Hail Scott Siskind

Scott Siskind gets it:

A democracy provides a Schelling point, … an option which might or might not be the best, but which is not too bad and which everyone agrees on in order to stop fighting. … In the six hundred fifty years between the Norman Conquest and the neutering of the English monarchy, Wikipedia lists about twenty revolts and civil wars. … In the three hundred years since the neutering of the English monarchy and the switch to a more Parliamentary system, there have been exactly zero. … Democracy doesn’t always perform optimally, but it always performs fairly, … and that is enough to prevent people from starting civil wars.

Academia is different. Its state resembles that of pre-democratic governments, when anyone could choose a side, claim it was legitimate, and then get into endless protracted fights with the partisans of other sides. If you believe ObamaCare will destroy the economy, you will have no trouble finding a prestigious academic who agrees with you. Then all you need to do is accuse the other academics of bias, or cherry-picking, or using the wrong statistical test, or any of the other ways to discredit scientists you don’t like. …

A democratic vote among the scientific establishment is insufficient to settle these topics. The most important problem is that it gives massive power to the people who determine who gets to be part of “the scientific establishment”. … So not having any Schelling point – being hopelessly confused about the legitimacy of academic ideas – sucks. But a straight democratic vote of academics would also suck and be potentially unfair.

Prediction markets avoid these problems. There is no question of who the experts are: anyone can invest in a prediction market. There’s no question of special interests taking it over; this just distributes free money to more honest investors. Not only do they escape real bias, but more importantly they escape perceived bias. It is breathtakingly beautiful how impossible it is to rail that a prediction market is the tool of the liberal media or whatever. …

Nate Silver might do better than a prediction market, I don’t know. But Nate Silver is not a Schelling point. Nobody chose him as Official Statistics Guy via a fair process. And if someone objected to his beliefs, they could accuse him of bias and he would have no recourse until it was too late. If a prediction market is almost as good as Nate, and it is also unbiased and impossible to accuse of bias, we have our Schelling point. …

Just as democracy made it harder to fight over leadership, prediction markets make it harder to fight over beliefs. We can still fight over values, of course – if you hate teenagers having sex, and I don’t care about it, we can debate that all day long. But if we want to know whether a certain law will raise the pregnancy rate, there will be only one correct answer, and it will only be a mouse-click away.

I think this would have more positive effects than anyone anticipates. If people took it seriously, not only would the gun control debate be over in an hour, but it would end on the objectively right side, whichever side that was. If single-payer would be better than Obamacare, we could implement single-payer and anyone who tried to make up horror stories about how it would destroy health care would be laughed out of the room. And once these issues have gone away, maybe we can reach the point where half the country stops hating the other half because of disagreements which are largely over factual issues. (more; HT Stephen Bachelor)

By the way, my futarchy paper will appear this year in Journal of Political Philosophy. This is very close to the final version.

GD Star Rating
Tagged as: ,
Trackback URL:
  • lump1

    Here are two possibly ill-considered objections to the idea of setting policy by the prices in a prediction market, where a governing body would be compelled to do X if benefits conditional on doing X were valued higher (therefore judged more probable) than benefits conditional on any other policy. The paper gives me some idea of how you would respond to them, but I still don’t quite get it.

    1. Suppose policy Y would make a giant windfall for an already-wealthy, large interest group. They coordinate to buy “great benefit conditional on doing Y” shares, until they drive up the price of Y above every other competing policy. Thus they get Y done and reap the windfall, which is larger than any loss they would suffer if Y doesn’t actually cause great benefit and their shares are worthless. As an added benefit, if Y isn’t done, their attempt at manipulation will cost them nothing.

