In terms of learning new things, the vote count helps tremendously! You can tell that a security-related suggestion earning 50 up votes is sound (of course considering context), technology-wise. I've learned a lot about passwords, plaintext, server-side hashing, salting and related best-practices solely from HN comments.
Displaying scores might make the other aspects troubling (since group-think is supported unnecessarily, often disregarding novel thoughts or disagreements). But I've learned to ignore such things, especially since most of the conflicts are "opinions" anyway, so the value addition is somewhat limited.
In terms of actual facts and expertise though, nothing can beat the vote count!
Having said that, I am still not sure which one I prefer given that there is inherently a tradeoff between the two aspects.
> In terms of learning new things, the vote count helps tremendously! You can tell that a security-related suggestion earning 50 up votes is sound (of course considering context), technology-wise.
No, absolutely not! You can only tell that other non-experts agree in some path-dependent fashion.
I'll occasionally see highly-upvoted nonsense in an area that I'm expert in. This is very bad.
This is some kind of cognitive bias.
EDIT> I should clarify that it's entirely possible for experts to disagree. So this isn't the you disagree with me so you dumb argument.
But that's a great opportunity for the expert to step in and say something valuable in reply. The high upvote count of the "wrong" response will lead to your response getting more readers.
The fact is we're already half way bought in to this "social bias". Otherwise just get rid of the voting altogether. Get rid of karma. In fact, get rid of associating usernames with comments. But I think everyone realizes, even if they don't like to admit it, that social context provides some value, even if not perfect.
No the don't, they provide a social signaling function that can be just as powerful as a vote count. When a "popular" name is attached to a post it gets read more closely and frequently voted up. If HN users were truly interested in letting the "quality" of the content stand on its own then comments would not have names attached to them and the content would truly stand on its own.
> when evaluating someone's comment it would be good to see other comments they've made that might provide insight into their biases.
In other words, let the popularity of a username influence the visibility of the content. If you are suggesting that people (including yourself) actually go back and check out user comment histories with any frequency I think you are mistaken.
Karma is not just popularity, it is also a measure of perceived authority and insight that a comment provides to a discussion. The consequence of losing this signaling function may not have decreased the quality of the comments, but it has certainly decreased the utility of the comment sections at HN.
I disagree. Since the new system was implemented the signal:noise ratio seems to have gone up. The fact that HN is more difficult to read now is a good thing; it forces people to think for themselves, and improves the overall level of discussion.
I do agree that the voting should be shown after a few days, but for right now the changes seem to be making things better.
I disagree with you. The signal/noise ratio has gone down considerably. What I'm noticing now is that people are writing longer comments, but the comments have considerably less substance.
People can't stand out based on upvotes (actual community approval) so they're trying to stand out based on post-length.
At a very basic glance it appears that what you're saying is true, but I've found the opposite.
Removing the obvious noise comments like "+1", "lol", bad jokes etc. how are you defining signal and noise?
This is my entire point: I don't care about the points when I know about the topic, it's when I'm not qualified to decide what is signal and noise that they are valuable.
I can't agree that the signal noise ratio has gone up since the system changed, though this is a subjective, unmeasurable point.
It may improve the "level of discussion" in some abstract way, but it certainly does make it less useful for individual readers.
It was common, before, for me to jump into the comments of a dicussion, check out the few highest rated ones, and then read the article with that information in mind, coming back to read other comments after that. With nothing to differentiate upvoted comments (unless they happen to be at the same thread level in the same thread), there's no way for me to do that, now. I don't view this as a good thing.
You don't seem to accept that other people may have different tastes than you. I often use the vote count to be more efficient in getting a grasp of the issues quickly, without reading everything. Great for you if you like to read everything from start to finish. Not everyone does, at least not all the time.
> You don't seem to accept that other people may have different tastes than you.
No. I'm well aware that others have different tastes.
What I'm arguing is that this feature that some people find convenient and harmless is not in fact harmless, and actually systemically degrades the site over time.
It's a systems problem, not a problem with any one individual.
If points aren't being shown, why have them in the first place? What's the point?
Points should reflect the quality of the comment. If it's not being used for that, it's worthless to have.
Showing them days later, after the discussion is finished, is also fairly worthless.
I'd much rather have some indicator of quality posts. We already have an indicator for poorer quality posts. Their is value in knowing what your peers think.
I must agree with this, I find this feature one of the most missed. It doesn't even have to show the number of upvotes, just anything that will tell me which of the posts is more highly upvoted in a relative manner (e.g. normalise all comments based on the highest ones on the page and use a 5-number/color/whatever scale).
Having the information normalized as you suggest would be great. Breaking it up into a ~5 point scale (especially if it took advantage of visual cues) would save some mental effort. Perhaps relative font sizes?
I still hold they are far superior to nothing.
Edit: case in point: I'd REALLY like to know how many people upvoted mmaunders original comment.