The strongest objection
A government may believe that stopping its own development would hand decisive advantages to a rival. Secret programs and unequal enforcement could make restraint unstable. If a unilateral restriction merely changes who develops a dangerous system, it may provide little protection and could worsen other security risks.
Why a race does not settle the question
Winning a development race is useful only if the resulting capability remains governable and supports a tolerable security order. If the principal danger includes losing control of the winner's system, being first does not resolve it. Competition can also pressure participants to deploy before they have adequate assurance.
The prevention response is therefore international: negotiate limits, verify consequential activity, coordinate relevant supply chains, and make violations costly. That response must address credible fears of cheating and unequal vulnerability. A shared catastrophe risk may create room for cooperation, but it does not erase conflicting interests.
The alternative should also be specified honestly. Continued competition is a strategy with its own assumptions about control, deterrence, and the stability of a large capability advantage. It is not a neutral baseline whose success can simply be assumed.
The hinge of disagreement
Can an agreement detect and constrain dangerous development more reliably than a race can produce a controllable winner? That is an empirical and institutional question. W360 favors building the conditions for enforceable prevention, while acknowledging that participation and verification are essential feasibility requirements.