Prediction markets (a.k.a. putting your money where your mouth is) are popular among rationalists, but kinda unpopular with governments. It is too easy to classify them as gambling. But if we remove the money, people have less incentive to get things right.
But there is also a non-monetary thing people deeply care about: prestige. This can’t be used on a website with anonymous users, but could be used with famous people who care about what others think about their opinions: journalists or analytics. So here is a plan:
A newspaper (existing or a new one) could make a “Predictions” section, where experts would be asked to assign probabilities to various outcomes. If they guessed correctly, they would gain points; if they guessed incorrectly, they would lose points. The points would influence their position on the page: Opinions of predictors with more points would be at the top of the page (with larger font); opinions of predictors with less points would be at the bottom (with smaller font). Everyone starts with some given number of points; if someone drops below zero, they are removed from this newspaper section, forever. And a new predictor, with given starting number of points is selected instead of them.
There is no money directly involved here, so the government can’t object against gambling. But still, the predictors will care about their outcomes (because those who won’t, will be gradually removed from the system). And the readers will see the prediction market in action, although they will not be allowed to bet there. Or perhaps there could be a possibility for them to give an internet vote, too, and see how good they are compared with the official predictors. Actually, the new predictors could be picked among the most successful readers.
This whole thing is inspired by the fact that some people also care about virtual money; e.g. by gold in World of Warcraft. As long as you have the virtual currency, you can have a prediction market running on that virtual currency, and it’s like having a casino in the World of Warcraft world: not quite real, therefore not regulated the same way.
But still, the predictors will care about their outcomes (because those who won’t, will be gradually removed from the system).
So you are suggesting a system that relies on the opinions of people who got selected because they really want to see their names at the top of the page in big font?
If the only way to see their names at the top of the page in big font is to provide correct predictions… why not?
The classical prediction market relies on opinions of people who got selected because they really wanted to make money on the prediction market. What’s the big difference?
Okay… I can imagine that if someone’s goal is to bring attention to themselves, they might make correct predictions to get to the top, and then intentionally make shocking (incorrect) predictions to bring even more attention to them. Kinda like people with too much karma sometimes start trolling, because, why not.
In particular, making predictions is not costless. To consistently produce good forecasts you need to commit resources to the task—off-the-cuff opinions are probably not going to make it. Why should serious people commit resources, including their valuable time, if the only benefit they get is seeing their name in big letters on top of a long list?
Well, this is something that can be tested experimentally. It could be statistically tested whether the results of the top predictors resemble random noise.
Some people spend incredible amounts of time on internet, reading about stuff that interests them. I can imagine they could make good predictions in their area. (And there should be a “no vote” option for questions outside of their area.)
Wikipedia exists, despite it doesn’t pay its contributors, unlike other encyclopedias. And there is some good stuff there. Also bad stuff… but that’s what the competition between predictors could fix.
There is probably a limit on how difficult things can be predicted. But it could be higher than we imagine. Especially if the predictions become popular, so for many topics there would be predictors whose hobby it is.
There are some technical details to solve, e.g. whether the predictor’s prestige will be global, or topic-dependent. (To prevent people from systematically giving 10 great prediction in topic X, and then 1 bad but very visible in topic Y.) But that’s like having multiple StackExchange accounts.
Pundits already make predictions all the time, but no one scores them. Even ones that where the outcome is very clear, like finding a nuclear program in Iraq or Apple making a TV.
So I think it is important to identify what the problem is and make sure you are actually addressing it. Setting up special venues has some advantages, like making sure that the questions are precise and judgeable, and attracting the right kind of people. Prestige for pundits is basically venue. Moving up and down inside a venue is a pretty small scale of change. I suppose that it is possible that if your venue has lots of churn, then the stakes would be higher.
One venue for realish money celebrity predictions is Long Bets, but while the emphasis is on accountability, it sure isn’t on building a track record.
Also: does any government other than USA object to prediction markets? I suppose that ipredict’s bankroll limitations indicate an objection by NZ, but I’m skeptical that this is a significant limiting factor.
There is a site called pundittracker.com that tracks the predictions of pundits. Personally, I don’t think it’s all that interesting, in large part because the most concrete testable predictions pundits make are in areas not particularly interesting to me. But at least someone is trying.
