Bad Problems Don't Stop Being Bad Because Somebody's Wrong About Fault Analysis

Linch’s post on LessWrong highlights a recurring logical fallacy in online discourse: treating a descriptive explanation for why a problem occurred as if it answers the normative question of whether the problem is acceptable. Three key examples illustrate the pattern. First, a complaint about a misleading headline is deflected by noting that writers don't write headlines—while ignoring the misleading effect. Second, a friend's concern about a prosaic safety problem at a major AI company going unfixed for months is met with a detailed account of organizational restrictions, implying the problem was correctly prioritized away. Third, criticism of a cruise ship releasing hantavirus carriers onto public airplanes is countered by pointing out that the WHO lacks legal quarantine authority—yet the danger to public health remains regardless.

The core argument is that explanation is not exoneration. Even if the party being blamed couldn't have acted differently, the underlying bad outcome (misleading readers, unaddressed AI risk, potential pandemic) still exists and still needs fixing. Linch notes that this fallacy is especially pernicious in AI safety, where discussions often get bogged down in ‘why the company’s incentives made it impossible’ rather than asking whether the safety measures themselves are adequate. The post cautions that ‘ought implies can’ is only a valid rebuttal when the problem is genuinely impossible to solve, not merely inconvenient. For tech professionals, this is a reminder to keep the focus on outcomes, not organizational excuses.