isn’t that they’re incompetent.

    Let me see if I can think of 10 ways GS could game the metrics on the reduced incarceration stats:
    1) Get Roca to work with the lowest-risk candidates instead of the ones they can best serve (maybe they are already doing this, this is actually the super boring null strategy and it will almost certainly work)
    2) Retain a good defense lawyer on staff to reduce the conviction rate for Roca clients.
    3) Take the other side of the deal and bet against Roca somewhere else, to extract arbitrage.
    4) GS could expect to lose money on this deal, but it might have a positive expected value when you include the publicity. Maybe it distracted people from some concurrent scandal. Maybe it bought them goodwill they’ll cash in for a political favor.
    5) If GS colludes with Roca’s clients to execute crimes that net them more than their initial investment but keep GS’s hands clean, GS can come out ahead even if it doesn’t get any of its principal back.
    6) Use Roca’s influence over “high risk” people to affect things GS already has investment positions in. For example, let’s say they short a real estate investor concentrated in one area, and then that area has a crime wave.
    7) Blockbusting! We all know “high risk” is an euphemism for people who look a little different from the fancy people, right? Just encourage Roca’s clients to hang out in formerly expensive neighborhoods, buy fancy properties on the cheap from terrified old racists, then flip those houses for a quick profit. Maybe have a business partner do this, for plausible deniability.
    8) GS wants the business of someone who is involved with Roca.
    9) GS is just an intermediary selling “social impact bonds” to clients and pocketing the spread
    10) A GS investment fund invested in this, not GS itself, the fund investors lose if it doesn’t pan out, no skin off GS’s nose, but either way GS reaps the goodwill.

    As for what I think is actually happening? Probably someone at GS strongly believes in Roca. Probably they’re not using the same though process on this as on their other investments – I’d be surprised if the people who evaluated this treated it the same as all GS’s other ventures. This particular hypothesis was privileged because it looks really, really good. (And you know what? It kind of is.)

    • suntzuanime says:

      3 and 5 don’t seem possible. 3 is impossible since the bonds aren’t paying out the full value of the intervention, so it shouldn’t be possible to successfully profit with mere arbitrage. They would have to be able to produce some value to make up for the leakage there. 5 is impossible because surely they could already collude with criminals if they wanted to, no need to buy a bond that they expect to receive no money from.

      2 sounds like a really solid end-around though. You could frame it as a positive, the clients get justice, the clients get to not go to jail which really sucks, and society gets to not pay for the clients to go to jail. But I doubt it was what was envisioned.

    • Douglas Knight says:

      None of 5-8 has anything to do with social impact bonds. GS always has the option of simply donating money to Roca.

      Only 1 and 2 constitute gaming the metrics. I don’t see why you included both 9 and 10.

      In fact, 9/10 is partially true. GS is betting that Roca will be as effective after expansion as it was in years past. Other investors are betting that it will be much more effective.

    • Scott Alexander says:

      1: Easily solveable by the government randomly assigning candidates, which in fact they are doing.

      2: This sounds kind of “nothing in the rules say dogs can’t play baseball” ish, in that even though it is not officially banned, in practice everyone knows if they tried it they would be told that obviously it is against the spirit of things.

      3, 4, 6, 8, 9, 10: These all seem like cases of Goldman Sachs also having some other clever money-making opportunity on the side, but they don’t actually reduce the benefit of the bond and might actually be prosocial.

      5: You know, committing crimes is against the law. If Goldman Sachs wanted to commit petty crimes to make money, I’m sure there’s an easier way to do it.

      7: Again, if Goldman Sachs wanted to do this, the social impact bond doesn’t really matter one way or the other except as a way of paying $9 million as a way of meeting criminals (not even real criminals! potential criminals!) I’m sure there’s a guy named Bruno at a local gas station who could do it for cheaper than that.

      • MugaSofer says:

        “This sounds kind of “nothing in the rules say dogs can’t play baseball” ish, in that even though it is not officially banned, in practice everyone knows if they tried it they would be told that obviously it is against the spirit of things.”

