Hacker News Reader

Comment by rlvesco7 | Hacker News Reader

"We're not going to fix this problem. I'm also fine with that."

Just thinking out loud here in terms of potential ways to stem this although perhaps this is futile.

What if HN had a checkbox for ad-hominem and/or off-topic comments?

Check boxing a comment as an ad hominem would add to your "ad hominem" score (separate from karma). Likewise for off-topic. This wouldn't stop a person from posting ad hominem or off-topic comments, but the idea of accumulating points for bad behavior could dissuade (some) people from making too many. Who would want to be branded the off-topic or ad-hominem gal?

Perhaps it wouldn't work -- but at minimum I think it would be an interesting experiment in promoting civility and thoughtful discussion.

View on HN

Because we're a community of geeks, these threads always fill up with random technical countermeasures. I am dubious that any of them will work. I think showing other people's comment scores was a bug and I'm glad we fixed it†, but I don't think we need more features to address the cultural issues.

I do think we'd benefit from more guidance, but 'pg is pretty hands-off on this stuff. Graham could say today "no Apple vs. * debates", and about 100 people on HN would immediately start actively shutting down pointless Apple debates.

But he doesn't want to do that. I'm sure there's a reason for it. Part of it is probably that no matter what the ratio of light to heat, YC benefits from having HN active and perceived as open and unmoderated.

† Not interested in debating this; take it to 'pg

The system he is suggesting is exactly what /. had, and that didn't work out too well. I have a theory that the decline of any great community begins with a conflict. In the case of /. when the SCO matter started, the issue was so black and white that there was no need of discussing it - and the community itself was so homogeneous in their opinion on this matter that there was no real discussion going on. In the case of HN, I think we're seeing a variation of the same thing - it's Apple vs everybody else, and the issue is so polarising that the discussions are almost never real discussions, they're just a bunch of people trying to prove they're right. And that is where the rot starts.

I am as guilty of this as anybody else - since I find myself rarely able to comment on any technical discussion on HN, most of my posts have been on these topics, where one does not need any more than an ideological standpoint from which to talk. Perhaps the recent precipitous decline in the quality of the comments is directly linked to the kind of stories that we've been discussing, and the kind of people who tend to comment on them.

> Graham could say today "no Apple vs. * debates", and about 100 people on HN would immediately start actively shutting down pointless Apple debates. But he doesn't want to do that. I'm sure there's a reason for it.

Yeah, because having a bunch of righteous do-gooders flagging topics as inappropriate isn't as much of a feature as you'd think.

I think we should all have the ability to add arbitrary meta-data to other posts and to write/tweak our own scoring system to recognize the pieces of other user's rankings we saw as useful.

So basically implement a system like Slashdot?

I think that the vaccine HN had was size. Just like a modern republic vs. ancient Athens, HN has grown to the point where the obvious answer just doesn't seem to work. I think that Slashdot has already had to deal with these growing pains, and HN(and reddit) chose to ignore /.'s problems and pick the obvious solution. Unfortunately the behavior of mobs/tribalism tears this choice down.

Which limiting yourself to a set of must-read users and not consuming all of HN might not be a replacement for that vaccine, but it sure helps keep the plague rats at bay.

I'd very much hope it's not just like Slashdot.

Any strict voting/counting based system is going to be biased to highly popular (vs. interesting) articles.

A Likert-scale ranking system (vote on some scale, usually 1-3, 1-5, 1-7, etc., occasionally even (to exclude uncommitted median votes), and including a multiplier (log-based) for popularity) seems to be one of the more reliable ways of meaningfully indicating general interest in content.

Tracking the variance / standard deviation in votes is also helpful in identifying whether an item is of low interest (or disliked), vs. polarizing.

Reply on news.ycombinator.com