Futarchy: Vote Values, But Bet Beliefs
hanson.gmu.edu
hanson.gmu.edu
The good from this policy is concentrated on the corn producers, and the harm is spread diffusely across all people who consume sweets. If I am an average consumer of sweets, I see little benefit fighting this policy today, since it will only change the price of a soda by a few cents.
Under a "Futarchy", the corn lobby can spend lots of money to "predict" that a sugar tariff will make the country better. They might even believe it, but I will have little incentive to vote with my wallet against such a policy, since the harm will be diffuse and extremely difficult to conclusively pick out from the noise using econometric techniques. Result: a strong tariff, which almost all economists agree is a bad policy. [1]
Just because Futarchy is vulnerable to some of the same problems as democracy as practiced in the USA doesn't mean it is a bad system, but I fail to see how it is functionally different than a plutocracy. [2]
The problem with prediction markets is that they are only accurate in so much as one's incentive to "waste" money producing an incorrect prediction is smaller than the amount of money to be made by having that incorrect prediction.
Hanging all law off of the results of prediction markets creates HUGE incentives for me to game the prediction markets, then extract rents elsewhere in the economy to more than recoup my investment.
[1] https://en.wikipedia.org/wiki/Tariff
[2] https://en.wikipedia.org/wiki/Plutocracy
(Edited for grammar, citations).
And I think Goldman Sachs wins that fight every time. Which is the whole purpose - to let the smart money guide the policy.
And even if that was the case - define smart. Smart in the interest of Goldman Sachs is to make more money. Smart in the interest of humanity, society or a community can be (and often is) something entirely different. Who are we creating policy for?
Eventually it does. Manipulating markets == handing your money to someone else, so the rich manipulators will exponentially [1] lose money as the smart speculators take it.
In this case, it's "smart in the interests of predicting which policies achieve the democratically selected objective function". For example, smart in predicting whether Obamacare will bend the cost curve (assuming that's the goal of obamacare).
[1] Assume 50/50 bets. The speculator bets $1, and takes $1 from the rich man. Next bet, he bets $2, takes $2 from the rich. Next bet he bets $4. The rate of growth doesn't have to be ln(2), but it is exponential.
Frustratingly, it doesn't specify what happens to outstanding "conditional" bets if the policy isn't passed, but a reasonable guess is that it's returned less transaction costs, which means that every time Goldman Sachs does the public the service of vetoing a bad policy it loses money. Even if speculators' transaction costs are zero it's irrational for them to bet against a market actor that will take a profit (from selling corn) even if they pay more than the market price to for their long position. Assuming the corn industry is really committed to supporting tariffs and that the tariffs won't harm speculators' ability to earn from other markets, the Nash equilibrium strategy for speculators is not to bet.
To do that they would have to bet a lot of money that the tariff will make the country better (by an objective metric which is decided upon by the voters), and if that turned out wrong, they would lose a lot of money.
Only societies without strong centralization and with incentives on the elites that stand opposed to collusion have the promise to evade what happens to democracies. An example is that which prevailed for a fair time in Medieval Iceland, but there are few others: the opening society of the US gave way to growing centralization pretty quickly, and the original vision didn't last much past two generations. There were not strong enough incentives to prevent collusion of the elites against the masses.
It's a measure of how much money is spent back and forth and not much else.
Some consider this money moving to be "progress" but it actually only highlights how many problems a society's people are willing to pay in order to solve.
The medical industry is a classic example of this. People are suffering and paying an arm and a leg for cures and treatment. GDP reflects the money aspect of this while ignoring the human aspect.
[0] http://www.youtube.com/watch?v=LFjApWH8R9c
A speech by Noga Alon [1], Baumritter Professor of Mathematics and Computer Science, Tel Aviv University
[0] http://www.daviddfriedman.com/The_Machinery_of_Freedom_.pdf
Political scientists aren't all idiots, people. Social contract theory exists for a reason.
