Hacker News Reader

(Added in edit to explain someone else's comment and reference) ... it doesn't beg the question, it raises it. But putting that to one side ...

They're not all the same URL, and when a submission is old enough it's no longer tested against. Minor variations in the URL will also thwart the dup detector.

Several suggestions have been made in the past about improving the dup detector, but it's unlikely they'll be implemented.

View on HN

Not sure if this has been suggested or not, but what about a list of "similar articles" below yours (based on a search of past articles based on title/content/etc) so that you can verify yours is unique?

It's an interesting idea, and could be useful. I think that finding a selection that matches your intended submission could be difficult. Would be cool, though.

Technically not difficult, but would add quite a bit of complexity to HN in order to fetch and store at least a fuzzy hash of the article bodies.

From there it's mostly a case for a bayesian classifier (LSI) and some elbow grease.

FWIW, the "similar articles" footer on most blogs is implemented that way.

Reply on news.ycombinator.com