I personally sometimes won't bother commenting on a low comment simply because there is little chance of an interesting coversation happening there.

The other thing that's messed up now is duplicate submission. They should be forbidden again. Find a way to get the duped article to the front page if we're trying to make sure it's seen (maybe with a different decay constant), but allowing the resubmission takes away the original conversation.

Second, you'll need to produce a bit more anger and irrationality before anybody starts labeling you as a troll :)

Frankly, the only time I ever check anybody's profile is when they say something really valuable or really obnoxious. So by the time I get there I've pretty much made up my mind about the character of the individual in question. I can't imagine forming an opinion of somebody based solely on their karma score on some website.

I do see what you mean, I didn't think you were discarding comments from low karma commenters. I mostly wanted to bring up a point that is often missed in these discussions, that karma points might be very dependent on not only participation and quality, but on timing.

Given that using average karma as a filter is so blunt, an alternative way to express your method would be "I might see a high average karma as a reason to re-evaluate a comment that I initially discarded". Then it would be even harder to misconstrue into something sinister and karma elitist as I did :)

* Edit: from the comment below by mrduncan that shows how average calculation is done, my previous comment will not affect my average a lot. The max comment is sensibly ignored as a potential outlier.

- adjusting to other comments on the same post doesn't account for timing

- adjusting for time after the post is gameable by necromancers

- adjusting by the total karma of the post commented on is incomplete; it also needs some relation to the time lapse from posting to comment

- penalizing short comments might help discourage snarky one-liners

Are there browser plugins to give an adjusted average karma score this way, yet?

Either their comment is useful to you where you live today or its not. You don't have to justify your judgement by yet another subjective.

(By a... checking... 3.4 karma/comment user who was sub-1.0/comment last week; not having changed his opinions or commenting style.)

In some ways its very Karmic in that my goal is to maximize all users Karma over my own individual Karma. I get really excited when a comment of mine results in multiple 20+ Karma comments because that means I'm really contributing.

[1] http://news.ycombinator.com/item?id=1944871

I think it works like this: Take the last N values from the last T time, discard the largest (and perhaps the smallest) and compute the average of what's left. If there aren't enough, don't bother.

For more details, read the code. It's available here somewhere ...

    ; Ignore the most recent 5 comments since they may still be gaining votes.  
    ; Also ignore the highest-scoring comment, since possibly a fluff outlier.

Most people submit articles in some part because they receive karma points for interesting submissions. It's not just by magic that good articles "tend to find their way here."

Maybe we shouldn't penalize articles that don't gain attention, but it should have a determined negative score in the back end, while displaying 0 or 1 to users on the front end. The initial score could be calculated somehow through the karma (or possibly comment karma average) of the submitter and length of time the submitter has been a member of HN.

When a newer, but established, user submits an article, it could need up to 15 upvotes before it actually starts gaining the user karma/points, for example. The score could then start to adjust to it's actual total as it becomes more active.

My only issue with actually penalizing articles that don't get attention is because some articles may be submitted a few times over the course of a day from different sources, yet the later submitted ones in what I have unscientifically observed seem to be the most active. The actual information in the article is completely relevant, but due to whatever probabilistic cause, they just don't get noticed. I don't think it's "right" to penalize a submitter because no one else was on the ball, so to speak.

Jasonkeste's idea for tying reputation to comments could be done with an icon. Our (New Zealand's) equivalent of ebay does this quite well.

Perhaps we could measure meaningfulness by the size of the tree of replies and assign karma according to that. That would even solve the post-to-upvote ratio problem.

I just don't know what to do with the downvote-for-disagreement thing. Not everyone regard it as a problem, but I'm not sure I agree.

Maybe Slashdot-style metamoderation...

One measurement to make might be the votes divided by comments.

Let V = votes

Let C = comments

Let IT = 1.0

Let R = V/C

If R > IT then the article is "worth IT"

Let "the paper it's written on" be 0.1.

If R < "the paper it's written on" then it's not worth the paper it's written on (not that it's written on paper, but you get the idea).

But all this is tinkering around the edges, and doesn't feel like it's significant. Perhaps a new model of how a site like this works at all is necessary.

