[detrites]

The problem is that leaks prove, every time they happen, that there definitively is a given amount of successfully-hidden information, up until that time. We may also safely assume leaks are the exception not rule, and so not complete.

It'd be interesting to see a "time to disclosure" metric, of average times from classification to leak, to get a lower bound on secrecy, but it could only be reasonably considered a lower bound, the upper could be anywhere.

It's also possible agencies responsible for secrets may have worked hard to maintain an aura of being incompetent at keeping them, in order to hold an advantage over their adversaries by misdirecting efforts to the wrong areas.