Latest AI Code
tensortalk.com
tensortalk.com
HN's approach is straightforward: we treat an article as a dupe if it has had significant attention in the last year or so. That's frustrating when there's a post you particularly like, but all we can really do is try to be even-handed in general. The limited spots on the front page are the scarcest resource HN has.
Perhaps a review with the aim of considering a less binary solution to this issue is in order?
I like the tagging mechanism, HN could learn from this :)
In fact, with all the AI code available, it should be easy to make tagging automatic :)