I don't want to sound negative, but.. there is so much content produced every day on sites like HN that I feel a bit like the past discussions don't matter anymore at all.
I disagree strongly; there's a lot of gold in HN's archives. Some things don't change, or don't change that quickly, and there are some topics (even in the young field of CS) that are classic, and some great discussion has already taken place on HN about them. Why do you think that doesn't matter? HN isn't even that old.
I don't have a silver bullet, but if you have a topic you're interested in, the search provided by Octopart is very good (bottom of the page), and it's not that time consuming to dig through the results. That's how I found mahmud's comment on Lisp compilers, because I was looking for discussion on HN about it.
Obviously some topics are harder to search for (for instance if you wanted to read about what HNers have said about Techcrunch over the years) but for the good and classic stuff (programming, math, etc) I think this works reasonably well.
It would actually be really cool to see what HN threads are linked to the most from HN. I bet that would be a decent way to find gold, especially on stuff you weren't specifically looking for.
I disagree, older content can be much better, particualrly on somewhere like reddit where there has recently been a hge influx of users diluting the current content, the older threads such as those on askreddit were of a higher quality and had much better discussions.
