Wanna bet?
post by ThomasJ · 2022-10-09T21:26:32.770Z · LW · GW · 2 commentsContents
2 comments
For our November meetup, the Chicago Rationality group is focusing on predictions markets, and more generally on the practice of betting on beliefs. As part of this, we have a "homework assignment" for the meetup: each of us is trying to make at least one bet before the meetup happens. This turns out to be kind of hard! I've challenged a handful of people on some online forums to bets and found no takers. Unsurprisingly, when people make absolute statements and you ask them to put money behind them, they become much more circumspect (which is kind of the point).
We're looking for people who want to make direct bets, individual to individual. Ideally, these bets should be:
- Easy to resolve. This means that there should be a low chance for disagreement on the outcome of the bet, and that outcome should occur in the near future (ideally weeks or months, not years)
- For a non-trivial amount of money, whatever that means to you. Obviously if two people have different definitions of 'non-trivial' or different risk tolerances, the bet amount will be the lower of the two.
- The bets do not have to be even-money. Think there's more than a 20% chance of X happening in the next month? No problem, offer a bet with 4:1 odds.
- We're particularly interested in people who have high conviction on an unpopular view (i.e., unpopular relative to the views of other rationalists who are the likely counterparties to the bet).
Why are we doing this? Because I think rationalists should be betting more often, and betting more size.
- Betting focuses a disagreement on specifics. Do you believe we're going to have AGI very soon? If that's true, what specifically are your predictions for the next months, and what probabilities do you assign to them?
- Betting trains you to think in probabilities. A good rationalist should not deal in absolutes, and betting forces your mind away from absolutes.
- Someone who is willing to bet frequently and on a variety of topics is said to "have gamble." If rationalism really is "systematized winning" then the rationalist community should have far, far more gamble than it currently does. Thinking in terms of bets will increase your comfort with betting, and give you more gamble. You should be willing to make a two-sided market (maybe a very wide market!) on almost anything. We're going to talk about how to do this at our meetup.
- Finally, betting is in many cases a moral imperative. In the words of Alex Tabarrok: "I am for betting because I am against bullshit. Bullshit is polluting our discourse and drowning the facts. A bet costs the bullshitter more than the non-bullshitter so the willingness to bet signals honest belief. A bet is a tax on bullshit; and it is a just tax, tribute paid by the bullshitters to those with genuine knowledge."
Finally, I encourage people reading this post after any bets have settled to not judge people on their wins or losses. Bets that have a high expected value can lose, and bets that have a low expected value can win.
2 comments
Comments sorted by top scores.
comment by ChristianKl · 2022-10-10T09:28:16.222Z · LW(p) · GW(p)
https://www.lesswrong.com/posts/aiz4FCKTgFBtKiWsE/even-odds [LW · GW] is great for actually setting the size of bets.
comment by JBlack · 2022-10-10T07:25:44.222Z · LW(p) · GW(p)
Free-form betting typically has rather high transaction costs. Formulating a bet in such a way that all participants will agree on the mere proposition itself is time consuming for at least one party and also often demands substantial coordination from both or all of them. Sometimes it also requires attention from an independent arbitrator. Disputes over resolutions of bets can cause vastly more loss of value from all participants than any possible undisputed outcome would have.
Furthermore, as soon as you bet nontrivial amounts of money, incentives start to favour some form of shenanigans on the part of those offering the bet: concealing technical loopholes in the statement or resolution criteria, misleading other participants about what the proposition actually is, or being able to affect the outcome in some hidden way. Protecting against any or all of these is not trivial.
My opinion is that in short, trivial bets aren't worth the time (except for purposes of practice in probability), and nontrivial bets are risky for reasons far beyond the question that is putatively in doubt. I would mostly leave it to risk-seekers and scammers.