I agree with the halo effect. On the other hand knowing that you are building a brand in your handle gives motivation to put in more effort. Not everyone thinks like Job's father ("you will know that the other side wasn't painted" (or whatever that quote was))

I thought an interesting idea would be to run a test of the top karma-ers and have them post replies w/o their actual handle and then compare the karma rating they get.

There are of course a few problems with this.

If the comment is of high quality the fact that it's associated with a brand new handle might paradoxically get it higher votes. "Hmm this new guy knows his stuff.." Add: The other issue is that there may be a little cousing of Hawthorne effect on the part of the commenter. Trying even harder to prove a point that it isn't their handle.

Sometimes the top comment is from someone where you recognize the handle. And the comment is good. But it doesn't seem that good. Same many times PG says anything. He gives brief answers and the way they rank sometimes it's like SCOTUS is talking to someone low on the food chain in the legal profession. (Same with what Fred Wilson says on his blog btw.)

I don't typically upvote anything PG does. To me it's like tipping the owner! And it's not like he needs the karma score. Off the top I don't ever remember seeing a PG comment that is not near the top of the thread or subthread. (I would love if PG were to do the anonymous handle idea and then publish the results.)

I downvoted a recent comment by pg that for quite a while was (and STILL is) the top comment in its thread

http://news.ycombinator.com/item?id=4253156

but that pg later acknowledged didn't have a sufficient factual basis.

http://news.ycombinator.com/item?id=4254592

Conspiracy theories are very enticing to the human mind, so they often get too many upvotes here on HN. Although I did downvote pg's first comment in the thread mentioned here, I was outnumbered by many more people who upvoted it, mistakenly.

Doing so gets people to read both. Besides in the absence of the original comment (incorrect as it may be), the followups lose context.

Reserve down votes for opinions that do not deserve a reply. Such as pure trolling.

I have seen familiar (absolutely not famous) handles from IRC on HN on two occasions. Made me happy. I can't turn this into a more rational point, though.

In a perfect world I would research each comment, do basic background checkes against stated facts, think through the implications of the comment, etc. and then form an unbiased opinion. I would do this with every comment thread. After I had done this I would be uniquely qualified to form an opinion on every comment, and I would be wise because I'd read and understood all the comments. I'd also have spent most of my day doing so.

The reason comment scores, and usernames as a proxy for scores, are interesting is that it's a good way of cutting through the noise. Most people just don't have the time to read through all comments on a thread and form an opinion on each one seperately.

That's what I'm really missing about the points - I don't care how many points a certain poster or a certain comment has, but I do care greatly to have an indicator of quality so that I can skim through a thread and pick out the best comments. Particularly on long comment threads where I don't know much about the subject, or on busy days.

The reason comment scores are bad is that it leads to leads to anchoring. http://en.wikipedia.org/wiki/Anchoring This is probably why dumb and/or mean comments get upvoted (as discussed in tokenadults top comment).

This seems like a benefit to me. If I'm skimming through a comments page on a topic I'm not fully familiar with, seeing an opinion I would otherwise disagree with voted up should make me update; it's evidence against my current opinion.