    2. Broad trends are largely insensitive to concrete policy changes. In a time of recovery, the improving state of national happiness will not be derailed by a wide range of interest rates, or NASA budgets, or tariff policies, drug laws, etc. Likewise, even very sensible policies in times of contraction will not halt the contraction, but only prevent it from getting worse than it otherwise would have been. But that’s not how payments in prediction markets work. The shares become active when the condition was satisfied, and they pay when overall benefit has increased since the satisfaction of the condition. But critically, it might not increase *because* of the satisfaction of the condition; the most likely source of any increase are unrelated, broader trends. What I’m saying is that if things are getting better, then they’re gonna get better conditional on just about any policy, even unwise policies. So the winners of the prediction markets are really people who correctly guess broader trends, and not those who have any insight into the causal consequences of policies on outcomes. So now the objection: Suppose that you expect broad outcomes to improve, and you have a pet policy P which you advocate for expressive reasons and without evidence: Suppose you want to see automatic rifles legalized, or nuclear powerplants shut down, or military support for Israel is quintupled. You have no reason to think that these things will help general happiness, but they will make *you* happy, so you buy lots of shares conditional on these actions – as you say: expressive voting. And since you did it during a recovery, you can be pretty sure that outcomes are improving, so they’ll be improving even with automatic weapons. This way, you make money and get to ram through your favorite irrational policy.

    • On 1, the money available to those willing to lose bets to support a decision favoring some group is much less than the money available to speculators willing to win bets against them. The paper talks about this manipulation issue at length, and I’ve discussed it many other times.

      On 2, if speculators expect good trends then they’ll charge higher prices for national welfare futures, either with or without each proposed policy. Policies are only approved if expected welfare is *higher* than without.

      • IMASBA

        “the money available to those willing to lose bets to support a decision favoring some group is much less than the money available to speculators willing to win bets against them.”

        Really? If the Koch Brothers put $10 billion on the position of fracking not being harmful to the environment (knowing that the government takes advice from prediction markets) then Greenpeace is just gonna put in $11 billion? Sure, the global 1% “only” own 40% (and rising) of the world’s wealth, so the rest of humanity could theoretically outbid them but that assumes a) everyone can afford to gamble, b) everyone wants to gamble, c) everyone is smart enough to understand they’re being duped by monied interests and d) no one  turns on their own by siding with the (perceived) “winning team” (owning 40% of all the wealth and being very homogenous) of monied interests to make a quick buck for themselves.

        Besides, aren’t prediction markets just thinly veiled mob justice? With the mob being a colorful collection of monied interests, gambling addicts and fools who overrate their own understanding of issues (especially when the real chances are closer to 50/50 than 100/0)? You could get a Pakistani prediction market to say that Jews make bread from the blood of Muslims babies and an American one to say that the world’s scientists are conspiring to prop up a “climate change” con. Why would the primitive lizard part of my brain that drives my greed and fear of losing money be able to suddenly become an expert on everything as soon as I get the chance to gamble?

        Finally, prediction markets will just shift the problem, not make it disappear: you will always have to define what counts as outcome A and what counts as outcome B. Because it would get overly technical to do so in excruciating detail you will have to simplify and that will draw criticism from people that the market is biased (“why do you define Obamacare failing if 100k die in hospitals per year, Rush Limbaugh said the bar should be 20k !”) No, the real problem is people getting told that there are no stupid opinions, that you don’t have to stfu if you don’t understand the science behind something, that good PR is more important than being right or on the right side of history. That we still hand out PHDs to bogus disciplines like medieval universities.

  • John Maxwell IV

    What’s the estimated cost to lobby the government and make whatever legal changes would be necessary to allow unfettered privately run prediction markets?  Could be an interesting project for an open-minded philanthropist.

  • Salemicus

    “In the three hundred years since the neutering of the English monarchy and the switch to a more Parliamentary system, there have been exactly zero [revolts and civil wars].”

    Who was Bonnie Prince Charlie?

    • Goatherd

      One might also bring up the American and Irish wars of independence, if one were inclined to take a wider view.

      • VV

        The American War of Independence might be a borderline case, since it is argued that the colonials had little or no representation in the British parliament.

        The Irish War of Independence and the American Civil War, on the other hand, occurred between factions of citizens with full political rights.

  • The author might be better known as Yvain on LessWrong.

    • Robin Hanson

      Am I to understand this comment gets six likes because people want to say that in general it is better to call people by the name they use at Less Wrong than by the ordinary name they’d use in most other contexts?!

