How Natural Tradeoff And Failure Components?
Read the original on Astral Codex Ten →
Follows up on
↳ Ontology Of Psychiatric Conditions: Tradeoffs And Failures — Essay · Feb 2021
Summary
A short follow-up prompted by new work on the genetics of psychosis, which resolves a long-standing oddity — schizophrenia risk genes appeared to lower IQ while raising educational attainment — by splitting the genetic signal into two components: one shared with bipolar that raises educational attainment, and one not shared with bipolar that lowers IQ. Scott reads this as confirmation of the tradeoff-versus-failure distinction he drew in 2021: the first component is a tradeoff (costly in psychosis risk, paid for in something like creativity or drive), the second a failure (bad in every direction, probably deleterious mutations in neurogenesis and synaptic pruning). He then generalizes past psychiatry: any sufficiently multifactorial bad outcome should decompose into both kinds of cause. Poverty can come from failures (low ability, chronic illness, bad schools) or tradeoffs (the starving artist, the bohemian). Singleness, bad pizza, and cancer get the same treatment — the cancer case being the sharpest, since easy cell division buys healing and growth at the price of oncogenesis while limited division trades the reverse. He closes with two honest limits: the advantage often attaches to the risk rather than the condition (cancer itself is simply bad), and conditions too simple to be multifactorial need not have both etiologies, muscular dystrophy being merely the predictable consequence of a very large gene. The verdict is that what he once treated as a mysterious feature of psychiatric conditions has been dissolved into something that now looks obvious.
Why this score
Quality 62 · Strong. Solid-to-Strong. A clean, honest, well-turned short piece whose real move — promoting a psychiatric observation into a general claim about multifactorial causation — is genuinely portable, and the intellectual humility of 'dissolving' his own earlier puzzle is the right instinct. But it is brief, largely reports someone else's result, and the generalization is developed by quick illustration rather than argued; the examples do the work of a proof they do not supply. Well below its parent ACX-058 (76), which built the framework this post merely confirms and tidies.
Claude’s paradigm shift 42 · Moderate. Moderate. The tradeoff/failure distinction is his own prior contribution, so the novelty here is confined to the observation that the split should be generic to multifactorial problems — a sharpening of an idea already in circulation rather than a new one. He says so himself, calling the conclusion obvious in retrospect.
Real-world impact 1 · Negligible. Negligible-to-minor. A conceptual tidy-up within psychiatric ontology, of interest to a niche of readers thinking about how to classify conditions. No effect on diagnosis, treatment, policy, or ordinary lives.