The nature of much of the NSA's signals intelligence work -- outside of codebreaking -- is attempting to work with huge collections of raw data. That means a lot of what they're going to be doing is collecting a metric shitload of information, trying extraction algorithms against them, finding out that they haven't yet got anything useful, and repeating the process.
And I'd bet they have the same problem many collectors and museums do -- having gone to the trouble of acquiring something, it's very hard to declare that it's obsolete and uninteresting and that you really can deacquisition it without being accused of failing your mission.
In other words: This sounds like exactly what I would expect business-as-usual to be at the NSA. I really don't find it surprising.