When a betting market clearly estimates that a proposed policy would increase expected national welfare, that proposal becomes law.
This is not how a normal market works. In a normal market, everyone gets to choose their own strategy for how they are going to act. The market only conveys information to them, through prices, about how good (or bad) their current strategy is.
In futarchy, the "market" would determine a strategy that everyone would then have to follow, because it's the law. In other words, it still retains the key feature that makes our current system of government broken: whatever is considered "a good policy idea" is imposed on everyone, instead of being tested by allowing people who think it's a good idea to try it out individually.
To really fix the current system, we need to really look hard at all the things we currently think have to be decided by a single policy for everyone. There may be some things (like national defense) that actually do need to be decided that way. But a lot of things don't, yet our current system forces us to adopt a one-size-fits-all policy for them. Futarchy does not fix that: in fact it is likely, if anything, to make it worse by making it easier for special interests to simply buy the policy they want, by eliminating the middleman (instead of having to buy politicians and elections, just buy into the betting market).
We do not have a single concept of "national welfare". We have a multitude of things that various people think are part of national welfare, but different people pick different sets of things, and they often conflict with each other. Many of these conflicts are fundamentally irreconcilable.
In our current system, we force people who do not agree with a certain policy goal to abide by laws aimed at achieving that goal, if there is a sufficient majority to pass such laws. That is bad. Futarchy would not change that. Imposing private law anarchy on everybody because the betting market says it would best achieve national welfare is just as bad as imposing, say, the Federal Defense of Marriage Act on everybody because a particular Congress passed it into law.
But futarchy adds another bad thing to this already bad system: now not just individual policies, but the measure of success for any policy, the definition of "national welfare", gets decided by majority vote, and that measure of national welfare gets imposed on everybody, whether they agree with it or not. So futarchy imposes more things on everybody against their will than the current system does. That makes it worse than the current system, not better.
> Imposing private law anarchy on everybody because the betting market says it would best achieve national welfare is just as bad as...
It sounds like you're defining literally any society as "imposing" its traits on its members. I suppose that's true in a way, but I don't see it as a very useful definition, since there's no way to simply not have society (apart from something like transhumanism).
Yes, I agree, and I didn't say anything that contradicts this. My comments were about futarchy, not about Friedman's anarcho-capitalism.
It's the difference between "everyone votes on which car should get made, and only the winning car gets made" and "everyone pays for whatever car they want." Most economists and people agree that the latter works better for cars, and I don't think that the production and enforcement of law is an easier task than automobile production.
You should direct these comments at Hanson, not at me. These are criticisms of futarchy as well as of the democracy we have now, and I agree with them. In fact I am arguing that futarchy is worse than the democracy we have now, because it requires "everyone votes on what the measure of national welfare should be adopted, and only the winning measure gets adopted" in addition to "everyone votes on what laws should get made, and only the winning law gets made".
In a sense I am, but I should first clarify the comment of mine that you were responding to. I wasn't saying that private law anarchy imposes things on people per se; I was saying that it imposes things on people if they are forced to adopt it because the national betting market said that would be the best way to improve national welfare. The whole point of private law anarchy is that people choose it voluntarily.
With regard to the more general point, yes, living in a society at all imposes certain restraints on people, but there are different ways that can be done. One way is for people to rationally understand the benefits of living in a society as opposed to living as Robinson Crusoe individuals, and to be willing to accept restraints that are necessary parts of getting the benefits. The other way is for society to impose the restraints regardless of whether the people being imposed on agree with the need for them. In any real society, there will be some element of the second way; but I think all of our current societies are far too quick to adopt the second way instead of letting the first way work.
Not quite. I think that a private law/anarchy system would achieve the ends that most ordinary people want to achieve, better than any system that exists now, if people actually understood how it worked and were able to act accordingly. But that's a big "if".
And you think speculators will agree with you, and in a futarchy would therefore reject anarchy proposals.