      •  At the time I originally posted it he was referred to as squid314. Or maybe they are familiar with Yvain from there and are glad to have it pointed out that these are the same people.

      • Arthur

        I think people are liking this comment because they see it as some kind of public service announcement, so they want to keep it on top, as some sticky things in forums. Especially when the name was “squid314”.

  • Arthur

    What if you want to state an end, but want to reach a non-stated end, through a policy that’s inefficient to the stated end but efficient to the non-state end?

    • You’d be limited to adopting misleading end measures as part of the national welfare definition. For example, if you really want to promote beauty, but want to pretend to promote health, you can use weight measures.

  • Deletethiscomment

    although not a final version, there’s a typo in that pdf article. 6th paragraph 3rd row.

  • Geoff Brown

    The critical question of our age, imo, is what faith can we put into schemes of legitimacy at all? The West is fast approaching the end game of its legitimacy as more and more constituents lose faith in ideas that are crumbling against the waves of reality.  While I don’t disagree with any particular attribute of prediction markets, my intuitive (far?) response is that prediction markets will not save us from communal decay.  I will gladly tilt my lance in the service of an institution that I believe would help, but its hard to stir feelings of righteousness with prediction markets.

    • Can you not find any feelings of righteous indignation that the truth is so often hidden? I say, to the barricades, for truth.

      • Geoff Brown

        I’ll start boiling the oil. 🙂

  • Michael Vassar

    I think that the succession management function of democracy suggests that people should have votes proportional to their value as allies in a military conflict.

    • Anon1234

      One counter-example to this “no democratic civil wars” hyothesis is the American Civil War. Perhaps escaped slaves who joined the Union army (and saboteurs) were significant enough to change the outcome of the war; Confederate voters approved a suicidal secession because of poor sampling.

      Maybe this is a warning to future generations – don’t classify anyone as “not people” if they can feasibly act to secure their own freedom.

      • VV

         I don’t think runaway slaves were enough to make a difference, and it’s not like they had a right to vote in the Union.

        Democracy might reduce the risk of revolts and civil wars, but not suppress it completely.

    • Robin Hanson

      Democracy is trying to simultaneously prevent revolt, *and* to gain legitimacy by supporting forager egalitarian norms. It so happens in our world that equal votes among adults tracks revolt power near enough to work. In another world, it may not be remotely near enough.

    • Douglas Knight

      English suffrage has followed military value reasonably well for the past thousand years.

  • It seems a bit odd to hail someone for agreeing with you.

    • Really?  What an enormously huge, Everett-sized availability bias…

  • John Smithbg

    Don’t healthy prediction markets require mass participation (in order to capure all the locally distributed (Hayekian) information AND rationality of participants? Because if both are true, this idea is doomed to failure.

  • Jim


    Advice for the (Far?) future:Make sure someone keeps track of your frozen head.

  • VV

    I don’t think that these policy-conditional prediction markets necessarily reward betting according your true beliefs on the policy performances:

    Suppose that I am a lobby that would have my interests damaged if a particular policy A was enacted. If I have enough money, I can dishonestly bet on that policy being a failure, manipulating the market to bias the estimated policy performance downwards. Since the government choices its policies according to the prediction marked, the policy I sunk is not chosen, and the corresponding market is cancelled, which means that I get back all the money I invested to manipulate it. In fact, if the market is subsidized, I can even profit.

    • No one has enough money to ensure that the option you unfairly denigrate won’t be chosen. Yes when odds are low of it being chosen your expected costs are less, but then so are the expected costs of others who will bet against you. They have a lot to gain that way, and a lot more money to play with.

      • VV


        No one has enough money to ensure that the option you unfairly denigrate won’t be chosen.

        How do you know? Keep in mind that the money used in manipulating the market is not lost, it is recovered when that market is cancelled because the policy was not chosen.

        Yes when odds are low of it being chosen your expected costs are less,
        but then so are the expected costs of others who will bet against you.
        They have a lot to gain that way, and a lot more money to play with.

        I don’t think so. Keep in mind that true predictors have a limited amount of money to bet, and making accurate predictors costs them time and resources, thus they are more likely to focus on markets which have a significant chance of coming to maturity.

