Not a big deal, just a note on a slow friday.<p>Submitting articles to HN sometimes has a neat feature that check if URL was already submitted (increasing the voting of the first one). This to avoid duplication.<p>Just seems that the check does not parse URL encoding. As a test look at:<p>http://news.ycombinator.com/item?id=1991171<p>submitted as: http://www.davidflanagan.com/2010/12/let-it%2Dsnow.html<p>and<p>http://news.ycombinator.com/item?id=1989457<p>submitted as: http://www.davidflanagan.com/2010/12/let-it-snow.html<p>Back to programming mode.
I suspect that people occasionally use the simplicity of the URL detector as a feature, deliberately changing the URL in trivial ways so they can submit items they know or suspect will be popular.<p>More often, though, people just don't check at all and submit items with no checking at all. This leads to duplicate submissions, a crowded "New" page, and split disussions.<p>I just wish people would read more, so they'd have a chance of knowing that the cool thing they've just seen and want to submit has already been submitted, sometimes many times over. The Dup Detector is sufficiently simply (for which read "brain-dead") that it can't be relied on.<p>ADDED IN EDIT: Cool! A down-vote! I always appreciate the opportunity to learn something. If you've down-voted this, or have a reason why someone might, please reply with more explanation so I can understand the alternative point of view. Thanks.
There are many more ways to mess with the dupe filter (this is by no means an exhaustive list):<p>- adding or removing www. from the url (if it has it in there, and if the site redirects)<p>- deleting a url and re-submitting it if it does not pick up enough votes in the time it take so scroll off the newpage<p>- adding a ? or # to the end of the URL<p>The dupe filter is a way for people that play nice to continue to play nice, I don't think it ever was advertised as hyper secure. It's like the lock on the barn door, it stops the most trivial attempts and serves as an indicator of desired behaviour but it is not so secure that it can't be tampered with.<p>A really good dupe detector would actually be a lot harder than it probably seems.