SDPS / Method qualification

A run can look finished before it is finished.

Structural Discovery and Proof Search asks whether constrained computational search can produce a mathematical result that is valid, genuinely new relative to its baseline, and independently inspectable. The current work is testing the harness—not claiming a discovery.

What failed

Independent review found that the Docker path could certify terminal CPU and wall-time evidence while its final inspection still reported the container as running. The evidence could therefore say “finished” before the execution boundary proved that the run had ended.

Why it matters

A discovery system cannot make a trustworthy claim if its execution record is ambiguous. Resource limits, isolation, final outputs, and termination state are part of the evidence—not administrative details around it.

What we know

Known Hadamard constructions serve as positive controls. Narrow restricted searches provide exact local negative cases. Those results test pieces of the verifier and protocol; they do not show that the system can discover a new matrix or theorem.

Next gate

Repair termination evidence, reproduce the repair independently, restore deterministic CI, and preserve the accepted protocol baseline. No scored search, paid computation, or public discovery claim proceeds before those conditions are met.

What is not claimed

No new Hadamard matrix, theorem, broad nonexistence result, or competitive discovery advantage has been established. Passing synthetic controls would validate only the tested harness behavior.