Resolution criteria:
This market will resolve to "Yes" if, by December 31, 2028, a large language model (LLM) achieves a ranking of Diamond or higher in StarCraft II. The LLM must:
Compete in standard 1v1 matches against human opponents.
Attain a matchmaking rating (MMR) corresponding to the Diamond league or higher or otherwise demonstrate clear ability to beat Diamond league human players 50% of the time.
If these conditions are not met by the specified date, the market will resolve to "No."
Update 2025-04-19 (PST) (AI summary of creator comment): Additional Decision-Making Requirement:
The LLM must make decisions on all actions and cannot outsource decision-making to external or specialized tools built for gaming.
While the LLM may use tools to control units, it must generate its own decisions and cannot simply rely on the tool's outputs.
@ArtimisFowl good point. The model cannot outsource decisions to tools. In principle I'm fine with it using tools to control units, etc, as long as the model is making the decision on ALL actions. It's not just outputs, the model can't "consult" with specialized tools built for games either.