Scott Alexander

The new era of prediction markets has provided charming additions to the language, including rules cuck, someone who loses an otherwise prescient bet based on technicalities of the resolution criteria.

Resolution criteria are the small print explaining what counts as the prediction market topic happening.

For example, in the Khomeini example above, Khomeini qualifies as being out of power in quotes if, quote, he resigns, is detained, or otherwise loses his position or is prevented from fulfilling his duties as Supreme Leader of Iran within this market's time frame.

The primary resolution source for this market will be a consensus of credible reporting.

End quote.

You can imagine ways this definition departs from an exact commonsensical concept of out of power.

For example, if Khomeini gets stuck in an elevator for half an hour and misses a key meeting, does this count as him being prevented from fulfilling his duties?

With thousands of markets getting resolved per month, chances are high that at least one will hinge upon one of these edge cases.

Kaoshi resolves markets by having a staff member with good judgment decide whether or not the situation satisfies the resolution criteria.

Polymarket resolves markets by... Oh man, how long do you have?

There's a cryptocurrency called UMA.

UMA owners can stake it to vote on Polymarket resolutions in an associated contract called the UMA Oracle.

Voters on the losing side get their cryptocurrency confiscated and given to the winners.

This creates a Keynesian beauty contest.

that is, a situation where everyone tries to vote for the winning side.

The most natural shelling point is the side which is actually correct.

If someone tries to attack the oracle by buying lots of UMA and voting for the wrong side, this incentivizes bystanders to come in and defend the oracle by voting for the right side, since, conditional on there being common knowledge that everyone will do this, that means they get free money at the attacker's expense.

But also UMA goes up if people trust the oracle to plan and plan to use it more often, and goes down if people think the oracle is useless and may soon get replaced by other systems.

So regardless of their other incentives, everyone who owns the currency has an incentive to vote for the true answer so people keep trusting the oracle.

This system does work most of the time, but tends towards so-called oracle drama, where a seemingly prosaic resolution might lie at the end of a thrilling story of attacks, counterattacks, and escalations.