Socket reported on 18 November that MITRE, which runs the CVE list of known software flaws, has admitted a timestamp problem. Jerry Gamblin examined the 52 weeks to 16 November 2026. MITRE gave 3,901 of its 4,509 records, or 87%, a publish time of exactly midnight UTC. None of 100,235 records from the other 425 issuers carried that time.
Alec Summers, MITRE's CVE project leader, said on LinkedIn that one publishing route makes the client supply a time. Because the old client has concurrency problems, MITRE picks one fixed value, midnight, for stability. The stamp is a design shortcut, not a measurement. Gamblin compared each stamp with the record's first appearance in the CVE List change log. In 202 sampled records, the 144 stamped at midnight appeared a median of 15.8 hours later. The remaining 58 showed up after a median of 4.6 minutes. Summers said the old route ends in Q1 2027. He said users have not shown a need to fix old timestamps.
A metric is only as honest as its least-checked field. The 15.8-hour gap would make MITRE look slow, though it reflects a placeholder. Ask any vendor reporting time-to-publish: which timestamp field feeds it, was it checked against a second source, and how are placeholder times handled?