If we had a technique that would defeat almost every lock in the world, we might be prudent to delay releasing that until people have had time to change the locks.

Same general benefit as coordinated disclosure, except it's a big game-theory problem where we're likely to eventually have sufficiently capable defectors.