I think part of the rationale for the 10-year expiration was to clean out the "deadwood" on the course list, and make the Registrar, VC, & state certifiers file cabinets manageable (although renewals tended to have the opposite effect).
We are now in the era of electronic document storage where space is not an issue - the entire course database can be put on a server, or burned onto a couple DVDs.
The average runner knows nothing about course expirations and cares even less, as long as the course is accurate. It should be the responsibility of the race director to make sure that is the case. A course that comes out significantly short (or long) will quickly be detected by GPS-wearing runners, and holy hell will erupt on the online message boards before the spilled GatorAde on the course has even dried.
My own opinion is in agreement with Pete; however, I am willing to compromise with a single ride and new certificate being issued (although that act defeats the purpose of Pete's proposal).
FWIW, I watched yesterday's Boston Marathon in the press room sitting next to Alan Steinfeld, former NYC Marathon race director and CEO of the NYRR. During the early part of the race I brought him abreast of this issue, and his opinion was the same - no new measurement should be required.