search

manifoldmarkets / manifold

Manifold Markets: A market for every question
https://manifold.markets
MIT License
418 stars 157 forks source link

Feature Request: Arbitrators #538

Open hugbubby opened 2 years ago

hugbubby commented 2 years ago

The current answer for "What happens if a market creator resolves a market incorrectly, or doesn't resolve it at all?" is:

Nothing. The idea is for Manifold Markets to function with similar freedom and versatility to a Twitter poll, but with more accurate results due to the dynamics of prediction markets. Individual market resolution is not enforced by the site, so if you don't trust a certain user to judge their markets fairly, you probably shouldn't participate in their markets.

That's good as far as it goes, but the potential for incorrect resolution is still anti-fun. It creates barriers for traders; now they have to think more about the person creating the market, instead of the probably-funner question of the event. It creates barriers for market-makers; if they don't somehow convince the community they're trustworthy people, they might be left with a market that doesn't actually measure the event that they want. It also deflates community excitement when, especially on questions with multi-year timelines, the market-maker just forgets that they made a market and the community ends up wasting money/time on markets that never pay out.

So, one solution that wouldn't necessarily add to the long-term workload by the Manifold devs would be to give users the ability to appoint community arbitrators. Market-makers could pay others to either resolve questions for them outright, or be a backup resolver in case the original market-maker goes MIA. Those arbitrators could build up and lend that community trust over time in the domain of resolution without having to actually make the markets themselves.

Edit: We could even have a system of using secondary markets ala https://manifold.markets/jfjurchen/is-j-f-jurchen-honorable

ahalekelly commented 2 years ago

I was just thinking about this too. The market creator could nominate people to vote on the resolution, with the number of votes required for resolution as a parameter, because there's a tradeoff between resolution security and speed. Was inspired by these markets https://manifold.markets/jfjurchen/is-j-f-jurchen-honorable https://manifold.markets/ahalekelly/is-j-f-jurchen-metahonorable

akrolsmir commented 2 years ago

We've talked about this kind of thing before, using markets or juries or judges to back up the arbitration of a market. In practice, I think autoresolve to PROB would also handle most of this use case, though @sipec might say we're not moving forward with that in the near future.

Overall, arbitrator delegation is one of those "nice to have, but not high priority" feature areas; it doesn't feel like it'd solve an especially pressing pain point, at least not at our current scale.

hugbubby commented 2 years ago

Overall, arbitrator delegation is one of those "nice to have, but not high priority" feature areas; it doesn't feel like it'd solve an especially pressing pain point, at least not at our current scale.

Beware this getting out of hand as the environment of trust slowly degrades. If you do actually grow this will be an untenable situation eventually and you will have to address it. I suspect it's a problem already and the longer term trades on many already created markets just haven't resolved inconclusively yet.