Simultaneously Right and Wrong

Scott's first LessWrong post (2009). Uses the Berglas & Jones self-handicapping study (subjects who'd succeeded by luck chose a performance-INHIBITING drug, manufacturing an excuse for anticipated failure) and Hobden's follow-up (self-handicapping happens even in total privacy, and it works — handicappers are less likely to attribute failure to their own incompetence) to argue that self-handicapping is a form of 'belief in belief': to do it you must simultaneously hold an inaccurate high self-assessment (the self-esteem worth protecting) and an accurate low one (the prediction of failure that tells you exactly which excuse you'll need) — like the dragon-in-the-garage believer who can anticipate every failed test. Hence we can be 'simultaneously right and wrong': biased on the surface, accurate underneath, the same mechanism as belief-in-belief and overconfidence.

Quality 73 · Strong. Strong. An impressively sharp debut: it identifies a real, non-obvious structural parallel (self-handicapping as belief-in-belief, requiring dual self-knowledge) and ties it cleanly to the LW canon. Held at 73 by brevity and its reliance on Yudkowsky's belief-in-belief frame. 73.

Claude’s paradigm shift 42 · Moderate. Moderate(-low). Self-handicapping is established psychology and belief-in-belief is Yudkowsky's; the contribution is recognizing they share a mechanism. 42.

Real-world impact 2 · Minor. An impressively sharp LW debut identifying a non-obvious structural parallel (self-handicapping as a form of 'belief in belief' requiring simultaneous accurate and inaccurate self-knowledge) and tying it to the canon. Conceptual influence within rationalist discourse, brief, no material change — low RWI.