I've encountered very few examples of outright deliberate fabrication of evidence. That isn't to say hoax claimants are universally honest. They aren't. You have, for example, Sibrel deliberately cherry-picking the evidence, and you have Sam Colby deliberately misrepresenting it. You have David Percy pretending to be an expert and then misinterpreting it. And at the most innocent end of the spectrum you have people defaulting to nefarious explanations for unexpected things that creep into the record. As illustrated here, often you can solve the problem by going to the original (or a better) source. Most of what we use today for evidence are convenience samples and sources. They're easy to obtain, but not always the best kept. Audio recordings will acquire glitches from copies through multiple formats, some of them lossy or prone to artifacts. Video recordings lose resolution and time stability. Photographs lose resolution.
Regarding the latter, I remind you about the reseau fiducials. When I shot using a reseau plate, I tested several hypotheses. The first was saturation and halation. No dice; the fiducials remained clear. Then I tried scanning, both with commercial transparency scanners and consumer-grade. Again, no luck. It wasn't until I compressed the images with a lossy compression algorithm and shrank them to web-distributable sizes that the fiducials disappeared "behind" bright patches. Over the years we've listened to, and debunked, countless theories for how the fiducials got lost. We could have cut to the chase. When I was finally able to get my hands on 4,000 dpi scans of the camera originals, the fiducials were not missing from them.
Logically we call this "subversion of support." If possible, it's the best way to refute a hoax theory -- or any theory based on alternate explanations for allegedly anomalous data. To subvert support means to point out that the thing being explained doesn't really exist and therefore requires no explanation.