An idea: prestige-based prediction market.
Prediction markets (a.k.a. putting your money where your mouth is) are popular among rationalists, but kinda unpopular with governments. It is too easy to classify them as gambling. But if we remove the money, people have less incentive to get things right.
But there is also a non-monetary thing people deeply care about: prestige. This can’t be used on a website with anonymous users, but could be used with famous people who care about what others think about their opinions: journalists or analytics. So here is a plan:
A newspaper (existing or a new one) could make a “Predictions” section, where experts would be asked to assign probabilities to various outcomes. If they guessed correctly, they would gain points; if they guessed incorrectly, they would lose points. The points would influence their position on the page: Opinions of predictors with more points would be at the top of the page (with larger font); opinions of predictors with less points would be at the bottom (with smaller font). Everyone starts with some given number of points; if someone drops below zero, they are removed from this newspaper section, forever. And a new predictor, with given starting number of points is selected instead of them.
There is no money directly involved here, so the government can’t object against gambling. But still, the predictors will care about their outcomes (because those who won’t, will be gradually removed from the system). And the readers will see the prediction market in action, although they will not be allowed to bet there. Or perhaps there could be a possibility for them to give an internet vote, too, and see how good they are compared with the official predictors. Actually, the new predictors could be picked among the most successful readers.
This whole thing is inspired by the fact that some people also care about virtual money; e.g. by gold in World of Warcraft. As long as you have the virtual currency, you can have a prediction market running on that virtual currency, and it’s like having a casino in the World of Warcraft world: not quite real, therefore not regulated the same way.
So you are suggesting a system that relies on the opinions of people who got selected because they really want to see their names at the top of the page in big font?
If the only way to see their names at the top of the page in big font is to provide correct predictions… why not?
The classical prediction market relies on opinions of people who got selected because they really wanted to make money on the prediction market. What’s the big difference?
Okay… I can imagine that if someone’s goal is to bring attention to themselves, they might make correct predictions to get to the top, and then intentionally make shocking (incorrect) predictions to bring even more attention to them. Kinda like people with too much karma sometimes start trolling, because, why not.
Money is a MUCH better motivator.
In particular, making predictions is not costless. To consistently produce good forecasts you need to commit resources to the task—off-the-cuff opinions are probably not going to make it. Why should serious people commit resources, including their valuable time, if the only benefit they get is seeing their name in big letters on top of a long list?
Well, this is something that can be tested experimentally. It could be statistically tested whether the results of the top predictors resemble random noise.
Some people spend incredible amounts of time on internet, reading about stuff that interests them. I can imagine they could make good predictions in their area. (And there should be a “no vote” option for questions outside of their area.)
Wikipedia exists, despite it doesn’t pay its contributors, unlike other encyclopedias. And there is some good stuff there. Also bad stuff… but that’s what the competition between predictors could fix.
There is probably a limit on how difficult things can be predicted. But it could be higher than we imagine. Especially if the predictions become popular, so for many topics there would be predictors whose hobby it is.
There are some technical details to solve, e.g. whether the predictor’s prestige will be global, or topic-dependent. (To prevent people from systematically giving 10 great prediction in topic X, and then 1 bad but very visible in topic Y.) But that’s like having multiple StackExchange accounts.
Pundits already make predictions all the time, but no one scores them. Even ones that where the outcome is very clear, like finding a nuclear program in Iraq or Apple making a TV.
So I think it is important to identify what the problem is and make sure you are actually addressing it. Setting up special venues has some advantages, like making sure that the questions are precise and judgeable, and attracting the right kind of people. Prestige for pundits is basically venue. Moving up and down inside a venue is a pretty small scale of change. I suppose that it is possible that if your venue has lots of churn, then the stakes would be higher.
One venue for realish money celebrity predictions is Long Bets, but while the emphasis is on accountability, it sure isn’t on building a track record.
Also: does any government other than USA object to prediction markets? I suppose that ipredict’s bankroll limitations indicate an objection by NZ, but I’m skeptical that this is a significant limiting factor.
There is a site called pundittracker.com that tracks the predictions of pundits. Personally, I don’t think it’s all that interesting, in large part because the most concrete testable predictions pundits make are in areas not particularly interesting to me. But at least someone is trying.