        Can the government actually welsh out on a bond because “you violated the spirit of the agreement”?

        I mean, they CAN, but … illegal, right?

        (Goldman Sachs would never get another such bond again, however. I don’t seriously anticipate they would be this stupid.)

      • Desertopa says:

        On the one hand, pulling a “no rule dogs can’t play baseball” in such a visible context would be likely to hurt their business later, but on the other hand, gaming incentives is a very significant part of GS’s stock in trade.

        If there’s some way to game the system that would result in a near term benefit at the cost of possible loss of business in the long term through nebulous and hard to predict forces or reputational effects, if anything, I suspect that the culture would favor gaming the system.

  11. Joe from London says:

    “I don’t know if it’s exactly a good idea to bring in the people who caused the financial crash to help the people who came up with the prison system.”

    A lot of blogs and journalists refer to “the people who caused the financial crash” and it’s usually just a form of banker-bashing. If this idea came into being, a desk at Goldman Sachs would probably be formed. It’s certainly possible that one or more people from that desk would have previously traded subprime mortgages, but it isn’t accurate to refer to those traders as having caused the financial crash.

    • anodognosic says:

      But of course, they’re not talking about *literally the same people*. They’re talking about the institution, one of a set whose corporate culture was, in fact, largely responsible for the crash.

      • Joe from London says:

        Yeah, it’s clear neither what’s intended, nor whether what I *think* is intended by that is accurate. It’s a common soundbite, but it’s not clear that there’s a specific belief you hold that won’t be shared by people who don’t think “goldman’s culture caused the credit crunch”

  12. Isn’t the “sabotage the control group” problem solved by either not telling them who the control group is or using the entire rest of the population as the control?

    • MugaSofer says:

      Well, not if they sabotage the entire country.

      Which would be a really terrible strategy on their part.

      Right?

      • Eli says:

        When your incentive is to capture wealth and power rather than creating wealth and power, sabotaging the entire rest of society in order to gain some power by comparison is actually a great strategy, provided nobody steps in to stop you.

  13. Randy M says:

    “Right now we have a system where we don’t really help people in need, unless the need becomes desperate, in which case we would feel bad about not helping, so we do, but then the cost of helping has gone up by an order of magnitude.”

    I think this is true only because the lines between needy and not-needy and between desperate and not desperate are subjective; and also because people who receive aid before becoming desperate are the unseen compared to those whose need is greater. For example, we have free public education that surely keeps some in the needy category out of the desperate category, similarly police patroling bad neighborhoods can keep people out of desperation. Plus, wouldn’t effective altruism want the resources pointed to the neediest, triage and the like? $100 dollars will go futher in terms of dead-babies when people are literally starving vs people who need a therapist to avoid depression, if for no other reason than one subset is a lot easier to identify.

    Also, I think blaming society for only helping the desperate is kind of missing the point of the country; that is, it was founded on the idea that people can mostly run their own lives without paternalistic coercion, with the help of long-standing voluntary instituations (church, family). Then again, society has gotten significantly more complicated, the upper limits of living standards have increased dramatically (so the needs of impoverished, while less absolute, may be more accutely felt), those voluntary institutions have been aggressively de-legitimized, and we have had significant demographic shifts, so those assumptions may no longer be valid, if they ever were.

    “So go to the potential criminals and tell them “I’ll give you $50,000 to not commit any crimes in the next few years. …
    The fact that this [prison scheme] would work probably says a lot about the inefficiency of prison compared to any other conceivable way of dealing with crime”

    If you want to call your priors facts you need to at least show your math. (Also, you admit it would only be a short term solution, and then use it to indict the current model–which I agree sucks, but that’s incidental.) Basically you are assuming crime is 50% need. While an impressive concession if you still want to label yourself liberal,
    consider the possible scenario of people stealing a lot more from each other in cars with much nicer rims.

  14. Michael Edward Vassar says:

    Greed isn’t actually the same thing as rational self-interest. Rational self-interest is a sub-type of rational interest in anything, which with a low enough discount rate looks like rational interest in anything else. Greed in practice is more like being susceptible to Pascal’s Muggings if they are delivered with the right emotional tone, or like a subset of conformity. It doesn’t actually produce much, but gets used a short-hand for rational self-interest.

  15. JRM says:

    As Banquo notes, a huge part of this becomes gaming the metrics very fast.

    Proper incentives are vital. Suppose, hypothetically, in Statifornia, they create an incentive plan for probation departments to avoid sending people to prison, instead telling them to meet their needs so that they need not go back to prison. A pot of money is made dependent on their success.

    You will find that probationers are much more successful under this plan. Well, according to the probation department – violations will go uncounted. If someone is hazardously close to prison, maybe you don’t refer it to the nasty ol’ DA’s office. Greed is only good if it makes things operate efficiently.

    When crime reduction incentives become sufficient, crime counting becomes an issue. Maybe that guy did stab himself. He said he stabbed himself twice. Maybe she was drunk and fell on her face real hard and wasn’t tuned up by her violent boyfriend.

    As these problems exist even without private enterprise involvement, I’d expect them to increase with same. I’d very much like to see the experiment done – but I am not so optimistic as to the results.

    (I live in Statifornia, am a nasty ol’ DA, and think the prison system has significant public benefits.)

  16. Shmi Nux says:

    As others alluded to, even if there is a way for a company to do well by doing good, there are usually many more ways to do better by doing less good, at least in the short term. And, if you agree that corporations are generally psychotic and short-sighted, they will attempt to take advantage of loopholes the way a jackass genie would, in the name of shareholder value.

    To make your ideas work you not only need to identify the areas where a private company can do well by doing good, but also to make it a stable equilibrium, where deviating from doing good is unprofitable.

  17. Doragoon says:

    I don’t trust a man who talks about ethics when he is picking my pocket. But if he is acting in his own self-interest and says so, I have usually been able to work out some way to do business with him. –Heinlein

  18. peterdjones says:

    It looks like Sweden is Raikoth, Scott. You gave a list of free this”s and early intervention that’s that neatly matches the actual behaviour of “socialised” countries with public healthcare systems. The only thing missing is the market.

  19. Anthony says:

    You never wonder if greed has an ulterior motive, because it’s already the most ulterior motive there is.

    Lust.

    Planned Parenthood offering free IUDs for women who need them, and the government paying them the money it saves from not having to put the kids through school.

    Reduced *welfare* spending is already claimed as a reason to support PP and taxpayer-funded abortion, though there’s not a *direct* linkage between the two. And it would be politically hard to make the alignment about education funding as that involves federal vs local spending issues. It would also be hard to measure given that the program already exists.

    We could have a health insurance company giving free preventative care to the poor, and the government paying them out of decreased emergency room visits.

    We’ve found that this doesn’t work. With a more direct linkage, those programs would have been dropped by the health care funders, and probably not just for poor people.

    A university accepting students without tuition, and the government paying them out of the increased tax revenue when they take higher-paying jobs.

    This assumes that a university education is what actually makes the difference. Outside engineering and accounting, that’s a hard case to make, and harder to measure.

    I don’t know if it’s exactly a good idea to bring in the people who caused the financial crash to help the people who came up with the prison system.

    The biggest problem for all such proposals is to design the reward payoff in such a way that it’s easier (cheaper) to actually solve the problem than to game the system. And that’s hard. Would you give the task of designing ungameable reward matrices to Goldman Sachs?

    • Doug S. says:

      Would you give the task of designing ungameable reward matrices to Goldman Sachs?

      Actually, I would, as long as they don’t get to buy into the reward system themselves.

    • peterdjones says:

      Education can make difference if you are starting off from a low enough basis, ad in the Mauritius Miracle. However, the returns are diminishing. The UK used to have a free tertiary education approach, that was justified by the naturally higher tax revenue from graduates, but iy could no longer be justified in the face of expansio, since expansion only means graduates taking lower payed jobs.

  20. peterdjones says:

    If you want to reduce the US incarceration rate, you could try asking someone within track record of less incarceration, rather than engaging in a speculative scheme, Since every other country has a lower incarceration rate, that would give you a lot of options as well.

    • Randy M says:

      But be sure to exclude countries with higher crime rates, because that would defeat the purpose. And of course any that don’t count certain activities as crimes that we still wish to discourage through force of law.
      And exclude countries with different racial/ethnic/socioeconomic mixes, because all that factors in (unless you are willing to change that…).
      Does that still leave us with a decent population to learn from?

      • peterdjones says:

        All the western European countries have lower crime rates.

        The larger ones, France , Germany and the UK, are quite ethnically mixed.

        The standout socioeconomic factor seems to be the higher inequality in the US.

        • Randy M says:

          Well, good! Are they all doing the same thing differently, or different things differently?

        • peterdjones says:

          @Randy:

          If doing the same thing means having some sort of criminal justice system, then, yes, they are doing the same thing differently.

        • nydwracu says:

          The UK is 87% white, 7% ‘Asian’, and 3% black.
          France is 85% white, 10% North African (I don’t know how genetically distinct they are, but they don’t look very distinct), 3.5% black, and 1.5% Asian.
          Germany is 91% white, 4% Turkish, 2.5% Asian (but that might include whites from Australia and New Zealand) and <0.7% black.
          The US are 63.7% non-Hispanic white, 16.4% Hispanic, 12.6% black, and 4.8% Asian.

          I wouldn’t call those “quite ethnically mixed” in comparison. There’s a large and important difference between 63.7% and 91%.

    • Anthony says:

      We tried “less incarceration”, and got lots more crime. Most countries with lower incarceration rates have either different racial mixes or are much less safe than the U.S. Can you provide an example of another country with a similar racial mix, lower incarceration, and lower crime?

      Or we could bring back the whipping post for minor crimes. Give convicts the choice of jail or whipping.

      • Multiheaded says:

        Can you provide an example of another country with a similar racial mix, lower incarceration, and lower crime?

        You’d think a “reactionary” would say, America 50 years ago…

        Give convicts the choice of jail or whipping.

        Has that ever actually been practiced? Singapore, the rare example of an advanced nation with corporal punishment, doesn’t let prisoners avoid jail this way…

        • Anthony says:

          Would you be willing to return to the policing and punishment policies of 50 (or 60) years ago? Here’s a sample:

          No Miranda. Perhaps we can replace this with a requirement to video-record all confessions, no matter how extracted. Let juries decide if the suspect was confessing, or just telling the cops something to make the beating stop.

          Whipping post. Get drunk and disorderly and arrested on Friday night, plead guilty, choose the whipping post, you’re home in time for work on Monday. And you’re sitting up straighter. Make this a choice, and maybe don’t allow seriously unhealthy criminals to get that choice.

          Racial profiling: We call it “stop and frisk” now – it worked pretty well in New York City, and it could go nationwide. Actually, New York’s policy wasn’t racially-biased – the percentage of people found with illegal weapons was basically the same across racial groups. But you’re stopping 4 people for every one you catch with something.

          Make jails a lot less comfortable. Pass a law that says if it’s acceptable at USMC boot camp, it’s not “cruel and unusual punishment” in jail.

        • peterdjones says:

          50 years ago…the US was more equal?
          🙂

        • ozymandias says:

          I believe people were calling stop-and-frisk racially discriminatory because blacks and Hispanics were far more likely to be stopped, not because they were more likely to be found with weapons. In fact, weapons being found an equal percentage of the time across groups increases the chance that it’s racially discriminatory: they can’t justify stopping more blacks because blacks are more likely to have guns. Also it’s less like “stop four people for every one you catch with something” and more like “stop nine people for every one you catch with something and 499 for every one you catch with a gun”.

          Also, making punishments more severe doesn’t necessarily increase the deterrent effect.

        • Anthony says:

          50 years ago, income inequality among U.S. whites was lower. Other inequality – not so much.

          Ozy – don’t have time to look up the reference, but I recall that the rates of finding something actionable were between 23% and 27% for all races, with blacks being in the middle of the range. That tells me that the cops are not targeting blacks for more attention than is justified.

        • suntzuanime says:

          The difference between USMC bootcamp and prison is that in USMC bootcamp you want to turn the prisoners into killing machines.

      • Douglas Knight says:

        No country has the same racial mix, but why not restrict attention to just whites? The Canadian and American white murder rates doubled and then returned over about the same period of time. America increased its white incarceration rates, but Canada’s didn’t change. So why believe incarceration did anything?

    • Scott Alexander says:

      “If you want to reduce the US incarceration rate, you could try asking someone within track record of less incarceration.”

      I’m not sure what you mean by this.

      • peterdjones says:

        I mean there is plenty of data from other countries, although apparently it is resistable if one assumes it’s all about race.

  21. Brian says:

    Ah, Love, could thou and I with Fate conspire
    To grasp this sorry Scheme of Things entire!
    Would not we shatter it to bits – and then
    Re-mould it nearer to the Heart’s Desire!

    I picked up a gorgeous illuminated copy of FitzGerald’s translation of the Rubaiyat a few weeks ago. Marvelous. I actually recognized it from references in the Principia Discordia, where they paraphrase a different quatrain:

    A leg of lamb beneath the bough,
    A flask of wine, a book of verse
    And you–beside me, singing in the wilderness
    And Wilderness is Paradise enough.

    Too much, man.

    • Scott Alexander says:

      I also have an illuminated copy! If you like it, be sure to look up the (several hundred) versus FitzGerald didn’t include – the Whinfield translation is pretty good.

  22. MugaSofer says:

    Excellent post, Scott.

    “Planned Parenthood offering free IUDs for women who need them, and the government paying them the money it saves from not having to put the kids through school.”

    Wouldn’t that be the money it loses in decreased tax revenues? Or are they supposed to compensate for this through … immigration, or baby farms, or something?

    • suntzuanime says:

      I think the idea is that children of “women who need them” (a lovely euphemism!) were never going to contribute much in the way of tax revenue anyhow.

  23. The particular article I read about this today was How Goldman Sachs Can Get Paid To Keep People Out Of Jail. It was the name “Goldman Sachs”

    Goldman Sachs are cozy with the regulators. In large part, they are the regulators. They get to regulate their competition, their suppliers, and their customers.

    So will probably get to decide how their success or failure at keeping people out of jail will be measured and judged

  24. Eli says:

    I think many people are against government social programs for a lot of the same reason that The Last Psychiatrist is against maintenance of certification exams (a position I totally called).

    No, they are against government social programs because they believe the poor deserve what they get. They believe that markets allocate deservingness, not capital.

    Yes, I realize I have “failed to steelman the enemy position”. The problem is that I’ve actually spoken to capitalist hard-right-wingers, and this is really what they sincerely told me about their ideology. Sometimes a position is actually just based on morals that you or I might consider abhorrent but which someone else really holds, rather than on any kind of factual thesis that we could steelman.

    Oh, and we used to actually have universities that accepted students without tuition in exchange for funding from the increased tax revenue. They were called “public universities”. They were defunded because they spawned hippies. I am not joking. Look up what Ronald Reagan did to the University of California system: he cut its funding and basically imposed tuition burdens simply because the students were perceived as ideological opponents of Governor Reagan’s voting base (blue-collar cultural conservatives hating on the libertine children of white-collar liberal professionals IS A THING). The idea was that having to pay substantial or even large sums for schooling would force those damn dirty hippies to stop fucking around with drugs and free love and get back to work like everyone else.

    A generation or two later, we have a student loan crisis, and universities talk of cutting departments that don’t directly contribute to getting students high-paying jobs just out of school.

    So despite appearing to us today as a horrid crisis, this is actually the semi-intended result.

    • peterdjones says:

      It might be possible to bring them round by noting that directing educational resources at the wealthy rather than the smart isn’t the way to maximise human capital and economic efficiency.

  25. Thor Russell says:

    I’ll give you $50,000 to not commit any crimes in the next few years. $25,000 now, in order to help you solve whatever problems turned you to criminality. And $25,000 at the end, after you’ve successfully avoided jail, as a reward.”
    Going to have to bite on this ..

    Well if you offered them a job in a bank instead then they would get more money and you may avoid the next financial crisis… oh and of course put the bankers in jail for symmetry if no other reason. It may make the prison guards job much easier.

    Seriously though I cant think of an easy way to keep people out of jail involving money without giving them a perverse incentive to get into the scheme by committing crime.

  26. Dodecahedron says:

    Prediction markets are promising because they use greed to fix epistemology. Neocameralism is promising because it uses greed to fix governance. And social impact bonds are promising because they use greed to fix social problems.

    I don’t think prediction markets use greed per se to fix epistemology. Just consider a prediction market that is set up like a game, where you can’t buy any real life goods with the money you earn, but you can still reinvest it in other predictions. Then there’s no real world greed in it (because the play money doesn’t let you command others’ work), but it will still work due to basic feedback. That is, the players who do well get more play money and so can invest more, while those who do badly get less and their signals are attenuated out of existence.

    From that perspective, greed (in the sense of playing the game for the power it grants you) only matters inasfar as it attracts people who are very good at predicting.

    Quoting Wikipedia:

    A common belief among economists and the financial community in general is that prediction markets based on play money cannot possibly generate credible predictions. However, the data collected so far disagrees.

    so that supports the concept that it’s not the greed (as such) that makes markets good, but the feedback that properly filters claims. The motivations of the players don’t matter as long as the players themselves are good enough.

    • Scott Alexander says:

      Maybe a better term would have been “self-interest” or something like that. The person playing the prediction market is playing to maximize their amount of play money or points or whatever – either because they gain status through that or because it’s a fun game.

      In any case, they want something other then the warm glow of signaling that America will definitely win the next war because America is awesome, go America.

      • Douglas Knight says:

        I think Dodecahedron’s point was not about what motivates the players, but that the market amplifies the opinions of people with past track records.

        • Dodecahedron says:

          My point is that standard explanations tend to focus too much on the greed or self interest part of it. If we assume the wisdom of crowds, the self interest doesn’t have to be all that strong: the players that play badly get excluded in short order anyway. We only need self interest to be strong enough that some good players get into the game in the first place.

          Self interest could take many forms, some of which don’t look like traditional greed at all.

          Say, for instance, you run a play market and people play it because it’s fun. Is the self-interest of enjoying playing a fun game greed? Not really, and it’s probably more gentle to the real economy.

          Or say that the US Government produces a market and says “the people with the best track record will be awarded medals for their service to the nation” (or be allowed to meet the president, or somesuch). Furthermore assume these medals aren’t of all that high value monetarily speaking. Then the market will select for those who find being rewarded for service by the US Government, well, rewarding. As long as there are enough of them that at least somebody in the pool is good at predicting, the market will work.

          The predictors might desire a medal to signal their faithful nature to the US, or they might do it because it’s tangible evidence that they’re helping their nation and such gives them more of a “glow”. The play market doesn’t care in any case.

          You could obviously say that all of this is self-interest, but if you go by that definition, pretty much everything is done in self interest. Even the aforementioned warm glow of “America is awesome” is then desired from self interest, because warm glows feel good.

          Maybe, for more complex prediction, play markets that are simply fun will turn out better anyway, for the same reason extrinsic motivation can hurt performance in creative jobs. But again, if what I’ve said is true, then the market will adapt to it by attenuating the players whose performance is degraded by that effect.