I can tell the difference between a stupid comment that got a lot of points for being stupid, and a useful comment that got a lot of points for being useful.
When I look at amazon and see that twilight has 3000 5 star ratings, I don't assume this means it's a great piece of literature, but it is useful information to me: it's a popular book.
Same goes for comments.
I'm not stupid, I can tell the difference between "popular because it's funny/pandering" and "popular because it's useful"
Let me make sure I understand correctly what you mean in light of what pg said about the reasons for the current experiment (quoted in my post to which you reply). Is it simply a mistake to say that high comment scores on comments that are mean or dumb is a problem? I ask, because pg has said that there has been a recent problem with comment scores.
One of the big assumptions that's being made here is that people are dumb calculators.
I think that there is a huge body of research showing that all human beings, without exception, high-IQ or low-IQ, are "cognitive misers"
http://www.psychwiki.com/wiki/Cognitive_Miser
http://blogs.discovermagazine.com/cosmicvariance/2009/11/04/...
(there is a good self-test at the link above)
http://lesswrong.com/lw/2ey/a_taxonomy_of_bias_the_cognitive...
in the sense of preferring less effort-demanding problem-solving techniques to more accurate, but more effort-demanding, techniques. Snap judgments spare mental effort for everything else that we are busy with.
Here on HN, the only thing most participants contribute to the community is good comments and good submissions and votes about other participants' submissions and comments. A few founders, pg and his core of curators (editors),
http://ycombinator.com/newsfaq.html
do actual work to keep the site running, and they get the best compensation for their volunteer efforts if all the signals about story quality and comment quality are true, thoughtful signals rather than noise.
http://news.ycombinator.com/item?id=2403696
He wrote, "The problem has several components: comments that are (a) mean and/or (b) dumb that (c) get massively upvoted."
So the founder of HN thinks that before the recent experiment there was a comment voting problem: (a) mean comments were getting too many upvotes, and (b) dumb comments were getting too many upvotes, and (c) too many of the comments that got the most upvotes were either mean or dumb or both. Let's stop and think about what that means. That means that, according to pg posting as of that moment, comment karma scores were often NOT reliable signals of good comments, comments worth finding rapidly when skimming a thread.
With that condition of HN less than a month ago in mind, how do the highest-voted comments visible in the bestcomments list
http://news.ycombinator.com/bestcomments
look to all of you recently? Are there fewer mean comments than before? Are there fewer dumb comments than before? Are the comments that are "massively upvoted" since the experiment began mostly comments that are reasonably kind and well-informed, helpful comments on the whole? In most of the treads you visit, do helpful, thoughtful comments seem to rise to a position of prominence, while mean or dumb comments gray out?
A link and comment in another recent metadiscussion thread largely sums up the back-and-forth about visible comment scores as a signal on comments in active threads:
http://news.ycombinator.com/item?id=2465357
>> Please bring back the comment scores. It helps a lot in parsing the comments and assigning a proportional weight to each when reading them.
> I had to think about this a bit, and I disagree so far. I'm finding that I'm not pre-judging comments as much. It's nice to be able to read someone's comment without knowing first that 70 or 80 or 3 other people thought it was worthwhile.
My impression too is that even with comment scores not visible, it is still convenient to browse threads to find thoughtful, informative comments, but now there is less anchoring bias
http://www.sciencedaily.com/articles/a/anchoring.htm
of most votes on a comment converging to one score level that shows up early in a thread's development, and more engagement by readers of HN in actively reading comments and upvoting (or downvoting) based on each comment's characteristics in light of the context of the thread. So far I can still find good comments quite readily here on HN. Indeed, I think that since the experiment began I am seeing more good comments more readily than before.
The main motivation stated by pg for the current experiment with making comment karma scores less visible is to "stave off decline of HN," and that is what will decide if the experiment was successful. If the previous visibility of comment karma scores led too many casual readers of HN to upvote mean or dumb comments, and too few readers to upvote thoughtful, informative comments and to downvote mean or dumb comments, the arguments on the side of reader convenience aren't going to be convincing. It isn't convenient for ANY reader of HN if the comment scores are a poor signal, and if bad comments become more prominent and good comments get skimmed right over by readers in a hurry. If a change of rules here makes every reader read comments more carefully and more thoughtfully, and vote based on comment inherent quality rather than on crowd appeal, that is a feature rather than a bug. For comment scores to be a good guide to every reader here, every reader can help by actively upvoting informative, helpful comments, and also by downvoting comments that are either mean or dumb--and especially comments that are both. As I recall, the experiment has also involved some changes in the effects of flagging, so flagging inappropriate comments is also helpful.
After edit: many comments in this thread ask about the karma rules and voting rules imposed by the software. We can all read the news.arc software ourselves
https://github.com/nex3/arc/blob/master/news.arc
if we would like to see what the rules do (except I think that maybe a few aspects of the current experiment are hidden from the current distribution of the source code), as previous HN threads have pointed out.
http://news.ycombinator.com/item?id=1307128
http://news.ycombinator.com/item?id=2034449