I'm not sure I agree with you on that. The best stories produce a high volume of high-quality discussion. Your proposed algo punishes submissions that generate discussion, which seems to be the opposite of what's desirable.

It's not perfect, but I'm finding it a pretty good predictor.

Additionally, it's very hard to have a high-quality discussion that is also high-volume. Once an article's comments get over several pages long, you can be sure that most commenters are not reading everything before commenting, so you get a lot of similar comments. Sometimes this is valuable (e.g. "Ask HN"), but most of the time it just worsens the signal-to-noise ratio.

http://www.upthread.com/

I made it for myself. I'm sure it misses some good stuff, but what does make it is almost always top-notch.

Note: there's a hidden span after the links with the score if you're curious. I have it hidden because it can be distracting.

Maybe we could get better results by identifying unanswered questions.

Perhaps you could implement a cost-refund system where it costs x karma to submit an article. If that article receives x number of upvotes, then you get them all back -- otherwise, they're gone.

For each subsequent article following a situation in which you did not get refunded, the cost doubles.

So, the scenario would be like this:

- new users can submit articles after they've attained 5 karma points, which are spent upon submission.

- If the article gets upvoted 5 times, the karma is refunded, and the cost is still 5 karma to submit.

- If the article does not get upvoted 5 times (or is flagged more than upvoted, or whatever) then the user loses the karma cost, and their next submission costs 10 karma. If the user doesn't have 10 karma, they can't submit.

There is a low entry cost, and good submissions aren't penalized, while bad or negative submissions are penalized progressively. There is no point at which the user is 'banned', but they have to earn progressively more and more karma to keep submitting. For high-karma users, this should allow them the luxury of gambling some karma on things they want shown, while lower-karma members have to be more mindful of whether they think this is worth gambling their karma on.

So if you do implement a system like that you're basically codifying the herd mentality much more strictly than it probably should be.

I would also venture that 'cutting edge' is HN's bailiwick, and therefore, it's really hard to get ahead of the tech curve here.

Incidentally, i submitted a link (http://news.ycombinator.com/item?id=2386915 ) about an hour ago, and after watching it depressingly fall off the first page of the new list, i think i'm giving up on ever submitting a link to HN again.

What's so frustrating is that i can identify content that is good, and relevant (the link above is about open data, citizen access to data, and information management), which are obviously relevant to the hacker news community.

But the only ways i can think to remedy this would be to either lobby my friends hard to vote up the link (something i'm loathe to do), or start pimping the links on other sites, in the hopes that people will try to submit it, see the existing link, and vote it up.

There's just no way i can do better than a piece of content like this. Oh well. I guess i'll stick to commenting.

  >  i submitted a link ... about an hour ago, and after
  > watching it depressingly fall off the first page of
  > the new list, i think i'm giving up on ever submitting
  > a link to HN again.

  > What's so frustrating is that i can identify content
  > that is good, and relevant (the link above is about
  > open data, citizen access to data, and information
  > management), which are obviously relevant to the
  > hacker news community.

That doesn't mean it's of zero interest to everyone. That doesn't mean it should be of zero interest to everyone. It does mean that you are mistaken in your belief that founders and hackers would be interested, if only they'd seen it.

You're going to fall into at least one of those categories, and i think that hackers in particular are going to fall into multiple of those categories (that i happen to know from personal experience).

But if you didn't think it was relevant, then i certainly don't think you're obligated to upvote. :)

And that's my point.

It should be of interest to everyone, and it's got nothing specifically related to being a hacker, being a founder, running a project, or other technical issues that are the point, purpose and focus of this community.

Or once were.

And before someone trots out the "satisfies intellectual curiosity" line from the guidelines let me just say no, it doesn't, and if you think it does then you have a very different concept of "intellectual" from mine.

So I'm not saying it's uninteresting in general, I'm saying it's uninteresting to this community as I understand it ought to be.

And in consequence it's clear that you have a different understanding about with this community ought to be. Maybe I'm now out-of-date, and it's changed into something with which I don't share significant overlap.

At the same time, if you don't find the generalist argument persuasive, i'd fall back on the point i skipped previously, which is that this is of direct interest to data geeks. One of the major information brokers, which is interesting both as an individual instance, and as a representative of all large information brokers is fundamentally broken internally.

