The most interesting question was about whether the NSA would prefer a perfectly secure internet or a usefully insecure one (roughly paraphrased).
That's not far from what I wanted to ask: given the offensive value of 0-day exploits (as seen with Stuxnet, regardless of who actually did it), can agencies in "Cyber Command" really be trusted to give theirs up via responsible disclosure?