I've found similar thing. What surprised and frustrated me sometime is my "shallow" one liner got a lot more upvotes than my longer, more thoughtful comments. It seems people have extreme short attention span and won't bother to read the longer ones.
This is simply because long comments very rarely provide more value than short ones. In fact, I would wager that shortness is a good quality that shows the poster tried to condense his point into its true essence.
Example: The parent post could be ~ 5 lines and contain the same point. I really like the way 4chan handles this sort of thing: The so called "green text". On 4chan it is unspoken rule that nobody cares about you. In order to make your text interesting enough to be read, you have to make it as short as possible while still carrying your point across.
The inherent problem with this is that certain posts simply cannot be made shorter without omitting useful information. When users are expected to write short comments, you'll quickly find that one-liner jokes and the like are the most upvoted, not the ones with actual content.
While 4chan may be a fun way to pass time, it's hardly the place I go when I want to read/join intelligent discussions or learn new things.
The degeneration of HN is likely due to the influx of new users from 4chan and Reddit, where comments are very short and often appeal to the lowest common denominator.
[S]hortness is a good quality that shows the poster tried to condense his point into its true essence.
This is often true, but I think the value of brevity comes mostly from the context, a threaded comment page. In that context, users expect a discussion, not a monologue. They're not just looking to passively absorb information by reading, they're looking to sharpen their own understanding by actively articulating a response. That's harder to do when you're trying to respond to a comment that makes multiple points. Better to break a long-form comment into several more tightly focused short comments so that people can more quickly pick out the topics they want to respond to.
Personally, I try not to spend too much time on HN (sometimes I enable noprocrast), so I need to quickly decide whether it's more worthwhile to read a single ten-paragraph comment instead of ten shorter ones presenting different points of view.
Usually, if the first paragraph is not absorbing enough and there is a risk that the rest is going to be rambling, I simply stop reading without any regrets. For me it's not an attention span issue, it's a limited time issue.
I've noticed this too. Unfortunately, my two most upvoted comments recently have been of the "throwaway" variety. Not mean, not trollish, but a couple of easy-to-agree with quickies that I banged off on the bus ride to work. One[1], perhaps legitimately upvoted so that the response would be more visible, was a simple statement that the submission was a couple years old and it would be interesting to see some newer data. The other[2] was a cheap, throwaway anti-SOPA line.
The danger here is that the easy-to-upvote throwaway comments become a self-fulfilling prophecy. Like it or not, karma is social validation. A long, thought-out reply that gets no upvotes is somewhat depressing; it makes you think that, even at your most thoughtful, your thoughts are unappreciated. Contrast that with a cheap and easy line that gets 40 points, and the social feedback is clear: your considered opinion isn't valued, but your cheap lines are.
It makes me think that the karma system itself might be fatally flawed for its lack of scalability. What works for a small, mostly homogenous community breaks when that community becomes larger and more diverse. We can deny it all we like, but I'm sure most people feel pretty good when they log on and see their karma significantly higher than it was the last time they checked. It's an ego boost, and few people are immune to enjoying that. When playing to the lowest common denominator gives you that fix of social validation, and a more thoughtful comment doesn't, it's pretty strong positive reinforcement for the less-desirable behaviour.
I'm not sure how to combat this in a scalable manner. It's difficult to think of examples of community-moderated forums maintaining quality in the face of rapid expansion. My previous suggestions of weighting the votes of those who consistently upvote quality comments higher than those of people who upvote fluff has met with little traction, and I can see some pretty big potential weaknesses there, myself.
In reviewing this comment, however, I have had a thought of something that seems ridiculously trivial, but might just work: pg has stated before his belief that comment length is a relatively good indicator of quality. I notice, however, that the comment box is relatively tiny. It makes a long, thought-out post appear, prima facie, no more substantive than a 3-liner. I would be interested to see what the result of making the comment box significantly larger would be. Would people, upon seeing that much empty space surrounding a throwaway comment, be likely to reconsider posting it? Are there easy cues like this that can be used to hack the behaviour of commenters? I'd be interested in seeing the results of running an A/B test on a seemingly simple change like this: Make half the users' comment boxes two or three times as long. Leave it that way for a few months, then take a random sample of the resulting comments and see if there's a difference in quality. I wouldn't be surprised if there was.
I don't know that this social problem has a technical solution, but you're right that the "points" system encourages shallow comments. Unfortunately some people will always see a number as a score, and start gaming the system to increase it.
The problem with many submissions recently has been they are taken directly from reddit, and have content that sits well with reddit but not HN. The "Programming prodigy passes away at 16" story is a good example: it's tragic news, but the fact that someone has died is not necessarily a good HN submission. Although much of the discussion about Erlang and Haskell went right over my head, I think this site did better when those stories were oft-submitted, if only because it tends to push away the people who would prefer fluffy "human interest" stories. I personally have no intention of creating a startup but find pointers to useful technical ideas and tools here - more of those, please
So I think we need to concentrate on the submissions, removing stories of marginal interest to hackers, and being diligent in upvoting good stories and comments, and downvoting crufty comments.
Yeah, submissions may be where the solution lies. I'm not sure if it's just me but has anyone else noticed that the New page gets completely saturated with spam? I have show dead turned on and I used to see a handful of dead stories. These days it's not uncommon for me to see 5 live stories with the rest having been killed. A month or so ago there was a discussion about HN's rank on Google. I'd be happy to see it close to the bottom of the first page if at all.
"It makes me think that the karma system itself might be fatally flawed
for its lack of scalability."
The motivation to be part of an online community should be the enjoyment
at the discussions and not the number of points.
I hate the points, and I hate it, that I look after them. I think that
just the presence of a karma system has a bad effect on the quality
of the comments. You're getting what you ask for.
It's symptomatic how often I read 'Please don't downvote me' or
'I hate to say this, but' in the last time.
