You don't need to catch every secure communication. Even a 1% identification rate is enough; less if you're willing to toss in some traffic analysis into the mix. Your goal, after all, isn't to prevent secure communication, or even to identify what's inside it, it's to identify the people doing it, because the secure communication itself, rather than the contents of the communication, are what you've banned.

But I don't think anybody -wants- this level of control. The Internet is too powerful a weapon.