5-HTTLPR: A Pointed Review

A landmark meta-science essay on the candidate-gene catastrophe. The serotonin-transporter gene 5-HTTLPR generated 450+ studies linking it to depression, amygdala function, stress-interaction ('orchid genes'), SSRI response, even nostalgia-proneness — a whole scientific edifice. Then Border et al (2019, n=600,000, techniques 20 years more advanced) found NOTHING: neither 5-HTTLPR nor 17 other 'depression genes' had any effect (gene-interaction p=.919). Scott extracts the lessons: candidate-gene studies were hopelessly underpowered (median n=345 vs the ~34,000 needed given polygenicity); 5-HTTLPR 'confirmed' whatever each researcher already believed (HPA, BDNF, bad parenting) because crappy science finds what it seeks; and the pharmacogenomic-testing industry (GeneSight, $410M) is built on this debunked foundation. 'ALL OF THIS IS LIES' — the unicorn-life-cycle-wrestling-Bigfoot metaphor for a field that studied 'something that doesn't exist.'

Quality 84 · Excellent. 84 — high Excellent. One of Scott's most important, most-cited meta-science essays; the 'you can build a whole 450-study field on a gene that does nothing' demonstration + the underpowering/prior-confirmation mechanism + the pharmacogenomic-testing warning are durable, important, and beautifully executed. Sits in the top replication-crisis tier (just below Control-Group-Is-Out-Of-Control on focus/scope).

Claude’s paradigm shift 66 · Notable shift. 66 — Notable, top of band. The candidate-gene-catastrophe demonstration + the 'a whole field can study a nonexistent effect, and no amount of context-sensitivity fixes that' point was a fresh, widely-cited contribution to understanding the replication crisis.

Real-world impact 3 · Moderate. 3 — moderate. Became a widely-cited explainer of the candidate-gene disaster within psychiatry/genetics/meta-science, with concrete spillover into pharmacogenomic-testing skepticism; discourse-level but genuinely influential.