In the former case--ongoing practices--it is hard to make a credible argument against oversight.
Some details may need to be withheld from the public to protect basic operation security. For example, the identities of overseas informants ought not to be disclosed, lest they be immediately imprisoned or executed by their governments. Nor should there be a Twitter feed giving up-to-the-minute location data on US special forces teams. These examples seem farcical, but they illustrate the point that at least some operational details genuinely deserve secrecy. (Unless you reject the underlying premise that the United States should conduct intelligence and military operations overseas at all. But that's a very different discussion.)
But even though specific details may need to be kept secret--for a reasonable amount of time, until they're no longer actionable by enemies--the nature of the government's ongoing practices ought to be disclosed.
The latter case--wherein lower-level officials are permitted to make judgment calls in time-sensitive emergencies--is rather different. This is the situation contemplated by the White House during the Cold War.
As the reasoning went, the American nuclear deterrent must be credible, or else the Soviets would seize the opportunity and launch an unprovoked attack on the US, Germany, or both. If a communication breakdown between the White House and the military could render the nuclear arsenal unusable, that would tend to diminish the credibility of the deterrent. On the other hand, the risk of a lower-level official making the wrong judgment call and starting a world war is appreciable.
Given these considerations, reasonable people could disagree on what the optimal policy would be. It is not unreasonable or morally repugnant to conclude that certain trusted military officers should be granted the authority to make a judgment call when a) the situation is an emergency, and b) the chain of command has broken down.
Again, the key distinction here is between ongoing practices and emergency measures in a hypothetical, worst-case scenario. That distinction must affect our moral judgment of the respective policies.