Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Experiment: No Comment Scores

news.ycombinator.com

Post reply on HN
Print view View original

161–170 of 327 posts

Re: Experiment: No Comment Scores

#161

Post by camccann »

araneae wrote: post #152

I actually expect that posts may become more extremely negative than they used to. At some point, you become sorry for a person with a highly negative comment; or at least I do. Sometimes I upvote comments that are negative because I think that the downvotes are unreasonable. Now there's no way for others to provide a check on the immoral downvoting behavior of others.

It seems that negative-scored comments still start to gray out, so you'll be able to upvote something you think has been unjustly penalized.

What you won't be able to do is aim things toward a desired positive score, such as downvoting a one-liner joke with 100+ points that you wouldn't downvote if it were at 20 points or less.

Re: Experiment: No Comment Scores

#162

Post by camccann »

kingnothing wrote: post #123
tumult wrote: post #102

Earlier quoted context omitted.

A simple number next to a post doesn't necessarily tell you much about whether or not you want to read it.

Sure it does. This is a community of individuals who share common interests. If the community values a particular comment, there's a high probability that I'm going to find value in it as well because of those shared interests.

And conversely, if I consistently find that the community values comments that I think are junk, I'm probably going to eventually leave the community (which is why I'm here, and not on certain other sites).

Re: Experiment: No Comment Scores

#163

Post by mechanical_fish »

DanielBMarkham wrote: post #104
kniwor wrote: post #97

But now I cant quickly skim off information from a thread. There are 87 comments in this thread at this moment. I am not interested in reading 87 comments about this experiment. I am however very much interested in seeing what the top few most insightful comments in this thread are. That feedback is very important. Right now I am feeling blind as a bat...

I feel your pain. I don't think I like the new system that much. At least so far. I'm used to scanning comments as a way to determine which articles to read. Postings with a lot of comment action usually indicate something emotionally appealing. Once I go to the comments section, then I filter by score, only looking at comments above a certain threshold (depending on how much time I have). I usually try not to commen…

This is precisely the problem. Without some idea of what earlier readers think of the comments it is much harder to skim.

The top-level comments presumably rise and sink based roughly on their karma (though the algorithm also seems to have other considerations, like momentum). But what of the second-tier comments and below? Ripostes are often where the action is.

I might find that I'm reduced to skimming for names that I recognize. That's how the rest of the web works. The web in general has no karma system, so the way to "upvote" something is to repeat it with your byline: You Twitter it or reblog it (or retweet it, or re-reblog it) with a link and your signature, implying a certain degree of implicit endorsement.

I expect that if this system holds we'll start seeing more "amen" comments. I kind of hope so, frankly: Now the only way I will spot an incredibly insightful but short comment buried downthread is if I see a bunch of comments from people I vaguely know saying "amen".

I'm not sure this trend will represent progress, however.

If we want to tinker, consider this: relative rankings on a page. Find a way to distinguish the top-voted 20% of comments from the second quintile, and the second quintile from everything else. (My instinct is not to bother distinguishing things below the top 40%, and of course negative-karma comments presumably are still doing their disappearing act, my favorite cute HN feature.)

At the risk of repeating the legendary eye-searing Orange Name experiment... we could turn the byline of comments in the top 20% a different color from the second 20%, which in turn is a different color from everything else.

Red is probably the wrong color for any of this -- too loud. Think "green". I'm thinking shades of green: Comments in the first quintile have very green bylines, comments in the second quintile have slightly green bylines, and all the other comments have their usual light grey bylines.

Re: Experiment: No Comment Scores

#164

Post by lhorie »

I'd be interested in seeing this happening for posts in addition to comments, so that there's more of an incentive to vote based on one's own opinion of the post than to fall for "scorebait" and voting on the heat of the moment because it seems everyone else is doing the same

Re: Experiment: No Comment Scores

#165

Post by camccann »

Prolitheus wrote: post #114
Radix wrote: post #105

Earlier quoted context omitted.

That would only solve one of the problems. Some of the users here are skilled enough that the average user doesn't know enough to keep up a disagreement. It's easier if the user names stay. For instance, the ggp is a security professional. I give more credence to his opinions than others.

It was only intended to solve one of the problems, as the other 'problem' is exactly what removing the names would seek to address: a person's authority/history influencing the weight of their comments within a thread. It'd really benefit people who've earned a negative following, but I don't see such a mob mentality on here as opposed to certain aggregators in which biases become apparent in just a few threads.

Yes, but sometimes a person's authority is actually relevant. Consider this short exchange from the other day: http://news.ycombinator.com/item?id=841887

On the other hand, when somebody who is deeply respected (be it pg, or any prolific high-karma commenter) posts on something where they have no particular expertise the comment shouldn't get special consideration due to the author.

I can't think of a way to reconcile these, and there's enough people who post here that are actual authorities in relevant areas that I wouldn't want to give up the first. Being able to know that, say, a comment about security was written by tptacek matters more to me than "oh look, pg made a two-word post that got 9001 points".

