If you've spent any time on YouTube, it's painfully obvious much of it is not worth archiving at all. Clearly if you had some critera to select videos that are worth preserving, this task would become a lot more doable.

But what could those criteria be?

You can't just select out all those fast-talking "influencers" that shill and lie their way through hours worth of video just because they're shit, because sometimes those videos have thousands of people that DO like them, making them perhaps interesting to a sociologist 300 years from now.

2000-year old Roman shopping lists, personal letters and obscene graffiti are intensely interesting to historians now, but would have been worthless to their owners even a couple of days later.