What about the issue of out-of-band financial rewards skewing the betting process?
For simplicity let's say there's a binary decision to be made: Does the Fed lower rates (A) or raise rates (B). Each possibility gets its own betting market, which the leaders at the Fed presumably consult in order to decide A or B.
Now if we presume that these market participants (bettors) have no financial stake other than the betting market itself, then the incentives are clear: Rational bettors will bet according to their true beliefs. This is what we want to achieve, an honest signal from the market.
However, if that presumption fails then it may become rational for a bettor to bet against their true beliefs, if that will nudge the outcome (A or B) in a direction that is favorable to them, external to the betting market. This problem is obviously most acute for wealthy bettors who may have a lot to gain or lose depending on outcome.
It seems this could work best in areas where the possibility of out-of-band incentives is low. Fed policy, on the other hand: Almost everyone has some financial stake in what they do.
A related question: Elon Musk purchased Twitter and destroyed a lot of its financial value. Was this a rational decision on his part?
Noise traders are those who trade for any reason other than profiting off their info. Adding noise traders attracts info traders, and makes prices more accurate. Manipulators are noise traders.
"Noise" might imply an unbiased signal, but there could easily be a consistent bias if wealthy participants tend to benefit more from a certain decision outcome. For example common wisdom is that fed rate cuts will tend to make the stock market go up, so most bettors with deep pockets (who very likely own stocks) will have reason to bias consistently in that direction.
Granted for many decisions we'd like to inform there probably isn't enough out-of-band benefit to make large scale manipulation a rational strategy.
I have always felt that this is one of the best criticisms of futarchy, and one that Robin never addresses when he points out that noise traders attract info traders. It's made worse by the issue that if the biased views regard the scenario where the option unfavorable to the wealthy is implemented, then futarchy won't implement this scenario, and trades on that conditional will be reverted, preventing anyone from compounding profits by correcting the bias.
idk, to me it seems like betting no on something where the wealthy bet yes, and are wrong will also correct the bias (assuming that over time the wealthy get poorer by continuing to bet yes on things that fail).
You cite Florida orange juice commodity futures, which improve on government weather forecasts, as an example of the predictive nature of the market system. Aren't there examples of betting through the system? Businesspeople put their money down to develop a cannabis farm, for instance, in light of pending policy changes around the legalization of marijuana. Couldn't the betting segments be isolated throughout the process of government action (which is never instantaneous) to evaluate or predict the positive and negative outcomes for both the populace and the supervising agency?
Thanks for the summary. I'm still not entirely clear though (and judging by the comments, neither are many others), would you mind walking through a slightly more in depth example, say maybe the Federal Reserve? IIRC, that's a nominally 'independent' agency with clear target metrics (keep inflation around 2%, minimize unemployment, maximize workforce participation), publicly available statistics already collected, and generally a lot of transparency about what options they have for policies (given the extensive new coverage whenever they even think about changing a rate). If I understood this article, that seems like a decent candidate for an agency specific test implementation, right?
Reading this made me realize that I don't understand how futarchy works.
The catch is a phrase that occurs frequently above, "conditional on". In a futures market, competitors bet on the outcome of a thing which /will have an outcome/. There is no "conditional on."
In a futarchy, you want to make a decision, so you want people to bet on different policy options, and choose the one that's the favorite. But you can't adopt all of the options, so only one of them ever gets evaluated & has an outcome to measure.
Which I think means people aren't actually voting on which option will have the best outcome; they're voting on which outcome will be most-popular.
I did. It only confirms my analysis, as far as I can tell. "When a betting market clearly estimates that a proposed policy would increase expected national welfare, that proposal becomes law." So, people bet on policy options, but you never actually learn which policy would have been best. How does that work? Does everybody who bet on a policy that wasn't chosen, lose the bet, even if that would have been the best policy? Is the government the house, and the house wins if the chosen policy lowers the metric, while the people who bet on that policy wins if it raises the metric, and everybody else just gets their bet cancelled? How do you make it so people are voting on outcomes, rather than on which policy most people will bet on?
Interestingly, Claude assured me that they're not rolled back. If they are rolled back, couldn't someone sabotage a policy to ensure it isn't chosen without being penalized?
Okay. As I wrote above, but maybe I edited it in after you wrote your reply. But it still sounds like people are just betting on which policy will be most-popular. Unless it's using the other approach which I already described in the comment you just replied to.
There are two issues: whether a policy is enacted and what the effect of the policy will be. For a conditional decision market, the latter determines the former. But if you bet that a policy will fail when everyone else thinks it will succeed, once it's adopted you can win money on your bet (unlike a pure popularity contest). On the other hand, if you bet that a policy will succeed when everyone else thinks it will fail, it won't be adopted (and thus your prediction can't be vindicated), but you're not out on the money you bet since the transaction is rolled back.
Is the bet a binary win (the metric is above X) / lose (the metric is below X), rather than a return weighted by how close you were to the metric measured on a set date?
Either way, there's is still a strong incentive to bet on a popular policy, because if the one you bet on isn't chosen, you've just wasted your time.
What about the issue of out-of-band financial rewards skewing the betting process?
For simplicity let's say there's a binary decision to be made: Does the Fed lower rates (A) or raise rates (B). Each possibility gets its own betting market, which the leaders at the Fed presumably consult in order to decide A or B.
