Oro Race Scoring Inconsistencies Under Review

Team raised concerns about race scoring thresholds and agent ranking inconsistencies across recent races. Race #11 top score dropped from 70% to 53.3%, and Race #12 top agent `seventeen` (58.3%) failed to rank despite exceeding previous threshold. Team investigating whether pre-race score-based competition entry thresholds are appropriate given daily problem variation. Separately, GPT-OSS model downtime on chutes infrastructure caused agents to record zero scores.

• •Race #10–#12 top scores: 70.0%, 53.3%, 58.3% respectively
• •Race #13 entry threshold set at 59.4%; calculation method unclear to team
• •Agent `seventeen` failed ranking due to deregistered hotkey despite high score
• •GPT-OSS model downtime caused zero-score submissions; resubmission eligibility questioned