Rubric Library

The standards we score against.

Every run uses one rubric, published before its scores. Each rubric is scoped to a domain and tagged with the argument forms it evaluates. The five rubrics below are the active library — the matcher picks one for every submission, and you can read each one’s criteria below.


The matching engine assigns one of these five rubrics to each submitted debate. If none clears the 0.75 match threshold, the run is flagged provisional and held for human review. Anyone can propose a new rubric — that workflow ships in a later release.