Pretend you are creating a new prediction markets platform of some type that you are responsible for. If you had to pick one direction, what do you think would be the best way to combat the, "Alts," problem, assuming you are using a points-based, non-monetary reputation system, and that Alts are indeed considered to be a problem on some level.
Assume that your task is to come up with a Market Design Change way to combat alt accounts (assume other methods such as Clear Rule Communication, Penalties, Bans are already being solved elsewhere).
Relating to this question below, it seems that the most popular way to address the Alts problem :
Granted, another ongoing poll right now does state that the alts problem is not a fundamental problem, however a secondary selection is that there is an erosion of trust that can occur, so this exercise / poll is valid.
1,000New users must be invited by existing users who vouch for the new user up to a certain manna amount. The user who invited a new user is ultimately responsible for the debt if the new user defaults. See https://www.ifaamas.org/Proceedings/aamas2012/papers/Z3_18.pdf and https://messymatters.com/yootles/ @dreev
@PatrickDelaney Yes that seems similar. And apparently BlueSky holds people responsible for inviting spammers in some fashion. Here the amount of trust is quantitative and denominated in manna/points. New users can't spend more than the people who invited them trust them up to.
@PatrickDelaney @dreev Also, this is an earlier paper on the same concept. If you want a copy and can't find a free copy let me know: https://www.researchgate.net/publication/220853186_Mechanism_Design_on_Trust_Networks
@DavidPennock I am signed up for BlueSky, having signed up through their website with no invitation and there appears to be no accountability. That being said maybe I'm part of a small randomized amount of whitelisted website sign ups...?
I would design it using a portfolio margin system, with forced liquidation, and with the ability to purchase options contracts to hedge so that temporary drops don't trigger liquidations.
This likely sounds complicated if you're not used to it, but could be made intuitive by making risk metrics visible to people. And options could have messaging like "Pay $X now for insurance to free up $Y of capital to bet elsewhere".
So there would be no huge negative balances unless there were a bug in the margining system that caused an untracked liability.
@Mira Wouldn't that just encourage a high number of alt accounts that are below the portfolio margin threshold for forced liquidation? Also, how would the options market get priced accurately? It sounds like without really tight option pricing, it's a merely a risk-free way to gently part non-sophisticated investors away from their money by those who understand arbitrage or have time to figure out how to exploit underlying mechanisms. Another thing that comes to mind is...*when* is pricing occurring? Seems like there could be a time-based exploit as well, if for example options are only priced once every 3 hours or something, it just opens up further exploits. It does sound complicated, frankly it sounds like a way to allow sophisticated users to siphon points out of the system (which gets into the, "eroding trust," problem. That being said, none of what I am saying here precludes building an, "advanced game," and a, "non advanced game," where the advanced game is specifically designed to essentially kick out non sophisticated players (an idea which was touched on at one of the Manifest talks). ...I also did say above that penalties and bans are to be assumed to be, "solved," so perhaps you were just thinking, "well if someone did that type of attack they would be banned."
@PatrickDelaney Small markets and fresh accounts with low balances wouldn't be eligible for anything but risk-free margin. Mid-sized markets would allow writing but not purchasing options(possibly using language like "get paid to create a limit order"), up to 2x leverage, and a house bot would handle moment-to-moment trading using those commitments as a backstop. The largest markets might have enough volume for full options and be eligible for increased leverage. You would get more permissive margining and leverage by concentrating everything onto one account; splitting it across would be more restrictive.
I have low confidence that anybody on Manifold except me could implement such a system robustly. But this mechanism would be what the system is designed around.
Actually, your market description here says "non-monetary". I'm not sure if Manifold's "mana" would be disqualified as monetary, so it's possible my suggestion would be too close to a monetary system for what you want.
@Mira My definition of non monetary is to distinguish reputation based platforms such as Metaculus or Manifold from those which provide an actual monetary payout such as as Kalshi or Polymarket.