An interesting question: is there any way to turn a karma system into a prediction market?
The obvious way to me is to weight a person’s voting influence by how well their votes track the majority, but that just leads to groupthink.
The key to prediction markets, as far as I can tell, is that predictions unambiguously come true or false and so the correctness of a prediction-share can be judged without reference to the share-price (which is determined by everyone else in what could be a bubble even) - but there is no similar outside objective check on LW postings or comments, is there?
I’d love to do a real money prediction market. Unfortunately western governments seek to protect their citizens from the financial consequences of being wrong (except in state sponsored lotteries… those are okay), and the regulatory costs (financial plus the psychic pain of navigating bureaucracy) of setting one up are higher than the payback I expect from the exercise.
The UBC is able to do a non-profit elections prediction market, and it generally does better than the average of the top 5 pollsters.
The popular vote market is you pay $1 for 1 share of CON, LIB, NDP, Green, Other, and you can trade shares like a stockmarket.
The ending payout is $1 * % of popular vote that group gets.
There are other markets such as a seat market, and a majority market.
The majority market pays 50⁄50 if no majority is reached, and 100⁄0 otherwise, which makes it pretty awkward in some respects. Generally predicting a minority government the most profitable action is to try and trade for shares of the loser. This is probably the main reason its restricted to the two parties with a chance of winning one if it were the same 5 way system, trading LIB and CON for GREEN, OTHER and NDP to exploit a minority government would probably bias the results. In this case in a minority the payout would be 20/20/20/20/20, but many traders would be willing to practically throw away shares of GREEN, OTHER and NDP because they “know” those parties have a 0% chance of winning a majority. This leads to artificial devaluation and bad prediction information.
By trading 1 share of CON for 5 GREEN and 5 OTHER, you just made 10 times the money in a minority government, and that’s the payoff you’re looking for instead of saying that you think the combined chances of Green and Other winning a majority is 1/6th that of the conservatives winning.
Of course they still have this problem with Liberals and Conservatives where trading out of a party at a favorable rate might just be betting minority.
I think the problem with a prediction market is you need a payout mechanism, that values the shares at the close of business, for elections there is a reasonable structure.
For situations where there isn’t a clear solution or termination that gets much more complicated.
An interesting question: is there any way to turn a karma system into a prediction market?
The obvious way to me is to weight a person’s voting influence by how well their votes track the majority, but that just leads to groupthink.
The key to prediction markets, as far as I can tell, is that predictions unambiguously come true or false and so the correctness of a prediction-share can be judged without reference to the share-price (which is determined by everyone else in what could be a bubble even) - but there is no similar outside objective check on LW postings or comments, is there?
I’d love to do a real money prediction market. Unfortunately western governments seek to protect their citizens from the financial consequences of being wrong (except in state sponsored lotteries… those are okay), and the regulatory costs (financial plus the psychic pain of navigating bureaucracy) of setting one up are higher than the payback I expect from the exercise.
The UBC is able to do a non-profit elections prediction market, and it generally does better than the average of the top 5 pollsters.
The popular vote market is you pay $1 for 1 share of CON, LIB, NDP, Green, Other, and you can trade shares like a stockmarket.
The ending payout is $1 * % of popular vote that group gets.
There are other markets such as a seat market, and a majority market.
The majority market pays 50⁄50 if no majority is reached, and 100⁄0 otherwise, which makes it pretty awkward in some respects. Generally predicting a minority government the most profitable action is to try and trade for shares of the loser. This is probably the main reason its restricted to the two parties with a chance of winning one if it were the same 5 way system, trading LIB and CON for GREEN, OTHER and NDP to exploit a minority government would probably bias the results. In this case in a minority the payout would be 20/20/20/20/20, but many traders would be willing to practically throw away shares of GREEN, OTHER and NDP because they “know” those parties have a 0% chance of winning a majority. This leads to artificial devaluation and bad prediction information.
By trading 1 share of CON for 5 GREEN and 5 OTHER, you just made 10 times the money in a minority government, and that’s the payoff you’re looking for instead of saying that you think the combined chances of Green and Other winning a majority is 1/6th that of the conservatives winning.
Of course they still have this problem with Liberals and Conservatives where trading out of a party at a favorable rate might just be betting minority.
I think the problem with a prediction market is you need a payout mechanism, that values the shares at the close of business, for elections there is a reasonable structure.
For situations where there isn’t a clear solution or termination that gets much more complicated.