Scott Alexander discusses the dangers of relying on 'Heuristics That Almost Always Work' through various examples, highlighting their limitations and potential consequences. Longer summary

Scott Alexander explores the concept of 'Heuristics That Almost Always Work' through various examples, such as a security guard, doctor, futurist, skeptic, interviewer, queen, and weatherman. He argues that while these heuristics are correct 99.9% of the time, they provide no real value and could be replaced by a rock with a simple message. The post highlights the dangers of relying too heavily on such heuristics, including wasted resources on experts, false confidence, and the potential for catastrophic failures when the rare exceptions occur. Scott concludes by noting that those who dismiss rationality often rely on these heuristics themselves, and emphasizes the importance of being aware of the 0.1% of cases where the heuristics fail. Shorter summary