Procedural Knowledge Gaps, part 2 — LessWrong