I think this is probably true; I don't think there would be a significant number of people who would be willing to bet in favor of a private law/anarchy system. But, as above, that's not because it wouldn't actually achieve the ends people want to achieve; it's because too few people actually understand how it works.
However, I'm not sure how this is relevant to my criticisms of futarchy. I'm not criticizing it because I don't think it will lead to a private law/anarchy system; in fact I'm not criticizing it on the basis of any particular outcome I expect it to lead to. I'm criticizing it on the grounds that it requires a particular definition of "national welfare" (the one that gets the majority vote) to be imposed on everyone. I don't think "the ends that most ordinary people want to achieve" can be captured in any such definition. More precisely, as soon as you settle on one particular definition, someone will come up with a way to game it, by finding states of affairs that look good in terms of the definition, but do not actually achieve the ends that most ordinary people want to achieve (although they do achieve ends that the particular parties who are gaming the system want to achieve).
I don't think it's possible to come up with a definition that isn't vulnerable to this failure mode. The only way to avoid it is to discard the whole idea of "national welfare", which of course also means discarding the idea that there are policies, single policies that can be imposed on everyone, that will improve "national welfare", if only we can find them.
No; I don't think there's any such thing as "true national welfare".
True welfare somehow is intrinsically unmeasurable.
No, that's not what I said. "Welfare" is a much broader term than "national welfare". I only said "national welfare" was unmeasurable, not "welfare".
But also, the vote on the measure of national welfare would be just as susceptible to manipulation as votes for elected representatives are under our current system, because voters would still face the rational ignorance issue: even if we assume, for the sake of argument, that it is possible to find a single measure of national welfare, it simply would not be a net benefit to most voters to learn enough about the various aspects of national welfare to be able to cast an informed vote on what would be the best measure of national welfare. So most people's votes would be determined the way most people's votes are determined now.
Futarchy seems promising if we accept the following three assumptions:
Democracies fail largely by not aggregating available information.
And we don't even need to go any further, because assumption #1 is wrong in two ways:
(1) Democracies fail because people have conflicting goals, not just conflicting beliefs about how to reach goals. Even people who have all the same factual beliefs can still have inconsistent goals. That is a problem in any system that requires everyone to adopt common goals in order to set policy.
(2) The "information" that would need to be aggregated doesn't exist anyway. Hanson assumes that there is some policy that, if imposed on everyone, would work; he never considers the possibility that the real problem is that assumption--that the root problem is the very act of imposing policies on everyone, not the particular policies that get imposed.
As for the problem of imposing policies on everyone, doing it by a mechanical procedure based on a betting market actually does strike me as worse than doing it the way we do it now, for a couple of reasons. But I'll respond further to that in the other subthread where we are having that discussion.
This is taken as obvious, but I don't find it so. There are many things that are not attributable to tangible national assets, but that can nonetheless change its course: things such as historical accident of where a particular thing was invented, actions of other nations, natural disasters, etc. I see every reason to think that this sort of thing is responsible for wealth disparity, moreso than repeated adoption of "dumb" policies.
It's a delicate issue, because you need to provide a long term stake to force the decisions to be optimal (so future welfare is evaluated correctly and optimized), but at the same time paying off those stakes leads to an unstable plutocratic scenario.
Some kind clever compromise could probably be reached, I think, and this is one of those wonderful ideas that will never be implemented.
I suppose you could weight them by importance (based on score voting maybe), have people bet on each policy's effect on every goal, and apply an optimization algorithm to the whole thing to get policy decisions, but not many people would understand and trust the system.
This is where the 'vote on values, bet on beliefs' part comes in (did you read the paper?). The voters' job is to specify how much importance they place on various goals or metrics, and the prediction markets are asked how to achieve the goals.
In principle I like the idea, I'm just trying to figure out how to make it better.
If your decisionmaking process is consistent then an objective function exists. That's just a topological fact.
http://www.chrisstucchio.com/blog/2013/metrics_manifesto.htm...
