Do leaks of classified information damage national security in tangible, meaningful ways? Probably, yes. Even the leakers must admit that they cannot entirely foresee the consequences of making public the methods and technology marshaled to preserve, protect, and defend the Constitution. Edward Snowden may be a genius, but one practical reason why leaking classified information is illegal is because the judgment of one person cannot possibly be submitted for the judgment of others, particularly those who have access to more of the big picture in a highly compartmentalized system. Those of us who don't have access to the big picture can really only speculate.
But national security is not a thing that someone can contain within a box. It is a description of a dynamic, personality-driven, externally influenced set of systems and procedures, where inputs, or threats, are often quite subjective, and where output criteria — whether the government can protect citizens from people who seek to harm them in a direct way — are not always obvious.
Something that makes us less safe in the short term can make us more safe in the long term. The exposure of a certain technique might discourage the bad guys from trying something, since they now know what the United States is capable of. (Logically, a disclosure is just as likely to discourage bad actors as it is to catalyze creativity among them, since there are many different reasons why someone would want to attack the U.S., or spy on the U.S., or otherwise cause harm.) Disclosures result in more visible accountability, which can change the way a potential adversary regards the legitimacy of U.S. actions. And disclosures can also limit unnecessary risks, where the rewards aren't sufficient to justify the action.
To put it another way, an unfettered CIA, the CIA of Richard Bissell, for example, is not necessarily a more effective CIA. An unregulated or lightly regulated NSA may be more susceptible to political influence and to tasking resources toward inappropriate or dubious targets.
Take the revelation that the NSA hacked into the email servers of Mexico's presidential office. The NSA can do this — it can break into the email of U.S allies — without presidential approval or any post-facto congressional oversight. And right now, as interesting as the "take" from that operation might be, I'm pretty sure that if Gen. Keith Alexander asked President Obama for permission to do this, Obama would have said no, or he would have placed some limits on the operation. Mexico is certainly capable of deceiving the U.S., and there is a legitimate reason to gather intelligence about the intentions of the country's leadership; to see, for example, if their internal deliberations about narcotics trafficking match what they tell the United States. But maybe the need to know this, the value added by this intelligence, is not worth the risk of being caught. Breaking into the email servers of an ally is not something I have a problem with, but I recognize that it is an extremely risky operation, one fraught with political and geopolitical consequences. In the short term, U.S.-Mexico relations will be tense. In the long-term, the NSA might think twice about hacking into the email of an ally. Or when it does so, it might think twice about how and where it stores such information.
The response to my conjecture is that U.S foreign policy must be backstopped, that presidents must have certain knowledge, and that only secret intelligence can provide this safety blanket for decisions involving life and death. I tend to agree with that in principle. I still don't quite understand whether NSA understands or is sufficiently aware of the nominal implications of such spying, even if it were to remain a secret.