Live data from Hacker News

Tell HN: Clickable domains and other new features for story quality

news.ycombinator.com

1–10 of 160 posts

Tell HN: Clickable domains and other new features for story quality

#1
Here are some experimental new features to help improve story quality on HN.

We've adjusted the dupe detector to reject fewer URLs. If a story hasn't had significant attention in about the last year, reposts are ok. That's been the policy for a while, but we've brought the software closer to it. It will still reject reposts for a few hours, though, to avoid stampedes. Allowing reposts is a way of giving high-quality stories multiple chances at making the front page. Please do this tastefully and don't overdo it.

When reposting, please don't delete the earlier post. Deletion is for things that shouldn't have been posted in the first place, such as if you regret having said something publicly.

When a story is a duplicate—that is, has had significant attention on HN in the last year or so—it's helpful to post a comment linking to the previous major thread, so users and/or moderators can flag the dupe. In addition, when a URL isn't the best source for a given story, it's helpful to post a better URL in the thread. We often see those and change the posts to use them.

Both these practices are common in the HN community and make a big difference to story quality here. Thank you all! The following features are intended to make them quicker to do. We built them to make moderation easier for ourselves, but hope they'll be helpful for community moderation too.

First, you can click on a story's domain to see the previous HN submissions from that site.

Second, when you're logged in, stories on /newest and on /item pages have 'past' and 'web' links. Click on 'past' to search HN for previous stories with that title. This helps with finding duplicates. Click on 'web' go to a Google search for the story title. This helps with finding better sources and catching spam.

Finally, when a story is the first post from a site, logged-in users will see the site name in green, by analogy with the green usernames of noob accounts.

These really are experimental and if any proves unhelpful, we'll toss it. We want HN to stay simple and coherent and not just be an agglutination of features. Your feedback will mostly decide what we do, so feed away!

Edit: ok, we tossed the green sites. More people disliked than liked them, and the same information is available just by clicking on the site name anyway.

Re: Tell HN: Clickable domains and other new features for story quality

#3

Very good improvements! However, the green-URL-for-noob-domains may be problematic. The green is used to give an indication for potentially low-quality content, but public startup/project launches would all fall under the noob-domain heading.

> The green is used to give an indication for potentially low-quality content.

I wouldn't say that at all! It's just interesting to know what's new.

We've been displaying new sites in green to moderators for a while now. When a high-quality story comes from a brand new (to HN) site, that's interesting. The converse, too: it's interesting when you can see that an obscure site isn't new to HN (because it's not green). Often someone posted it like 6 years ago and it got no upvotes; or sometimes there was actually a major thread. Now you can click on the domain to find out. HN has a rich history that's fun to explore.

Edit: I'll give you, though, that green sites and green usernames have higher variance. The good ones are really good (think of the author of a story showing up in the thread to discuss it), while the bad ones are more likely to be spam. I think it's fine to call attention to it in either case.

Re: Tell HN: Clickable domains and other new features for story quality

#4
The "past" link is a nice feature, but what I'd really like to see is a way of weeding out duplicate stories before they're posted. For example, today there were 9 similar stories posted (so far) about Facebook's new "dislike" button[1], causing none of them to receive a significant numbers of upvotes. Could we have a feature in the "submit" dialog that displays similar stories in the last 24 hours and then asks the user if they really still want to submit theirs?

[1] https://new-hn.algolia.com/?experimental&sort=byDate&prefix=...

Re: Tell HN: Clickable domains and other new features for story quality

#7
post #5

What about a way to merge stories that are duplicates but with slightly different URLs that didn't get caught in the filter? Perhaps comments from all the articles could get merged together under the post with the most upvotes.

We do exactly that when users find the duplicates and link to them from the thread, as well as when we find the duplicates ourselves.

Making higher-octane software that can automate more of this has been on our list for a while, but it's hard to know when we'll get to that.

Re: Tell HN: Clickable domains and other new features for story quality

#8
post #6

Very good improvements, thanks. I think the dupe detection would be even more useful if done during submission.

The dupe detection software, of course, does run during submission. Clicking on a search link is something that humans have to do, though. That's for catching duplicates that escape simple URL matching.

Writing software to identify which URLs are really about the same thing and which URLs are not is a nontrivial problem. I'd love to work on solving that in the general-enough case to be useful for HN, but we shouldn't let that stop us from doing incremental things to make life easier in the short run.

Re: Tell HN: Clickable domains and other new features for story quality

#9

The "past" link is a nice feature, but what I'd really like to see is a way of weeding out duplicate stories before they're posted. For example, today there were 9 similar stories posted (so far) about Facebook's new "dislike" button[1], causing none of them to receive a significant numbers of upvotes. Could we have a feature in the "submit" dialog that displays similar stories in the last 24 hours and then asks the…

Second this. I feel like I've seen more dupes lately as people rush to get a hot story in. The past and web links will help for sure but it won't stop the submitter unless they're prompted.

That being said, these are all great features. More information access without being overly cluttered is a huge win for everyone.

Re: Tell HN: Clickable domains and other new features for story quality

#10
post #8
post #6

Very good improvements, thanks. I think the dupe detection would be even more useful if done during submission.

The dupe detection software, of course, does run during submission. Clicking on a search link is something that humans have to do, though. That's for catching duplicates that escape simple URL matching. Writing software to identify which URLs are really about the same thing and which URLs are not is a nontrivial problem. I'd love to work on solving that in the general-enough case to be useful for HN, but we shouldn't…

I suspect what kawera meant is if the submission form notified you that the URL is a dupe via ajax, so that then you wouldn't have to bother to copy and paste the title (which can be a pain on phones and tablets).
Post reply on HN