This petition is probably just a confirmation of the majority of the names on that list
It's far worse than "just confirmation." If the NSA is training a machine learning algorithm to identify subversives based on various features/activities of those people, they need a set of labeled (i.e. subversive = yes/no) training data for the algorithm to learn from. They now have 100,000 extra training samples to help the machine learning algorithm better tune its ability to identify subversives. So, not only have the 100,000 people moved from the "probably subversive" to "known subversive" column, but the NSA is better able to predict subversiveness for all of the people that didn't sign the petition by comparing their features/activities to those that did.
That's a perspective I hadn't thought of, and will try to apply to other situations.
On the other hand, if the 100,000 people who already signed have now identified me as a subversive, I'd better go ahead and sign, in the hopes that it will maximize the odds of reigning in the NSA before they come get me.
It's far worse than "just confirmation." If the NSA is training a machine learning algorithm to identify subversives based on various features/activities of those people, they need a set of labeled (i.e. subversive = yes/no) training data for the algorithm to learn from. They now have 100,000 extra training samples to help the machine learning algorithm better tune its ability to identify subversives. So, not only have the 100,000 people moved from the "probably subversive" to "known subversive" column, but the NSA is better able to predict subversiveness for all of the people that didn't sign the petition by comparing their features/activities to those that did.