I'm almost tempted to install this just to see how much it hinders the experience of using HN. User identity isn't something I look at before judging a comment, and total user karma has an even smaller correlation to value of any specific comment or submission. Anecdotally speaking, I have high user karma but I know most of my comments and submissions are pretty middlebrow. As another anecdote, on the front page is a very interesting submission about Mega Man on Temple OS [0], but the creator of that program only joined after seeing his work linked to from HN (i.e. he didn't submit it himself). And the most interesting comment I've seen today is from this user with <200 karma [1].
I typically only look at the username when a commenter uses a personal pronoun and the submission involves a major company/organization/person; here's one in which karma score wouldn't have indicated as being particularly noteworthy [2]. It'd be kind of cool to see a plugin that used NLP to highlight these kinds of comments but I find that they are already "highlighted" by upvote count anyway.
I have a hybrid approach. After having an unsatisfactory conversation or reading a comment I cannot agree at all, I quit but nevertheless look at that user's comment history to see why this happened. Occasionally, I realize who was wrong was I, or that user had a good reason to say so in spite of other arguments. I try to use the history only to learn more, so I ensure that I'm no longer a part of the discussion (that has failed anyway).
As a neecomer I really appreciate that you don't judge based on the accounts Karma. I have to say that it scares to engage in a discussion with someone as experienced in HN as you are, but I can't help but to congratulate you for the amount of time you have put in an how it has turned into a great account.
As a lurker, I tend to spend time on the few comments I write, rather than post my thoughts on every topic that needs further discussion.
Unfortunately, it's the latter behavior that nets more karma, more often than not. Those with high karma tend to be those who post or comment more, separating the frequent contributors from the lurkers.
If you intend to use this userscript, please do so knowing that the numbers you're looking at have little (if any) correlation with the quality of their corresponding comments. And reconsider. Thank you.
I find the idea that more information can actually hinder you to be pretty fascinating. This is not something I consider everyday. I wonder what other situation/ scenario would this be applicable to?
I also notice people getting upset over down votes once in a while. High karma users definitely command a certain respect.
All in all, I like how HN down votes comments from even high karma users if it's out of touch with the topic or is a meaningless vent. Hope that continues.
I believe information-filtering/overload to be the core challenge and driver of inspiration when it comes to how humans understand the world. The amount of information in the world (particularly, that we store and transmit) has increased exponentially, but our attention spans have not.
`More information != Better` is evident in contemporary web design, not just the seemingly retro-because-why-change aesthetic of HN (I say "seemingly" because I don't think HN's design motivation is out of curmudgeonly inertia). Look at the single-column experiences of Facebook and Medium versus what existed in the early 2000. Hell, look at the latest startup unicorn, Snap, in which information is mostly vertical photos and videos, and in which it is impossible to even review a history of past content.
The entire premise of data visualization is to simplify information. I'd argue that data, in general, is predicated of simplifying the world. Think of how many times you've read a survey or poll or study and thought that it was wrong because you, personally, aren't anything like those in your age group. Or, if you've ever applied to college or a job, how your GPA/SAT/ACT score, or previous salary, doesn't reflect the depth of your knowledge.
The benefits of reducing information seems to also be inherent in our current biological state. The vast majority of what the human eye can see, in terms of total area, is relatively low resolution (i.e peripheral vision). The psychological tendency to rely on first impressions has been posited to have positive implications for survival. The perceived spottiness in human memory is also thought to be not just a physical limitation, but actual feature: http://articles.latimes.com/2011/may/29/opinion/la-oe-bar-me...
----
On a sidenote, I'll admit to paying attention when I'm downvoted. But that's a reflection of how rarely I get downvoted and how little I downvote other people. And I think that in itself is a reflection of HN's moderation and decorum (don't use downvotes just because you disagree). OTOH, I don't get angry. When I get downvoted on HN. I have a general faith that it was because I screwed up -- e.g. I'm not as funny/witty as I think I am, or I was unnecessarily rude to someone else. And generally, that's the case and I've felt that that feedback has made me a more considerate writer over time.
> `More information != Better` is evident in contemporary web design.
In a way it is. On the other hand I found the original Yahoo page with just about 100 blue links very readable and navigable.
I wonder if the art of speed reading has been lost. Navigating the web for me was much faster around 2000 than it is now. Too many sites have proprietary and byzantine procedures of interacting with them (example: click on a button before being able to enter something in a web form).
What I want is a way to flag comments, and threads on the homepage with comments, by users I have marked as interesting. This may be an okay proxy, but I have no idea how noisy it is since I normally don't see karma.
