>>1789
>be me
>want to find more quality MADs
>time and effort are limited resources
>want to write algorithm to sift through the videos that MAD reuploaders post on youtube in order to find quality content, while limiting non-MADs/MADs that I already have in my collection, as manually sorting through these would represent a waste of time (and thus resources)

In an ideal model, I'd want to be writing an algorithm whose output would abide by the following conclusion:
>A video will have been downloaded if and only if it is a MAD that I don't have in my collection
>This would also mean that an item wouldn't be downloaded if and only if it was not a MAD that didn't exist in my collection
However, there's no way to guarantee these conditions without something like my oversight of the downloads, so I'm just trying to get a reasonable approximation to maximize the amount of quality content that I can find per unit of time/effort. It's okay if I lose out on some quality content along the way, as long as the percentage of MADs that I haven't seen before is higher in my output than in my input.
Rather than opting simply not to download things, I could also put them into tiers based on the likelihood (as I've defined it) of them being original MADs, but bandwidth is another limited resource, and should probably be used on other methods with higher yield.