I think the concern with incidental information is that the non-incidental, the critical information, in a lot of cases already feels too broad. The ability to keep that incidental information—to first define it, and then also retain it—is quite challenging for us as citizens, for our community to really understand what that means to them. How do they know it's incidental? How do we as regular people understand what information is being collected about us, in addition to how hard it is to keep that incidental information? I think we're hearing a lot of arguments from law enforcement about how incidental information is not actually incidental, which is our first concern, and that, even when it is, it could be useful in the future. I think if you're looking at 25 or 30 years ago, that might be a notepad that's stored somewhere in a filling cabinet. I think right now it's these datasets from which that information is being cross-referenced against multiple other datasets that can potentially provide a lot of either false positives or misleading information, which actually take away from the very purpose of trying to collect this information in the first place.
On February 8th, 2018. See this statement in context.