Sometimes, old-school metadata can be important.
Fer example: an ex-military colleague once explained to me how newspaper clippings could wind up classified. It's not the content of the newspaper article that's significant; it's the little scribbles in the margin, or just the fact that someone in this office thought this article interesting enough to cut it out and file it.
Now, let's consider the Snowden Files. They're computer documents, released to the public by a person of dubious reputation. They look official... but how many among us know what actual official NSA documents look like, and would comment on the matter in public? And how many of those could verify the actual authenticity of the documents, or even the plausibility of the content, even in general terms?
Right. Hardly anyone.
But! We now have metadata. Specifically: various government departments have been sending forth directives cautioning their employees not to look at the classified documents (or not to do so on government networks reserved for unclassified material).
While I'm sure our masters in Washington would like us to believe that the NSA has been totally behaving itself, and the scandal is all a hoax, and anyone can create documents on a home computer and slap the NSA logo on them... the official reaction is based on the documents being authentic.
Not exactly the traditional "neither confirm nor deny", nor even a "Senator, you know I can't answer that question one way or the other in public, but if you'd care to arrange a private meeting I could answer it for you." Nope. We got public officials telling bald-faced lies, then, when documents of dubious provenance show up exposing their lies, their reactions confirm the authenticity of those documents.
Can't anyone play this game?