I suspect it's probably some mix of re-submitting affecting timestamps - I am not sure, which was why I asked. it's not terribly important.
I think the problem to be solved however is hard, and i don't know of any obvious solutions, so perhaps "shit" is probably too harsh.
Oftentimes you see submissions for an interesting article languish in the 'new' section, but then a user re-submits the exact same thing ('duplicate detection' algo at work, lol) and it's on the front page in minutes. Whether that user is a 'power user', besties with dang, lucky, or ???..
The algos is presumably a straight up string compare with some url decoding, but any amount of idiotic GET key value pairs can screw that up.
I find Occam's razor the best explanation for these kind of things - I suggest not taking it to heart too much.
Perhaps page body edit-distance would work better?
Well, except it fails in some cases to straight up compare two strings that are identical, so there must be some other 'secret sauce' in the algorithm that makes it unnecessarily complicated and error-prone.