zlacker

That wouldn't do anything currently. Only really atrocious comments get flagged, and they always have huge negative scores. Though of course if flagging had more effect, maybe more people would do it.

Another thing I've considered is having specific types of flags on comments, and having them have different effects. E.g. there could be a flag for incivility, and if you got enough of those (maybe in proportion to your total number of comments) you'd actually get kicked off the site temporarily.

>>pg+(OP)
Something to keep in mind though is that, as the size of the user base here grows, there's going to be an increasing number of people who feel comfortable using "this button will contribute to banning this user"-type controls less than judiciously.

>>pg+(OP)
Make flags public, and their outcomes. "16 people flagged this comment." "16 people flagged this comment, and it was tagged uncivil. User was banned for 2 days." "3 users incorrectly flagged this comment, and have lost flagging privileges for 2 days."

alex is right that automatically escalating based on #flags is a blunt instrument. You'd need to perform the escalation manually, you'd need multiple people doing it, and you'd need the decision-maker to attach their name to it ("kn0thing marked this uncivil") so the watchers can be watched in a lightweight manner.

>>pg+(OP)
One transition from "current" state to "flag does something" state might be to make the flag button more obvious, to make it have a more immediate effect, and to encourage its use.

My perception of the flag button now is that it's the bat-signal; it means that something is so bad that it actually warrants individual admin attention.

>>pg+(OP)
One of the Dutch news sites I frequent also has qualitatively different forms of upvoting. E.g., you can upvote for humor, upvote for insight, upvote for new information. By applying different metrics to different kinds of comments, it might be easier to create the kind of balance you want.

>>pg+(OP)
When I see something dumb and/or mean that has been highly upvoted, I don't consider flagging it, because I don't think it will have any effect.

However, I don't see an algorithmic way to use flagging in such a situation. The new control mechanism would be subject to the same problems that led to the comment being upvoted in the first place.

I'd like to see the mods experiment with a more authoritarian approach. When a comment sucks, I want the mods to impose their will upon the populace, informing everyone that the comment sucks, regardless of its upvotes. Flags could play a role.

>>dochtm+V4
What site is that? I'm curious.

>>pg+(OP)
There's a lot of discussion here about comments and votes, but only one or two remarks in this thread note the clearest differentiator I see between "old guard" and newbie users: the meaning of the up/down arrows.

Old guard would prefer to vote on a "Contributes/Detracts" axis, while new users vote on an "Agree/Disagree" or "Like/Dislike" axis.

Arguments about this erupt in threads, with newer users generally saying, "It's a democracy, this site is what the majority want it to be, and if most of us want it like this, your loss." But such an approach devolves into pop pablum.

I believe most other ideas here would be unnecessary if the meaning of the up/down arrows could be resolved either socially or algorithmically.

The "correct" solution would be to offer a quadrant, with contributes/detracts on one axis, and agree/disagree on the other. But that would require a rewrite.

Instead of a rewrite, I'd experiment with temporarily changing the arrows to say something explicitly supporting well reasoned comments:

  17 points by uptown 1 hour ago [ contributes | detracts ] link | parent | flag

Some would undoubtedly still interpret these as a rightness axis like "agree/disagree", so perhaps an even more familiar pair of terms:

  17 points by uptown 1 hour ago [ content | spam ] link | parent | flag

However, the term "content" might lead to voting up every valid content remark, so the positive word should be something with more of a value judgment, while still being a word that can apply to points of view with which one disagrees:

  17 points by uptown 1 hour ago | interesting* | spam | link | parent | flag

The idea with the vocabulary choice is that a neutral comment would not be clicked on, and "interesting" is directly in the HN charter.

I think this type of vocabulary is more in line with the desire to see well reasoned or contributory discussion flourish.

* Mouse over the word "interesting" could tooltip: "This comment made me think."

// This account is ~400 days old, but a prior anonymous account is ~800 days old, giving some perspective on the trend over time.

>>pg+(OP)
I think the idea of categorizing comments is good. I would cite as a somewhat-working example Wikipedia's article flags. Wikipedia has a serious problem with deletionism, but they have ways to mark articles in meaningful ways, like too long, or too short, or irrelevant, or excellent, etc. They can also mark individual contributions as "citation needed" (a flag I would really like to have on comments, see http://news.ycombinator.com/item?id=2423582 as an example of a citation-needed comment getting, I feel, way too many upvotes).