There's certain users that (often due to them being experts in a certain area) very often post interesting comments that I learn a lot from. I mainly find them by occasionally reading their comment history, but it would be nice to see them while the discussion is still ongoing.
A few examples: pcwalton, patio11, jblow, WalterBright
Slashdot -- as always the trailblazer in terms of comment self-moderation -- had that feature 10 years ago. You could mark users as friends or foes; their posts would be highlighted and get a (customisable) karma modifier. I don't like the terminology, but I think I like the feature.
While having inter-personal vendettas is probably bad for forum culture, it at least facilitates some kind of inter-personal relationships.
It's nice to judge each comment on its own merits, but it's also fun and interesting to recognise someone as a peer whose comments you liked -- or disliked! -- before.
It's also orthogonal to the karma system. I already have a personal (if completely one sided) relationship to a few prominent hacker news users. But outside of that circle of maybe 5 to 10 people, posts might as well be anonymous. Annotating other users makes it easier to expand this sphere beyond hacker news royalty.
(It's a joke, with poor delivery it seems. It also pointlessly replaces things in <head>. Also, it clobbers every element on the page, attached event handlers included, so it's expensive and destructive. To be clear, it's expensive because it treats the page as a string rather than using the DOM and replaces every element, not because regular expressions are expensive, which they aren't. As far as string replacements go, it's pretty quick. The underlying joke is it's a 1997 JS novice/Perl monk's attempt at replacing your O(n*m) script, which really needs no optimization, hence the dated <B> and "DHTML!"... I'll see myself out).
I'm using the HackerTagger extension in Chrome for extra context on the users. You could probably use that to some extent to highlight their posts/comments. (It will not show they commented on somebody else's post though)
Although others object to karma displayed prominently next to a username, I'm wouldn't be against it --if-- the karma was more meaningful.
The problem is that karma score combines points for submissions and comments. Those are very different activities to combine into a single mathematical scale. (Yes, I know that's also how other sites like Reddit works and a single score is "simple".)
Upvotes for submissions is saying something about the userid's curation and/or the type of topic. Some topics are "hot button" and attract upvotes even though the actual writing within the article/blog is substandard.
Upvotes for comments is saying something about userid's generative mind. His/her brain has an opinion and the fingers type it out.
As an analogy, karma combines upvotes for how a disc-jockey selects (curates) music with the upvotes for how a musician performs live music. This doesn't invalidate the DJ's effort but combining the scores means we lose more information than we gain.
I suppose the higher level abstraction we can use to justify a "composite" score of 2 very different activities is the concept of "participation score". In that way, karma==participation. In that case, does showing a participation# add useful information to the posts?
> It's based on comment scores only. My sense is that it's an irrelevant distraction and we should get rid of it. For example, we caught a bunch of users who were gaming it by deleting any comments that brought their average down. We fixed that by treating deleted comments as comments of score 0 for average-computing purposes. It's an example of how, once you publish a metric, people start to care about it and do things based on it, regardless of how meaningful it is. Comment average used to be used by a few algorithms (like comment ranking) but we turned that off as an experiment a while ago and nothing seemed to get worse. If anything, I think it may have helped a little.
The interesting "missing metric" for me is comment heat.
Rationale: I've occasionally made a polarising remark that appears to attract almost as many downvotes as upvotes. These remarks end up with maybe +3 karma after 24h of jiggling vigorously up & down but this doesn't represent the total number of votes.
With any aggregate measure, the distribution of values summarized may be as revealing as the value itself.
What I'd like is a way to hide the user globally while still tracking locally. By that I mean not knowing who a commenter is across other pages but knowing they're the same person replying back to you.
Something like HMAC(username, story_id) would work for this. It's be consistent on a given page, but not immediately apparent so people would either judge comments more on their quality or simply click the user more to see who it is. I'm curious to see which it'd be.
I'm not a fan of this. I like HN because we're all shown equally (unless you actively click on a profile link that is).
The only thing I'd like, as others have said is my own ability to highlight the names of users I think are interesting (to me) - much in the same way new users are green, maybe have my own personal list of users who are in, say, red.
I typically only look at the username when a commenter uses a personal pronoun and the submission involves a major company/organization/person; here's one in which karma score wouldn't have indicated as being particularly noteworthy [2]. It'd be kind of cool to see a plugin that used NLP to highlight these kinds of comments but I find that they are already "highlighted" by upvote count anyway.
[0] https://news.ycombinator.com/item?id=13971627
[1] https://news.ycombinator.com/item?id=13966080
[2] https://news.ycombinator.com/item?id=8736450