Okay, and now comes the labor-intensive part, which Keltham is frankly not looking forwards to. Personally? He does not actually like tackling epistemological problems rigorously. It's work! Not all of it is fun work either, when you've got to do this much of it! It wouldn't be fun even if what he was trying to destroy with truth was not the seeming of things precious to him!
What that looks like, in a case like this, is indeed not that you pick a prior, and then go through everything you've seen and assign Ordinary vs. Conspiracy likelihood ratios.
What it looks like... is a large research problem.