This actually increases transparency. For example, if some party wants to declare a War on Christmas (TM) or a War on Women's Bodies (R), they actually need to put 23 x Dead Santa or -6 x Women's Freedom into the objective function.
Instead of voting for the party that wants to reverse something that directly impacts their welfare or they otherwise consider to be stupid or immoral, the electorate is forced to listen to a set of explanations of mathematical equations that try to reverse engineer how it could have been passed, and how to remove it with the least side effects. If I want to pay lower taxes I'd prefer to vote for the party that makes lower taxes for people like me a core plank of their platform, and not a set of equations purporting to show how a tax cut could feasibly be approved by the market in the current economic climate provided variables P, Q, R and S were held constant and Z was reduced by a couple of basis points.
It's quite possible that a party advocating a high weight on lower-quartile average incomes and a negative weight on birth rate wouldn't want some nutters to propose the "compulsory abortions for offspring of the unemployed" law, and get it passed, but I'm not sure a welfare function that could avoid all such unintended consequences even exists, never mind one intelligible to the average voter.
Courts should invalidate any welfare definition change that seems too directly an attempt to favor a specific policy whose main rationale comes from other measurable outcomes
That - and your "miles of road" example - seems to me to invalidate changing the welfare definition to add welfare weightings for tax levels.
As for voting for tax cuts, you probably want taxes cut as low as possible while still providing certain vital services. (I'm assuming you aren't an anarchist.) In that case, you don't merely want maximize (-1 x tax rate), you instead want to maximize (-A x tax rate - B x # of murders). Lowering taxes may or may not be the best way to achieve that goal, but you are unlikely to be able to properly judge that (no offense intended - I can't either).
An objective function and it's unintended consequences may be hard to understand for the average voter but the results of policy are even harder to understand. For example, many voters are in favor of earning assistance for firefighters and soldiers (http://www.opencongress.org/bill/110-h3997/show). One unintended consequence of the "Earnings Assistance for Heroes" policy is a gigantic bank bailout.
By contrast this system asks voters to become technocrats and carefully predict the set of variable weightings that will allow the investors to pass legislation that reduces their taxes; that is if Big Capital doesn't decide to bet on a policy proposal that uses the freed-up budget to reduce their taxes instead. Frankly, if an average voter can't determine if they want the candidate that offers them lower taxes instead of the one that offers them better policing, they're certainly not capable of choosing between the two candidates' proposed set of HDI weightings and calculation methodologies. Nor does anything in the proposed mechanism stop policy proponents sneakily attaching undesirable policies to desirable ones: in this case if you offered the banking system bailouts they'd probably bet big enough on the bill to allow it to pass even if it had subsidiary clauses mandating the execution of firefighters' firstborn. It would be trivially easy to use this process to enact laws whose consequences were essentially non-statistical, like most civil liberties...
No it doesn't. It merely asks voters to decide how many dollars a prevented murder is worth.
in this case if you offered the banking system bailouts they'd probably bet big enough on the bill to allow it to pass even if it had subsidiary clauses mandating the execution of firefighters' firstborn.
Why would Goldman do this? Why wouldn't they just take billions from all the other speculators willing to throw their money away to get bailouts and then use their winnings to bail themselves out?
Further, there is a proposed mechanism to stop policy proponents from sneakily attaching welfare-reducing policies: Goldman Sachs, George Soros, and other speculators. These folks are smart enough to spot sneaky clauses and greedy enough to rob the market manipulators pushing them.
A remaining problem might be actually measuring those values in ways that can't be gamed too easily. Attempts to objectively measure programmer productivity have generally failed at this, and Women's Freedom seems harder to measure than that.
Another problem might be that the objective function will keep changing. I guess the way to handle this is to pay off bets based on what the function was when the bet was made, and to have a fixed schedule that alternates making bets and changing the function.
I don't think it would be even more obscure. In the current system, we mingle incestuously goals and means to the point where we can't even tell the difference. Forcing a separation is going to be helpful.
The paper scuttles itself in the first paragraph.