Hacker News new | past | comments | ask | show | jobs | submit login

Your argument assumes that once a story has appeared once or twice, the community has probably seen enough of it for lack of votes to count as a consensus. That's a natural assumption, but false: it overestimates how much attention /newest gets and underestimates the impact of randomness.

When I became public as moderator I heard from a lot of users (e.g. [1]) that the biggest problem with HN was high-quality stories not making it to the front page. At first I was skeptical. I was looking at /newest a lot as a moderator and hadn't been noticing much good stuff on the cutting room floor. But the users saying this were good, observant HNers, and I kept hearing it, and it got me worried. So I started looking at /newest more systematically, and then I was shocked. Not only were many high-quality stories being overlooked, probably most of the finest submissions—the out-of-the-way, intellectually interesting, totally unexpected submissions that make for the best of HN—were getting ignored. I'm still not sure how I could have missed that; I must have been looking at /newest at least as much as anybody.

Since last summer, we've put a lot of effort into addressing this. One simple thing we did was clarify that a few reposts are ok if a story hasn't had significant attention yet [2]. That's why it's ok that nkurz reposted this one.

ScottBurson hit the nail on the head, I think: giving good stories multiple cracks at the bat is one way to counteract the randomness of what gets traction here. Sometimes people abuse it, e.g. by reposting things that don't belong on HN or by being overly promotional, but those are fairly easy problems to solve and their cost is much smaller than the benefit of having more substantive stories—especially those out-of-the-way gems—on the front page.

The above isn't an opinion about the current article, which I haven't read, but it's clear that nkurz reposted it for the right reason, which is that he thinks it's intellectually interesting and well-written. That's the kind of thing we want users to do.

1. https://news.ycombinator.com/item?id=7494367

2. https://news.ycombinator.com/newsfaq.html




Thought on "New": splitting the load among more users would be useful.

There's so much submission to the queue that if you show everything to everyone, good quality posts simply slide off the page too quickly to be reviewed.

An alternative would be to set some sort of threshold below which submissions aren't seen by all site visitors, but are instead revealed to only a subset. That subset increases as votes come in. You're basically increasing the probability that any one story will be seen by someone. Set levels such that a submission is on the "New" page for some suitable period of time (an hour, six hours, twelve, 24, whatever seems pragmatically reasonable), during which someone ought to see it.

Alternatively: eliminate the "New" page entirely and populate the front page directly with distinguished new submissions, again though, each presented to only a subset of visitors.


Do these reposts have to be by different people, or can the original submitter of a story that goes nowhere resubmit it a couple more times over the next day or so to try to get lucky?


They don't have to be by different people.


It wasn't an argument, I was just trying to clarify the lack of prior attention.

You're right, I didn't realize that this was now part of the FAQ. Thanks for the clarification!


But this runs in direct opposition to the title-matching rejection algorithm that is in place. (If posters preserve an article title properly, their attempts to repost are intercepted and redirected to the predecessor.)

If it is intended that duplicates be allowed, then perhaps the discussion threads for such duplicates should be merged together, rather than getting such posters in the habit of not honoring original titles.


There's no title-matching algorithm. The duplicate detector only considers URLs.

Beyond that, I'm not sure I get your point. It's definitely intended that reposts be allowed if (and only if) a story hasn't had attention in about a year. You have to change the URL slightly to make the repost work, and that's by design. Merging threads doesn't solve the problem we're talking about here, where the previous post didn't get much attention, since you'd be merging into a dead thread. The problem merging solves is the opposite one, when you have multiple active threads on the same story, and we do merge threads when we see that.

Better dupe detection is on our list to work on, but it's a harder problem than it sounds and I don't know when we'll get to it.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: