Previously, previously, previously, previously, previously, previously, previously, previously, previously, previously, previously, previously.
Yikes! Surprisingly, there is a precedent, of sorts:
I'm flagged for a children's play where the kids sing come all ye faithful. Pretty sure they are just sitting on it until their due date passes.
Judging from the examples the content id algorithm is working fine.
The problem is youtube's assumption that the copyright claimant is right by default, even if the claim is absurd like owning the copyright to white noise.
Content ID system != content ID algorithm.
Their entire principle -- that software can automatically and correctly interpret copyright law and fair use, and that claimants always do so in good faith -- is fundamentally broken. Like, "artificial general intelligence, halting state" broken.
If they lessened a financial incentive, by shifting monetized ad revenue back to the original poster when the dispute is affirmed, with triple damages, we'd have a lot less of this.
Over on the Stack Exchange network (StackOverflow, etc.), they have a system that allows edits being made to questions to be accepted or vetoed. Every so often they will serve up a decoy edit to ensure you're paying attention. It seems to me that YouTube could use a similar mechanism to weed out overzealous content-bots, assuming they were arsed fixing this in the first place
I like Jim Sterling's approach to fuck with their system. It's entertaining, at least. https://kotaku.com/game-critic-uses-brilliant-workaround-for-youtubes-copy-1773452452