This is a data processing and scaling issue laid out in a way that should make us ask what's wrong, and what it is that can be done to fix these sorts of problems.

Put more succinctly, this is a problem in our domain.

I think you're getting into a long-deplored mailing-list tendency for people to ascribe general attributes to their community based on personal preference. "I like ice-cream, lots of other people probably like ice cream, so it's a good post for HN." Journalistic feature-creep, if you will.

My point being that yes, we are all tax-payers more or less, but that doesn't mean everything paid for with tax revenues is germane. Syllogistic posting rationales tend to cast a wider and wider umbrella, sometimes venturing into the realm of slippery slopes.

EDIT: fixed quote-italics

Additionally, i do feel that i have some sense of the hacker news demographic having both been a part of it, and watching the sort of content that does generate discussion.

So i'm cognizant of the problem, and i'd like to think that gives me at least some ability to compensate from the bias.

1) Match peoples' interests

2) Write really really well

So, you can either submit articles like "BubbleTwit (YC '08) gets $14B in funding, announces Natalie Portman is new CEO" or you can just keep writing and writing and writing until you get really good.

I'm mostly suggesting this because it's what I wish I would do. I'm not that good yet.

I peeked at your post... I think it is a fascinating subject, but your title and opening paragraph don't work that well. "Sustainable" doesn't mean anything without context. Your opening paragraph doesn't really create any kind of desperate need. What is the specific crash going to look like? Paint us a picture of destruction, and then lead us out of it. Great writing grabs people and holds their faces in front of something real.

I don't doubt that you've got real things to show people, so just keep working on the craft, and you'll get there.

credit where credit's due, i didn't author the original piece, and there are several things i would change about it if i had.

But yeah. i should blog more too :P But self promotion is frowned upon at HN.

The only cure I know for that is to lower the signal to noise ratio, and I don't know a better way to do it than to make submissions harder, or at least, more chancy for the shotgun-submission approach.

If every submission is something that people care about, then I believe (again, right or wrong) that more people are going to be willing to hit the 'newest' page instead of waiting for things to hit the home page, and there should be far fewer submissions.

Of course, this is all predicated on the assumption that auto-post bots are dumb, and that they'll lose you more karma if it's penalized than you gain, but that's an assumption I can't really stand behind, not having seen the numbers.

I think this is actually a pretty complex issue (social dynamics, reputation, incentives) so let us keep Mencken's Law firmly in mind

  For every complex problem there is an answer that is
  clear, simple, and wrong.

In the scheme I'm trying, it costs you karma to upvote or downvote, which gets converted to non-spendable clout. You get a small portion of karma everyday, which increases as you have more clout (and followers, another aspect I added). If you have spent all your karma, you cannot submit stories until you earn some. Here I was trying to limit the effect of users that are new, and of those that have not contributed appreciated content or comments.

That said, you've got me rethinking the idea of karma for submissions... It might be better to have submissions only increase clout. -Therefore, it will only slightly increase a user's karma-earning rate. Good comments will be the best way to get karma. Yes. I am going to try that. Thanks!

It just foolish that some high profile story give someone lots of karma advantage then good commenters because someone submitted it first.

  > Any chance we can disassociate upvotes on articles from user karma?

Interestingly, Oxford University used to mark final exams for Mathematics and associated schools (such as Computer Science) in this way. Questions would typically start with parts that were largely "book work" and move on to parts that required more careful thought. In addition, you could answer as many questions as you like. Therefore, to avoid people getting lots of marks by just answering all the book work, the papers were scored by squaring the scores for each question and then summing those squares.

That is: overall_score = sum(q in Questions | score(q)^2)

This practice stopped in ~2003 (now you can still answer as many as you like, but the mark for the paper is the sum of your top 3 marks for individual questions, so it's only really worth answering 3 as well as you can).

The only case I could see where your point is valid is if "karma abusers" withhold lower quality submissions in the hopes that they'll submit a really popular one. Do you know if they do this? Or am I misunderstanding the rationale?

Assuming that people are chasing karma (let's leave why to the side for a second, because I don't understand it either), then it's far more efficient to submit lots of low-scoring stories than to even bother thinking about high-scoring ones.