Hacker News Reader

Comment by pg | Hacker News Reader

We advocate that founders pay a lot of attention to users, not that they literally do what users ask.

What keeps users here is not features but the quality of the links and the comments. So that is what I spend the most time on. E.g. we have really good software for catching spam, and also semi-spam promoted by voting rings. There are also a number of subtle tweaks to prevent flame wars.

The last code I released (the day before yesterday) was to enable the system to ignore the flags of people who flagged too indiscriminately. Which in turn depended on code I wrote to study how big a problem bad flagging is. Like good defense, this kind of work is generally invisible. Even news.ycombinator.com/classic is something I wrote for myself to try to gauge whether quality was declining.

View on HN

I totally agree; it's just the incongruity.

The epic feature request thread seems vestigial: you often responded there 800-900 days ago but haven't for 428 days, I'd guess because HN has outgrown it.

Yeah, is kind of cheesy to use an ordinary comment thread. But a new, purpose-built feature request system is one of the many potential new features that, while a good idea in principle, would take significant time and yet have little effect on the quality of the site.

What about simply giving meta discussions (including features) a place of their own ?

That way if there is an interesting one (and it does happen, even if rarely) it has a spot and doesn't get killed because it takes up valuable space on the 'new' page.

I've been thinking about that for a while, if only because if there was an official place for meta-discussion, I could make people stop cluttering the real threads with it. On the other hand, maybe this feature would make the site more attractive to the kind of people who like meta-discussion, when I want it to be less attractive to them.

Well, since some of it is useful, give that it's own spot and leave the flag and kill options in place to control the meta discussion should it get out of hand.

Just like in wikipedia there is the 'talk' page, it's out of band but it is not out of control.

That seems to be a good middle ground.

How big of a problem is bad flagging?

It's highly localized. There turn out to be a small number of people who generate large numbers of bad flags. (Bad = flagging something that does not end up getting killed.) Which meant the solution was to be able to turn flagging off for individual users.

It might improve flagging accuracy if you included it on users' profile pages. Give people a metric and they'll tweak it. The result might be dramatic.

That's a really good one actually.

Just being able to measure how many - and which - flags were 'good' flags and which were not would already help.

I flag 10-15 stories per day and very occasionally a comment (maybe three times in all so far or so), but I'd have to track all those stories to see if they end up getting killed or not. Most of them are easy calls, the outright spam. But there are border cases where it would be good to know exactly what the guidelines are.

Especially since 'bad flagging' seems to be a problem, if there is a lack of clarity there then by resolving that 'bad flagging' as an issue might go away.

Reply on news.ycombinator.com