Now if we presume that these market participants (bettors) have no financial stake other than the betting market itself, then the incentives are clear: Rational bettors will bet according to their true beliefs. This is what we want to achieve, an honest signal from the market.
However, if that presumption fails then it may become rational for a bettor to bet against their true beliefs, if that will nudge the outcome (A or B) in a direction that is favorable to them, external to the betting market. This problem is obviously most acute for wealthy bettors who may have a lot to gain or lose depending on outcome.
It seems this could work best in areas where the possibility of out-of-band incentives is low. Fed policy, on the other hand: Almost everyone has some financial stake in what they do.
A related question: Elon Musk purchased Twitter and destroyed a lot of its financial value. Was this a rational decision on his part?
Noise traders are those who trade for any reason other than profiting off their info. Adding noise traders attracts info traders, and makes prices more accurate. Manipulators are noise traders.
"Noise" might imply an unbiased signal, but there could easily be a consistent bias if wealthy participants tend to benefit more from a certain decision outcome. For example common wisdom is that fed rate cuts will tend to make the stock market go up, so most bettors with deep pockets (who very likely own stocks) will have reason to bias consistently in that direction.
Granted for many decisions we'd like to inform there probably isn't enough out-of-band benefit to make large scale manipulation a rational strategy.
I have always felt that this is one of the best criticisms of futarchy, and one that Robin never addresses when he points out that noise traders attract info traders. It's made worse by the issue that if the biased views regard the scenario where the option unfavorable to the wealthy is implemented, then futarchy won't implement this scenario, and trades on that conditional will be reverted, preventing anyone from compounding profits by correcting the bias.
idk, to me it seems like betting no on something where the wealthy bet yes, and are wrong will also correct the bias (assuming that over time the wealthy get poorer by continuing to bet yes on things that fail).
You cite Florida orange juice commodity futures, which improve on government weather forecasts, as an example of the predictive nature of the market system. Aren't there examples of betting through the system? Businesspeople put their money down to develop a cannabis farm, for instance, in light of pending policy changes around the legalization of marijuana. Couldn't the betting segments be isolated throughout the process of government action (which is never instantaneous) to evaluate or predict the positive and negative outcomes for both the populace and the supervising agency?
Thanks for the summary. I'm still not entirely clear though (and judging by the comments, neither are many others), would you mind walking through a slightly more in depth example, say maybe the Federal Reserve? IIRC, that's a nominally 'independent' agency with clear target metrics (keep inflation around 2%, minimize unemployment, maximize workforce participation), publicly available statistics already collected, and generally a lot of transparency about what options they have for policies (given the extensive new coverage whenever they even think about changing a rate). If I understood this article, that seems like a decent candidate for an agency specific test implementation, right?
Reading this made me realize that I don't understand how futarchy works.
The catch is a phrase that occurs frequently above, "conditional on". In a futures market, competitors bet on the outcome of a thing which /will have an outcome/. There is no "conditional on."
In a futarchy, you want to make a decision, so you want people to bet on different policy options, and choose the one that's the favorite. But you can't adopt all of the options, so only one of them ever gets evaluated & has an outcome to measure.
Which I think means people aren't actually voting on which option will have the best outcome; they're voting on which outcome will be most-popular.
Correct me if I'm wrong.
You could follow the first link, on "futarchy".
Also, your original motto was, "Vote values; bet beliefs." Where are the values voted on, & how does that get merged into the policy choices?
I did. It only confirms my analysis, as far as I can tell. "When a betting market clearly estimates that a proposed policy would increase expected national welfare, that proposal becomes law." So, people bet on policy options, but you never actually learn which policy would have been best. How does that work? Does everybody who bet on a policy that wasn't chosen, lose the bet, even if that would have been the best policy? Is the government the house, and the house wins if the chosen policy lowers the metric, while the people who bet on that policy wins if it raises the metric, and everybody else just gets their bet cancelled? How do you make it so people are voting on outcomes, rather than on which policy most people will bet on?
No, bets on an unchosen policy aren't losses. They are rolled back.
Interestingly, Claude assured me that they're not rolled back. If they are rolled back, couldn't someone sabotage a policy to ensure it isn't chosen without being penalized?
Claude is wrong.
You could also try to sabotage a publicly-traded company after short-selling it.
Okay. As I wrote above, but maybe I edited it in after you wrote your reply. But it still sounds like people are just betting on which policy will be most-popular. Unless it's using the other approach which I already described in the comment you just replied to.
There are two issues: whether a policy is enacted and what the effect of the policy will be. For a conditional decision market, the latter determines the former. But if you bet that a policy will fail when everyone else thinks it will succeed, once it's adopted you can win money on your bet (unlike a pure popularity contest). On the other hand, if you bet that a policy will succeed when everyone else thinks it will fail, it won't be adopted (and thus your prediction can't be vindicated), but you're not out on the money you bet since the transaction is rolled back.
Is the bet a binary win (the metric is above X) / lose (the metric is below X), rather than a return weighted by how close you were to the metric measured on a set date?
Either way, there's is still a strong incentive to bet on a popular policy, because if the one you bet on isn't chosen, you've just wasted your time.
You can use the same amount of money to bet on ALL of a set of mutually exclusive options.
The ones I'm familiar with are binary. Maybe someone could make a scalar one.