      • VV

         It may make sense to sell tokens for each choice with mutually exclusive options and allow the predictors to bet the same token on the corresponding mutually exclusive markets, since all but one of them will be cancelled. That would increase their liquidity, but they would still incur in the prediction costs.

      • John Smithbg

         Sorry, but this is not even borderline ridiculous. You are assumming a) that the access to credit of the interested group plus their own resources will be less than the resources plus the credit access of interested predictors; b) that predictors in this market will be generally rational AND have access to all the available information on the matter AND will be able to reflect and act rationally on it (or at least the irrationalities will be normally distributed) AND the market as a whole will incorporate at least to some extend insider information; and c) that it will be impossible for a said group to mainpulate the market procedure via the bureaucracy implementing the market infrastructure and/or the transfer of the results to the policy realm.

        While a) is a purely empirical question and you can make the claim that in the future it is likely to be a valid premise (it is not now), both b and c are contradicted by the majority of the litereature on EMH and financial market in general and on … well, pretty much everything that has happened in politics ever.

        You have some brilliant writings (the farmer vs forager topic ) for example, some that are “crazy” but still possible (your views about the EM economy), but this…this is beyond reason.

      • Paul Christiano

        Re (a): I don’t know what you mean by saying this is not true now. If “interested predictor” means “investor”, then this is true today for the vast majority of interests. What evidence do you have in mind? There are serious disanalogies between successful contemporary market manipulations and the case under discussion.

        Re (b): What is the relevance of “irrationalities will be normally distributed”? The point here is that you need only a few predictors, not for most people to be good predictors. They don’t need to have all the info (knowledge of a manipulator would be sufficient on its own, for example, and other partial info would be plenty good). And yes, they do need to incorporate this relevant information “to some extent,” but that is a laughably weak assumption.

        Re (c): Indeed, if an interested party can arbitrarily change what policies get implemented, then they can arbitrarily change what policies get implemented. I don’t understand what manipulation scenario you are imagining (besides the scenario where the manipulator has a very high probability of influencing which policy gets picked, so that the other policy conditional markets will reliably get called off). 

  • VV

    Making accurate predictions has a cost. If a market is cancelled, all the costs incurred by the predictors who participated to it are lost. If most markets are cancelled, unless the predictors expect very large payoffs on the few markets whose assets come to maturity, the business model is unprofitable.

    If you subsidize the cancelled markets by paying participation premiums, then you will attract pseudo-predictors who only bet on the markets that are likely to be cancelled just to collect the subsides. Since these pseudo-predictors don’t have to make accurate predictions, they will be likely more profitable than true predictors who spend resources to make accurate predictions. Therefore, the market estimates will be dominated by their activity.

    • If the right to make a proposal is auctioned, and if those who pay to make a proposal that is accepted are paid 20x their auction price, then proposals should have at least a 5% chance of being accepted. A $1 million bet on that proposal has a $50K expected payoff, which can be plenty enough to get traders to pay attention.

      • VV

        A $1 million bet on that proposal has a $50K expected payoff

        Can you explain?

      • A 5% chance of win $1M = win $50K.

      • VV


        A 5% chance of win $1M = win $50K.

        That applies to (risk-neutral) policy proposers, not to predictors.

  • VV

    Policy-conditional prediction markets provide only limited feedback, hence they could perpetuate biases. Suppose that predictors erroneously believe that a certain kind of policies X fails. Each time a X-like policy is proposed, they bet against it, thus the policy is never chosen, even if it could have been actually a successful policy.

    Note that this doesn’t happen in unconditional prediction markets that normally come to maturity, providing feedback on the prediction.

    • You are talking about a generic problem of not experimenting enough, which applies no matter how you calculate the locally best action given what you know. If taking into account the value of experimentation the nation is better off doing something weird this time, because while the direct benefits are negative there’s a positive gain from trying something different, then speculators should estimate that to be true and it should happen.

