We also have to consider the (near future?) implications of machine learning on sifting through the "noise" that is generated with tools like this. In fact, by knowing that one is using such a tool, one could envision a scenario where this data adds more information about the user and not less - in particular about everything that one is not actually searching for or what OS the user is actually using. Consider that just clicking on a random search link with a browser automation engine probably doesn't look like a real user browsing without considerable effort on the developers part. Here we arrive at counter ML :)

>The idea that it's possible to overflow the governments secret lists is founded on the idea that the government cares if a list gets too long. I'm not sure that's the case.

It's not just about governments having extensive and precise information on us, it's anyone who might use it (i.e. hackers, insiders) for nefarious purposes.

Whether or not they "care" if the data is superfluous is unimportant, the goal is to make reliably narrowing down a target more time consuming.

Why posit this? It's but one of many possible considerations. My activities have nothing to do with "pressure cookers", nor am I exclusively concerned with "secret lists".