I remember that nonsense. It was an oversimplification of what Echelon is and what it does. Echelon was never "jammed" by the keyword saturation. Echelon dealt exclusively with semantic forests which rendered those messages irrelevant.
And even if those messages had to be processed, the very idea ignores the sheer computing processing power that the NSA possesses. And baby, it's stuff we won't see on the open market for another 5-10 years.
Someone (Bamford?) said that NSA measures their computing power not in number of workstations, but in acres.
Precisely. Which is why those idiotic email "echelon spamming" things are so stupid. This is not some wanker-esque keyword matching scheme but sophisticated mathematics.
Incidentally, far slicker technology than semantic forests is under development in the US private sector. There have been some very interesting theoretical developments in the computational side of algorithmic information theory, which is the grand-daddy of all this stuff. People have no idea what kind of wicked ingenuity comes out of the US.