Re: Experiment: No Comment Scores

#166

Post by unalone »

dcurtis wrote: post #121
pg wrote: post #56

Earlier quoted context omitted.

Ok, now there should be scores next to your own comments. Better?

You know how sometimes online stores won't tell you the price of something until you add it to your cart? Hacker News feels kind of like that now. Like price, comment score is the best indicator of context, class, and quality. I feel kind of lost.

Like price, comment score is the best indicator of context, class, and quality.

Fuck that. Are you saying that you're too lazy to read what people are saying and form your own opinions? That groupthink is just going to naturally turn out better than forcing every member to think about what they're reading?

For the longest time, HN has become more and more "accessible", at the cost of a lot of intellectual discussion. I've seen a lot of conversations where one side of the argument is downvoted to hell and it makes the other side look "correct". I just saw a thread two days ago where somebody in the debate responded to a guy who was at -4 by saying "Look at your comment score, looks like I'm right." That's shit. That's utter shit and it hurts Hacker News.

I'm fine with spending more time reading a thread if it means I'm legitimately thinking about what people are saying. This is a terrific improvement to the system, and it feels a hell of a lot less claustrophobic than it did when every single thing people said was being judged as if it were an objective statement capable of being "good" and "bad."

Hopefully this also stops people from downvoting statements based on disagreement. Feels less vindictive when you can't see what your vote's done.

Re: Experiment: No Comment Scores

#167

Post by unalone »

siong1987 wrote: post #112
ivankirigin wrote: post #45

I think you can remove the word "by"

"10 points by ivankirigin" is actually quite misleading. It looks like "ivankirign" gave the comment 10 points. We are having the same problem on how to show it correctly on GraffitiGeo. We ended up using "by ivinkirigin, points: 10", which in my opinion is still not perfect. Any idea on this?

Somehow associate the points with the upvote/downvote icons and not with the username?

Re: Experiment: No Comment Scores

#168

Post by unalone »

Confusion wrote: post #146
kniwor wrote: post #97

But now I cant quickly skim off information from a thread. There are 87 comments in this thread at this moment. I am not interested in reading 87 comments about this experiment. I am however very much interested in seeing what the top few most insightful comments in this thread are. That feedback is very important. Right now I am feeling blind as a bat...

For me, comment rankings also serve to verify the correctness of my thoughts about the article or a previous comment. I believe this is completely normal and reflects the ability to doubt yourself. The fact that I initially think a story or comment is valuable, does not mean it actually is valuable: I may be wrong. As there will usually be comments reflecting different sides of the issue, I need to know what the comm…

That implies you trust the community more than you trust your own judgment.

If somebody refutes you elegantly, then you should look into their refutation regardless of whether or not the refutation has many points.

Re: Experiment: No Comment Scores

#169

Post by NathanKP »

chaosprophet wrote: post #39
yters wrote: post #30

Earlier quoted context omitted.

I don't think this'll stop people from just skimming the top. But, it will stop them from finding those highly rated comments hidden by a low rated parent comment.

Which brings us to an interesting point: Shouldn't comment threads be sorted by the total votes for all comments in that thread???

That might make sense but it could still mean that a long, pointless, off-topic discussion would rise to the top, above other shorter, succinct comments that might end up buried at the bottom.

Re: Experiment: No Comment Scores

#170

Post by unalone »

DanielBMarkham wrote: post #106
tumult wrote: post #99

Earlier quoted context omitted.

Ones near the top are some kind of product of age (more recent) and score (higher). By the way, the kind of skimming you're describing is what leads to mob voting :) People skim for the highest rated stuff, read it, then rate it up more. Plenty of good comments never get rated up, and people who say the most inflammatory stuff (or just make stupid jokes) get the votes early on and stay in the "lead" as it were, the s…

Repeat after me: you cannot have a system with points and expect people not to play games with it. Just accept it as a given and try to work around mob voting in other ways. Perhaps there is a way to make the system work against it's own mob tendencies. The entire point of using karma is that it means something -- it has value. Heck, I bet to some folks it has a lot of value. I'd further bet that it has so much value…

Repeat after me: you cannot have a system with points and expect people not to play games with it.

Exactly! A lot of games just got killed, and now we're hearing some protesting about it.

The entire point of using karma is that it means something -- it has value. Heck, I bet to some folks it has a lot of value. I'd further bet that it has so much value to some folks that you could monetize it. But that's a discussion for another day.

You don't find it slightly pathetic that people would be judging themselves by the score of their comments on a web site? And I say this as an upstanding member of the Karma Kommunity.

There is, actually, one system of points that works brilliantly. Metafilter's favoriting system, which allows upvotes but not downvotes, lets top comments get notified, but doesn't penalize any other comments, and because MF uses a flat commenting system the conversation isn't at all affected by what people like and dislike.

Post reply on HN