There is a dupe checker, but only for exact matches (which if there is one instead of getting a warning it automatically lodges an upvote of that link on your behalf). The OP of this thread appended a # to the end of the URL to bypass the filter.

There's also a semantic dupe checker called RiderOfGiraffes. (Seriously, and with all due gratitude.)

Thank you, although not everyone agrees. Although sometimes they then get re-upvoted, many of the duplication notifications get downvotes. Here are two recent ones:

http://news.ycombinator.com/item?id=2392849

http://news.ycombinator.com/item?id=2392739

That's baffling to me, because I almost always click-through on the previous discussions. Maintaining discussion history adds value to the community.

then HN needs to implement URL normalization and equivalence from RFC3986. These are fresh in my mind because I extended urllib in Python to support them last week.

It isn't an easy problem to solve, working out if two URLs are equivalent, but the RFC goes some way to solving it which picks up easy things like adding a #.

see: http://tools.ietf.org/html/rfc3986#section-6

Thanks to hashbangs (or really just the prevalence of Ajax-only sites), it's now a lot harder to tell whether two pages with #something in the URL are effectively the same.

Sometimes the content on a page has changed, but still has the same URL; submitting it again isn't a dupe.

In those cases the item should be flagged - any discussion is potentially no longer relevant because it's about an earlier version.

No cool off period (although that is a good idea). I just submitted a good read and it head already been posted 616 days ago, and thus no one will see it.