tournament

You can rank competing hypotheses in a research vault using pairwise scientific debates with Elo ratings. The tournament approach generates automated matchups, executes structured debates, and updates a live leaderboard to surface the strongest ideas.

Elo-based debate scoring works by running structured, pairwise matchups between hypotheses and adjusting their ratings based on match outcomes. This iterative process produces a stable ranking order and logs provenance-friendly records of wins, losses, and Elo changes.

Yes, you can compare your hypothesis leaderboard with external systems using the optional federated mode. This feature maintains a separate federated leaderboard specifically designed for external comparisons while keeping your internal vault rankings isolated.

Yes, you need frontmatter metadata including name and description fields in your SKILL.md files to run a hypothesis tournament. This metadata drives the workflow and allows the system to track Elo, matches, wins, and losses across your hypothesis set.

The best way to reduce manual bias when evaluating scientific hypotheses is using structured, automated debates with Elo-based scoring. This method prioritizes underrepresented or uncertain hypotheses for matchups and relies on consistent rating calculations rather than subjective human judgment.

You can track match outcomes and provenance for hypothesis rankings through automated logging of match results, Elo rating changes, and leaderboard states. These records persist directly to your knowledge graph, ensuring a complete and traceable history of all tournament debates.