Peter Attia· MD
the past performance of a test tells you something not everything but tells you something about the future performance of a test
Loading…
The headline is broadly defensible, but the qualifications matter. Effect sizes vary by population, the strongest claims rest on shorter trials, and credible voices push back on how it's typically framed.
the past performance of a test tells you something not everything but tells you something about the future performance of a test
Every Sunday: the week’s new conflicts and verdict changes — and nothing else.
Native comments, Twitter mentions, and Reddit threads about this claim — surfaced together so the conversation isn't fragmented across platforms.
Bookmarking — the dossier-vs-overview split is the right call. Most of the time I want overview; sometimes I want receipts.
Would love a "what would change this verdict" RSS feed. Sign me up if it exists.
Bayes' fundamental idea was that you can only interpret a test in the light of what that test has predicted in the past. It's an extraordinarily important idea in the way we think about the universe, that the past performance of a test tells you something, not everything, but tells you something about the future performance of a test.