PCR-C — Control Before the Boundary
Pre-Critical Recursive Cutoff · infrastructure/control research layer
Research object
PCR-C is a staged infrastructure control framework. Its role is not to redefine the LUMINA-30 boundary, but to formalize one infrastructure-layer mechanism for acting before that boundary is crossed.
Why a control path can fail before a stop is requested
A nominal stop control is not the same as an effective intervention path. The relevant research problem is whether intervention remains structurally reachable before recursive or self-reinforcing escalation makes refusal ineffective.
Relationship to LUMINA-30
LUMINA-30 asks the boundary question: was effective human refusal still available before irreversible impact? PCR-C addresses an infrastructure-layer question: can control act before the system reaches the point at which that refusal becomes structurally ineffective?
Research use
- Reasoning about staged intervention before irreversibility thresholds.
- Separating formal stop authority from the continued reachability of the infrastructure control path.
- Connecting pre-incident review to infrastructure-level control design.
- Studying where intervention capacity becomes insufficient relative to system escalation.
What PCR-C does not claim
PCR-C does not replace incident review, human responsibility, the LUMINA-30 canonical boundary definition, or every other AI safety method. It is one research layer in the public stack.