Beta feature
Battles are experimental and not fully reliable. This mode reads the current archive only, runs a hosted AI comparison, does not save results to the database, and never overrides moderator judgment or the documented feat record.
Battle runs can take a while on purpose. The engine scans both dossiers, builds lane candidate pools, checks matchup and Reddit community context, judges category sheets one by one, and then stabilizes the final verdict so it does not contradict the lane results.
In practice it is still usually faster than waiting for a full manual verdict from human scalers, even if some battles need extra time.