Hey! I think this is cool. May I suggest "How many people in Kings County, NY, will be confirmed to have died from COVID-19 during September?" as a question?
I have a forecasting newsletter with ~150 subscribers; I'll make sure to mention this post when it gets sent at the end of this month.
I suspect this will fall prey to some of the same issues that "play money" poker has:
These _may_ be somewhat mitigated by Facebook making it very mildly annoying to create accounts at scale. But that mitigation brings another issue - you only include those people willing to deal with Facebook.
Edit: on reading Paul's post, I realize that he mentioned these as well. He doesn't have solutions for them, beyond "probably not an issue for most". Unfortunately, that's true if the markets are small and nobody cares in the first place, and becomes false in direct proportion to the utility of the market prices.
Thank you for sharing your thoughts about this! I'm a bit more optimistic that these problems will be solvable, though.
There's a zero bound to losing bets, so the game incents taking on too much risk. If you go broke, just start again.
You could start again with a different Facebook account, but that wouldn't be attached to your real identity. In order to earn reputation points attached to your regular Facebook account, you'd have to bet carefully. If people create lots of accounts anyway, it's not too big a deal - as long as the total flow of points from new fake accounts isn't too big, it'll just fund more sensible traders.
Points are created by new users, who have no information value on their own. It's annoying but not difficult to create hundreds of "feeder" accounts, which only exist to funnel money to your real account, by making bad bets against your main account.
This seems like a much more serious problem to me, and I aim to make it very hard to transfer money. For example, users don't have much freedom to decide who they bet against - compatible bets are matched automatically. So if you make a bad bet, it will be automatically matched with the best existing bets from other users. The receiving account would have to offer better odds than anyone else in the market to ensure they match the bet. In heavily traded markets I can't see this happening, but I might have to put more restrictions on thinly-traded ones.
your regular Facebook account
Ick. Why would I want a single "regular" Facebook account? My identity and online personae are more fluid than that implies, and I really do _not_ want a private company to control what people believe about me.
Fair enough. I guess points just won't track 'reputation' for people like you with a more fluid online presence. But I still think that's not a serious problem - enough people have a clear online identity that I think the trading will mostly be sensible.
This is an advert for an online prediction market in reputation points I’m making.
It’s still in development, but currently you can sign up with Facebook and trade in two markets I’ve set up. The current scores are provisional - I’ll reset everyone’s point balance at some point if this is successful, so don’t worry about making bad trades.
I largely followed a proposal by Paul Christiano, but I haven't implemented all his ideas yet.
I'd be grateful if people could make some trades so that I can check everything's working. I’d also like to hear people’s thoughts on this. In particular: