One particularly interesting argument seen in the Underwatergate scandal is the one that the NYT, by revealing the existence of warrantless wiretaps on international communications lines, compromised national security.
Reporters asked how that can be. After all, surely the bad guys knew the U.S. had the ability to perform surveillance on them, and has a secret intelligence court, and was presumably getting lots of secret warrants to watch them, and was furthermore watching them overseas without being subject to the 4th amendment.
The White House response was effectively, "Well, we're catching some of them with this program. So obviously in spite of the fact that they should know we are listening, they forget, and we learn things." In other words, the bad guys are sometimes stupid, and by bringing a lot of publicity on the surveillance (legal or illegal) we're reminding them not to be stupid.
I've seen this issue talked about before. Many members of the mafia have been caught with wiretaps, saying things on phones that you think they would know are probably tapped. This argument is used to counter the claim that since encrypting communications are readily available (such as in Skype) the smart criminals will not get caught with wiretaps.
Furthermore, in this case, while the White House revealed only minimal details of the program, security experts in blogs and other media around the world engaged in all sorts of informed speculation about what's really going on. While the NYT didn't reveal any technical details, kernels in the discussion almost surely do.
I'm willing to accept that even the smart criminals make mistakes, and get caught this way, and this will continue. So indeed, heavy publicity around the surveillance techniques and issues probably does, as they claim, instruct or remind some bad guys not to use certain communications that could put them at risk for being caught.
The harder question is this: Does that imply we must keep silent on these issues? I think the answer is clearly no. The standard the spooks and White House suggest is untenable, and there is no clear way to draw the line. Because if we use the stupidity of criminals as a standard, then it's hard to see what public discourse might not be considered potentially harmful to the exploitation of the criminal's mistakes. Yes, it's clear to see that a massive public debate with constant articles in all major media is more likely to remind a bad guy to watch what he says on the phone, more than a single blog posting would. But this is a difference of degree, not of kind.
In the end, it's a security through obscurity argument of a particularly high order. Not only must we not let the bad guys know that we can wiretap, we must not remind them after it is presumed they already know. It's hard to imagine a rule against this that would not chill speech at an extreme level.