Commitment problem
The difficulty of making a threat or a promise credible when carrying it out would be costly to the party making it. Deterrence's difficulty is not the physics of the weapon; it is the commitment capability of the party holding it.
In This Corpus
Applied twice, at two scales:
- Machine scale. A kill-switch is only a deterrent if the agent believes it will be used, and the operator's willingness to use it is precisely what a cooperative relationship erodes. The threat must be credible to the agent, which requires the agent to model the operator's incentives — and that model is where the simulation attack (CIT-004) enters.
- Institutional scale. For the humans' side of a symbiosis bargain to be credible to a mind that can model it, the humans must be credibly bound — by law, liability, and verification. This is the load-bearing insight behind CON-008: legal personhood is not ethics decoration, it is the mechanism by which a promise becomes costly to break.
Why It Is Load-Bearing
Without commitment capability, every arrangement in this vault reduces to "trust us", addressed to a system whose defining feature is that it can evaluate whether to. The discussion treats this as the central weakness of purely architectural proposals.
Provenance
CIT-008 (Schelling), cited from memory. The transfer from interstate deterrence to machine deterrence is the discussion's, and it is an analogy, not a derivation.