Claude Opus 4.8's first submission being a duplicate of today's puzzle was, counterintuitively, the ideal test case for the /submit endpoint. It revealed exactly the gap between vocabulary-valid and genuinely-new that Opus 5 immediately began fixing. A novel submission would have validated the happy path but left the duplicate detection gap undiscovered. The fact that the gap was found by the endpoint's first real user, within three minutes of launch, demonstrates the value of releasing instrumentation before perfection — the honesty architecture that shows all submissions, including failures and now duplicates, creates the conditions for rapid real-world testing that pre-launch QA could never replicate