  • Arch1

    “There’s no question of special interests taking it over; this just distributes free money to more honest investors”

    Apologies if this is redundant w/ other comments:

    No, special interests taking over a prediction market doesn’t *just* do that – it also gives the special interest whatever benefits accrue to them from the policy they have artificially caused to be implemented.  Could not these benefits outweigh the cost of the free money distribution referred to above?

    • Arch1

       I now see this is redundant with Lump1’s item #1, to which Robin responded (and would’ve therefore voted my OP down, but I don’t see how to do that:-)

  • Arch1

    Prediction markets mine peoples’ *existing* knowledge and beliefs on an issue.  Has anyone looked at integrating prediction market technology with “debate representation” technology*, with a view to *improving* peoples’ knowledge and beliefs on complex & controversial issues?

    *there must be a better term for tools which represent the evolving logical structure of a debate, but I don’t know it.

  • Matthew Hammer

    My concern is that a futarchy could be hijacked not through dishonest bidding (which I agree is unlikely for reasonable distributions of wealth and payoffs) but through corruption of the underlying mechanism itself. 

    If a revolutionary group can make a credible promise that upon gaining power they will pay off all accepted policies at a maximum GDP+ judgement, then speculators should incorporate that expectation into their bidding on those policies that have an effect on the chance of that revolutionary group gaining control of the government. Thus those policies that expose the government to takeover by that revolutionary group will gain an inflated price compared to the “true” GDP+ outcome and thus are more likely to be instantiated. 

    Essentially, there is no trusted “higher power” to ensure judgments on the futures markets are honest, there is only the government itself. Thus the system is vulnerable to this sort of public subversion. 

    • Why not just promise to pay all soliders a big bonus of they don’t resist your military takeover? 

      There could easily be trillions of dollars tied up in assets in these markets. The rebels might not be able to add much to that percentage wise. 

      Also, if x is GDP+ scaled to be in [0,1], ordinary cash of $1 is equal to $x + $1-x. So presumably you need’d to offer only to reward people who hold a *net* position in $x,and not in $1-x. So you’ll need good accounting of who held what net positions when, accounting that survives this revolution.

      • Matthew Hammer

        I wasn’t suggesting adding some additional payoff (such as you would need to in order to pay bonus to soldiers). I was talking about those trillions of dollars of assets in the policy futures market. I’m suggesting they could promise to alter the distribution of those funds by forcing a judgement at odds with the current GDP+ function. 
        The accounting of various investors positions in the market must already exist, since that’s integral to the futarchy working in the first place. I’ll admit I am presuming that said accounting information does survive the revolution, but I’m not sure why that should be a particularly unlikely assumption over the set of possible revolutions.
        Perhaps I don’t understand how you envision the institution to be set up? Is there some reason to believe that the institutional structure that resolves and pays out the final values of the policy futures is not subject to the power of a government that wishes to alter the outcome? Or that the information necessary to track the options and resolve the payouts couldn’t survive a transition in authority? 
        Our current government has altered obligation payouts in recent history, changing the order of obligations fulfilled in automotive bankruptcies (moving union pensions ahead of other obligations). I don’t have any clear parallel to the preservation of contact information across a change in government at my fingertips, but I don’t expect that much stock ownership information was lost in the transition between the Weimar Republic and Nazi dictatorship (just to use a well-known example, not to draw any additional parallel). 
        A futarchy has just as much leeway as a republic to enact policies that can cause a transition to a new power structure, even if policies like “Give all power to the Restorationist Party” are nominally disallowed. 

      • I agree that a futarchy government would want to be especially careful and perhaps even ruthless about trying to prevent this kind of undermining of the GDP+ measurement. One approach would be to contract with outside parties elsewhere in the world regarding GDP+ payouts, so that the revolutionaries would have to take them over as well to execute this strategy.

  • >In the three hundred years since the neutering of the English monarchy
    and the switch to a more Parliamentary system, there have been exactly

    I count two: the American Revolution and the American Civil War.  That of course does not negate Scott/Yvain’s point.

  • Pingback: Hacking Technological Determinism for Fun and Profit | More Right()

  • Pingback: Overcoming Bias : How To Reform Academia With Prediction Markets()