I’ve been fiddling around with Manifold Markets lately and wondering about the possibility of using prediction markets as commitment devices. I haven’t actually done this yet but I’d like to give it a go.
The way I see it, it’d work like this:
Say I want to commit to writing a book by some deadline.
I start a prediction market that resolves ‘yes’ if I finish it in time, and I put all my ante money on ‘yes’ so that I get my money back if I succeed and lose it if I don’t, with the amount being somewhat proportional to how much I wanna get it done. (So far a pretty standard commitment device).
Then other people can bet against me, standing to gain the money I’ve put at stake if I fail to complete it, and increasing the money I stand to gain if I succeed.
The biggest benefit of this kind of commitment device that I can see is that the rewards/motivation scale in proportion to how much I need it! If it’s likely that I’ll get the book finished anyway, not many people are going to bet against me, so the market ends up being a small amount of motivation that I didn’t really need. But the more likely I am to fail, the more people bet against me, the more reward money I get if I succeed!
- If you want to read my book, you can pay me to write it by betting that I won’t, and you’re guaranteed to either get what you paid for or get your money back plus interest
- Easily adaptable to team commitments - get everyone on a team to place a large bet in favour of your collective success, and they’ve suddenly all got shared stakes in achieving a clear outcome with a clear deadline, something many managers can only dream of (alternatively just agree to share the profits)
- If partial success is possible (maybe I want to finish as much of the book as I can), resolving the market to a clearly defined percentage of success maintains the incentive to complete as much of the project as possible even if 100% success isn’t feasible
- Only works for public commitments
- Potentially allows people to profit by sabotaging me?
- There still doesn’t exist an easy-to-use, real-money, anyone-can-make-a-market prediction market platform - best you can do right now is play money on Manifold or similar
- Requires people to have some knowledge of how likely you are to succeed - which, if you want people outside your immediate social circle to bet, requires you to post updates of your progress or something similar.
(This one’s possibly more of a feature than a bug - accountability can be a good thing!)
- Requires bettors to be well-calibrated - if they overestimate your chances, you won’t get as much motivation as you might need. If they underestimate you, you might get discouraged? Or you might just make more money…
I feel like this one shouldn’t be much of an issue, especially if you keep making more commitments - it’ll soon become clear how reliable you are.
Another way of thinking of this is as a form of both epistemic and disciplinary rigour - sort of a combo of the commits.to principle of making a note whenever you say you’ll do something and the principle of being willing to bet on your beliefs, committing using a prediction market is sort of a way of betting on your own reliability.
This could also be a great way to choose between multiple projects: if I make 10 markets for potential projects conditional on my actually attempting them, there’ll be some kind of spread between easy-to-finish projects with lower rewards and harder-to-finish ones with higher rewards, letting me choose one that’s easier, or more difficult, or one where’s it’s genuinely ambiguous whether I’d succeed if I tried.
I’ve noticed one person make a personal will-I-achieve-this goal on Manifold, but the initial probability was set to 50%, meaning the market maker’s success lost them money!
Has anyone else tried this, or would anyone be willing to?