Scott Alexander argues that psychiatric diagnoses, while imperfect, are useful tools despite potentially combining multiple conditions with different causes. Longer summary

Scott Alexander discusses the criticism that psychiatric conditions like autism or depression are not unitary categories but rather collections of different conditions with different causes. He argues that this view, while correct, is not as revolutionary as some think and doesn't invalidate the usefulness of these diagnoses. Scott compares psychiatric conditions to medical conditions like pulmonary edema or stroke, which can have multiple causes but are still useful diagnostic categories. He suggests that psychiatric disorders might ultimately be understood as computational conditions, with various biological, psychological, and environmental factors affecting the brain's computational parameters. The post emphasizes that while research into subtypes of conditions like depression hasn't been very productive, current psychiatric diagnoses remain the most useful tool available, despite their limitations. Shorter summary