I've noticed the same thing. I'll see a debate which is basically a thread of two people replying to each other with no other contributors. In this circumstance, it's impossible to distinguish between a situation where one person gets 10 upvotes and holds a well-agreed opinion and the other person gets 1 or 2 upvotes and is arguing his opinion just well enough to avoid being downvoted.
I understand that we're supposed to form our own opinions as to who we agree with, but sometimes its just not reasonable to take the time to do enough research. Sometimes, you want to learn from someone that actually knows what they're talking about.
Without some sort of vote indicator, it's hard to tell who has the most accurate opinion, except often in subjects like law, security, and seo where there are known experts that often chime in (e.g. grellas, tptacek, patio11).
I'd suggest that a form of "fuzzy vote counts" be implemented. Something to indicate either a relative score ("this comment is substantially higher voted than its parent") or just an approximate value ("unvoted", "few votes", "many votes") without a specific score.
Without some sort of vote indicator, it's hard to tell who has the most accurate opinion, except often in subjects like law, security, and seo where there are known experts that often chime in (e.g. grellas, tptacek, patio11).
You assume that votes are a good indicator of who holds the most accurate opinion. My experience says that they are merely an indicator of who holds the most popular opinion, or who argues their point in the most convincing fashion. The latter is a particularly insidious case, since accurate information does not always go hand in hand with good debating skills.
I would agree with you. Most topics don't even really have a clear measurement of objective accuracy. And, even when they do, it's rare that votes reflect expert's analysis of accuracy. Rather, votes on HN typically reflect one of the following:
O How well known is the individual - I'm even guilty of mindlessly clicking on a grellas comment when it has to do with Law, Security with tptacek, or patio11 on all sorts of things (Japan, Startups, SEO).
O How popular is their opinion - once again, I sometimes do this myself - when I want an endorphine rush, or my day is slow, I'll go create a comment that I know adheres to HN Philiosophy, just to enjoy racking up 40 or 50 points. Juvenile, I know, but it _feels_ good.
O How effective are they presenting - this is a bit of a mixed bag, and I'm happy to see that frequently very well argued positions are downvoted into oblivion because they are nonsense. :-)
But, with that said - the first approach actually isn't too bad - grellas/tptacek/patio11 actually _are_ worth reading, and their opinions really do count for more than a random individual - so maybe the "this personis well known and has a good track record" vote does have merit.
Regarding your second bullet - do you have an alter ego to test anti-HN sentiments as well? Not really in good taste, I know, but it would be interesting as a social experiment.
I notice that a lot of posters prefix their comments with a "damn-the-downvotes" remark, which shows that a lot of disussion is skewed toward karma-preservation. Both up and downvoting are fraught and maybe you really need both or neither one.
In this particular forum I think votes are a fairly good predictor of accuratese. They're definitely better than nothing.
For instance, I know nothing about security - without votes how am I to know that tptacek is an expert at security and not just some crazy script kiddie that has nothing better to do than post here with stuff he makes up? It seems I hold this community in higher regard than you since I would infer from massive upvotes that what he says is probably correct, and not that he has good debating skills.
That's a very fair point. Unfortunately, when I don't know the subject matter at hand, social proof is pretty much all I have to determine good opinion from bad. Keep in mind that the title of this submission has to do with learning from HN, so that's the perspective I'm taking as well. If you don't know anything on a subject, the most popular opinion isn't necessarily a bad jumping off point for further research. I'm also not saying that you should ignore lesser voted answers on that basis alone. I'm simply saying that in a vacuum of other indicators, vote count works in a pinch.
You still have that rough idea of social proof in the comment ordering. Of course there's the occasional very recent comment sprinkled in, but I think the error this introduces isn't worse that popular, but slight wrong comments upvoted with herd-thinking.
I like it this way even for discovering new things, especially since the lack of comment points prevent my brain from going into autopilot mode. Trying to grasp new ideas without a concerning mind just doesn't cut it for me ...
If you look at my original comment, you'll see I used the example of a conversation with only two participants as an example. In this situation, you don't even have the rough social proof of the comment order.
Ordering is flawed even alone, because it has certain biases in comment age as well. I've seen new comments with no votes shown above other, very highly voted comments.
I should mention that I'm not necessarily campaigning to bring back the vote counts, nor do I necessarily want them back (for reasons you describe well), but I do think they did have value, and we need a way to bring back that value while minimizing the bias. I'm not sure how to do that, but I do really like the "upvote, downvote, or abstain" concept mentioned in another comment on this submission, wherein the votes are hidden until you perform one of the listed actions. You can still view the votes, but you can't influence them after you know what they are.
No one should be suggesting votes = accuracy of opinion, this is demonstrably incorrect.
What I'm arguing is that how popular an answer is is actually a valuable piece of information, outside of its direct tie to how truthful a comment is. Having nothing there doesn't help me determine how truthful a comment is more than knowing popular opinion.
To my understanding, it is generally accepted that the primary purpose of karma points on HN is to encourage good discussions. One of the most common reasons I've seen proposed by people who want to bring back visible scores for comments is "it'll make threads easier for me to skim". Another common reason I've seen is "how can I know what I should think without visible karma points?" I feel that it is worth bearing in mind that neither of those reasons is "bringing back visible comment scores will improve the quality of discussion." I also feel that (with the exception of borderline-trollish comments like http://news.ycombinator.com/item?id=2495648 and "Bring back visible scores because I'm angry today!" threads) that discussion quality has improved since comment scores where made invisible, though obviously that position is up for debate. But, if we accept that the primary purpose of karma points is to encourage good discussion (and making, e.g. skimming comment threads easier is at best a secondary purpose), and (for the sake of argument) we accept that making comment scores invisible has improved discussion quality, then it is clear that the correct decision is to keep comment scores invisible, imho.
It's not perfect. But without it, anybody can just throw in their opinion or troll comment and if you're unfamiliar with the topic, you won't be able to tell the difference.
Votes are like a simplified version of facial expressions when talking in a group. Going off of a group's reaction is flawed too, but we do it anyway. It has more value to observers than no group reaction at all.
I understand that we're supposed to form our own opinions as to who we agree with, but sometimes its just not reasonable to take the time to do enough research. Sometimes, you want to learn from someone that actually knows what they're talking about.
Without some sort of vote indicator, it's hard to tell who has the most accurate opinion, except often in subjects like law, security, and seo where there are known experts that often chime in (e.g. grellas, tptacek, patio11).
I'd suggest that a form of "fuzzy vote counts" be implemented. Something to indicate either a relative score ("this comment is substantially higher voted than its parent") or just an approximate value ("unvoted", "few votes", "many votes") without a specific score.