> The government classification isn't about protecting the secrets of private citizens, only government secrets.
Using that logic, Facebook private messages supplied courtesy of the NSA would be highly classified, considering their collection mechanism would be PRISM or similar closely-held programs.
> I have some experience in this field - handling fingerprints of non-criminals which were classified at the bare minimum for the agency.
Fingerprint data is available to nearly every local law enforcement organization in the country. There's no mystery about the collection method. It's a foregone conclusion that it's aggregated up the chain.
As other comments have said, if immigration officials were using NSA-supplied data, it would likely be in vague, aggregate form that did not hint at the underlying collection mechanism.
That only flies if there is no other possible way for another agency to have acquired the information. An administrative subpoena would be one way to do it -- I expect there is a term for that sort of thing, although the obvious "warrant laundering" does not seem to pull up any meaningful hits in google.
BTW, unless something has changed since I was last in west virginia, fingerprint info is not widely available in the general case. You take prints from an arrestee or latents from an object, you send them to IAFIS for matching and they return hits which may not even include a copy of the matched prints depending on the circumstances.
It shouldn't fly. That's why in my original post NSA was listed dead last as a probable data source, far below normal federal law enforcement means (such as administrative subpoenas). Even those were far less likely than simpler explanations.
As far as fingerprints, I agree the interface available to IAFIS database end-users may be restrictive. However, due to the highly aggregated nature of the data, and the fact it's maintained for law enforcement by law enforcement (namely the FBI), people expect it to be available to a myriad of government agencies (which it is).
With Facebook private messages, you expect only Facebook employees (or partners) to have access, or in special cases government via warrant or subpoena. Major tech companies took a fairly large PR hit following the news of NSA having wholesale access to their systems, precisely because of these expectations.
If the headline was instead "FBI grants NSA full unrestricted IAFIS access." nobody would bat an eye, because it's expected. The FBI is heavily involved in national security affairs themselves.
My point was, the sensitivity of the collection method is a major factor in determining the classification level of the data. The PRISM leak is a prime example of this.
What I believe is that if the NSA are already monitoring someone for whatever reason and see evidence of law-breaking that they would call up an appropriately cleared contact in the relevant agency and say, hey you should look into this person. The agency is free to run with it from there, using whatever tools with minimal oversight are available to them and no need to even mention the NSA after that point.
In fact, I believe that the number of actual terrorists is so small that such (ab)uses of the system are inevitable. In the same way that all the federal money for SWAT training of police departments has resulted in SWAT teams being deployed for all manner of low-risk situations, there just ain't enough real work for them to do so the mission creeps.
Using that logic, Facebook private messages supplied courtesy of the NSA would be highly classified, considering their collection mechanism would be PRISM or similar closely-held programs.
> I have some experience in this field - handling fingerprints of non-criminals which were classified at the bare minimum for the agency.
Fingerprint data is available to nearly every local law enforcement organization in the country. There's no mystery about the collection method. It's a foregone conclusion that it's aggregated up the chain.
As other comments have said, if immigration officials were using NSA-supplied data, it would likely be in vague, aggregate form that did not hint at the underlying collection mechanism.