Third Party Validated, Predictive Markets: AI

Many markets out there on Manifold, particularly in the AI space, are resolved based upon a market maker's qualitative judgement.

The weakness in this type of market is ultimately it's a matter of the market maker's opinion, which is one singular judgement call, rather than an aggregation of multiple people's opinions (e.g. such as a poll).

This is an effort in collecting markets which use third party sources of information as a proxy for a judgement on a question, and then resolve based upon a numerical value passing a particular threshold.

A couple weakenesses in this approach 1) The map is not the territory. 2) We might extrapolate some information from data that's not really meaningful.

Strenghths in this approach: 1) Less ambiguous resolution criteria, 2) Ideally less worry about the market resolver being influenced by merely being convinced in the comments. 3) If you pay attention to a lot of these types of markets, you probably have a better picture of the ground truth of what's going on in a space, rather than a bunch of random people's opinions on the space, as those random market makers may or may not be experts.