Implementation-level vulnerabilities are a persistent threat to the security of computing systems. We propose using the results of partially-successful verification attempts to place a numerical upper bound on the insecurity of systems, in order to motivate improvement.
Thomas E. Hart, Marsha Chechik, David Lie