
> From a security perspective, it's still concerning to be relying on large proof-checking binaries, where the answer to why we're supposed to trust one binary is that the binary has supposedly been checked by another large proof-checking binary. I'd like to see more work on simplifying these binaries. Bootstrapability and minifying trusted (binary) blobs remains an important part of a much larger software *verification*/transparency effort
Posted new page "False computer-checked proofs" linking to 199 reports of soundness bugs in several important proof-checking programs: cr.yp.to/proofs/falseccp.htm…
View quoted post ↗






















