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.
because if you ever test 99 specificity uh that has 20 sensitivity now you drop the specificity to 80 your sensitivity management goes up by 20 right because it's just it's just the dumb luck yeah and your negative predictive value goes out the window so so that's the problem right
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.