There is a built-in duplicate detector, but it doesn't work very well. More and more the URL people end up with has loads of feedburner crap and other junk, and that breaks the URL dup detector.

I made this suggestion ages ago:

http://news.ycombinator.com/item?id=1012215

The idea is to use just the first part of the URL, and then to use the extracted title. Seems much more reliable - I use it to auto-detect duplications. It annoys me to have the same item submitted over and again, mostly because it splits the discussion. For me, most of the value here on HN is in the comments - often the discussion has more value than the original submission.