Apart from the enjoyment of revisiting the classics (not too often), these reposts play the important role of brining junior cohorts of users up to speed with the HN culture (which, whatever it is, is one continuation of the lineage of internet lore).
(oh, and in case it isn't clear, the reason I post those 'related' lists is not to point out dupeness (we do that in other ways) but to just give curious readers other things to look at.)
I think because it often means repeating the same points over again. People who contributed to the previous/earlier mentions may feel too exhausted to repeat their view, provide the same useful references again, or fight the same old misconceptions about the topic.
An interesting related example came up this week, when many people posted submissions about the same news story (anti-matter & gravity). However, all the source links were different. Some commenters may want to make the same comment in all the threads. But there may also be different facts or views in the different articles, so they might each provoke different discussions. [FYI: In that case, an authoritative source Nature appears to have 'won', even though it was not the first one posted. It has the most upvotes and comments, compared to wire-blurbing sci-tech news sites just summarizing the raw CERN PR - which was also submitted. The Let 1,000 Flowers Bloom approach worked well in this case].
However, that's really a disjoint case from what we're talking about here, because once enough time has gone by to clear the hivemind caches, an "old" topic becomes "fresh" again for most readers—especially, of course, the ones who never saw the article before, but also for ones who know it as a classic, and so on.
The FAQ defines "enough time" as "a year or so"—we leave some wiggle room there because sometimes it seems to need a bit longer than a year, and sometimes a bit less is ok. This has been the rule on HN for a long time, well over a decade, and it has held up well over time - I think it strikes a great balance.
Perhaps impose a moratorium period of (at least) 3 months for resubmission. Would stop a lot of dupes in new.