This is probably premature and extremely high-level, but hey a man can dream
I was thinking that it would be cool to come up with a standard for a permissionless covenant-based prediction market.
Anyone should be able to post a new market specifying a question (e.g. “will France win the World Cup”) and an expiry date. This should be contained in a covenant state, and possibly replicated in plain text in the payload field so that all participants are immediately aware of the new market.
Anyone should be able to mint new tokens against any existing market. The minting operation enforces that one always mints an equal number of YES and NO tokens, paying half the winning payout for each pair, ensuring the market solvability.
The tokens would follow the KCC standard and be immediately tradeable on the DEXes that support the standard.
The market is then resolved by voting. After the defined maturity, all token holders vote for the true outcome during a given time window, after which the holders of the winning side can claim the full payout.
Since there’s always an equal number of YES and NO tokens in circulation, if at least one token holder on the losing side votes honestly (while everyone else votes opportunistically), then the market resolves to its “true” outcome. This is not perfect, but no system is after all, and the resolution risk is simply part of the game.
That’s it, of course that requires finalized KCC and DEX standards and maybe some extra covenant magic, I just wanted to throw it out there to test the community interest!
I wonder if there could be a seperate reward/penatly stake for honest voting, in addition to the bet stakes. And then to slash if you provide an answer which is not the majority vote, or even when not providing an answer.
The problem ofc is the correlation between the stake you have in the bet and your voting power, I guess this is an inherit, unavoidable flaw(?).
Yes, one can imagine several mechanisms. I tend to prefer rewards over penalties/slashing. For example, the losing tokens could enter a lottery where a random holder gets a big payout, but only if they voted honestly (=same as the majority).
Also, you can make it so that voting is necessary to get any payout (so you can expect that almost all the winning side votes, while some losers will vote too in order to participate in the lottery).
Of course it’s not completely immune to manipulation but I would expect that at least for liquid markets and noncontroversial bets, it would generally converge to the truth (sometimes much better than centralized solutions which have recently failed spectacularly in a few cases).
Hi, for those interested in diving a bit deeper, here’s a detailed analysis of the protocol, explaining why it’s reasonable to expect it to resolve to the truth under some mild conditions (and a few limitations).
i think this is a PHENOMENAL IDEA!!! I am working on an app which juxtaposes official inflation data (free to use) with users ability to vote on their lived-inflation, so far more simple, but perhaps can be used as an MVP for your idea?
Glad you liked it! However this is a mechanism for a binary prediction market, only yes/no questions are supported, not predictions on a market variable such as inflation (you can still use threshold questions though such as “will the May CPI be higher than X”).
Keep in mind that the voting here is not a way to aggregate opinions, but rather a means to bring onchain an objective truth, so the question should ideally be as little ambiguous as possible.
Ok I see, thanks for clarifying so exciting and so desperately needed because at present the line between fact and fiction is so severely blurred that I dont think anyone can tell the difference. For instance: there was a rapported drones spotted over CPH airport. Videos and images were released. A month later the news reported that that was wrong, it was planes. Videos and images released supporting that story too. So now, whats the right story and whats the wrong story?
I’ve slightly amended the protocol, now it burns uncollected winning rewards instead of reallocating them.
Thanks to that, we can now prove a much more general result that also works when people hold both token types and are free to vote however they like with each token they hold.
The protocol acts as a truth amplifier. In extreme summary, the result states that if participation is not too concentrated and people believe that truth is slightly more likely to win, then they all end up voting for the truth, and the market resolves to its intended result without a central authority.
