Have you considered adding a slight cost for submitting articles? (perhaps 2-5 points of karma?) There's a reward for submitting anything first that other people are likely to submit, but no cost, so the new page is often clogged with industry buzz that drowns out more substantial submissions. Many excellent article fall off within an hour, and then they can't get reposted.
Having more articles on the front page that aren't based on a scramble for duplicate-submission karma would probably improve the overall discussion threads, too; those posts tend to draw a lot of shallow comments.
It would also help with spam. Win win win.
I also wonder what percentage of upvotes for submissions comes just from duplicate submissions - maybe those should be counted differently (or not at all)? If the front page is already full of threads about some news about Apple (or whatever), being first to submit a redundant (but distinct) post is disproportionately rewarded, yet reduces the signal/noise ratio even further.
The problem with this is it makes every submission into a karma game. If submitting costs karma, it will cause more people to think "will this earn enough points?" before submitting links.
... but that's not what you want people to think. You want them to think "is this interesting and on-topic for HN?" They're not the same question, as much as the former tries to approximate the latter.
If people focus too much on points, you're likely to get more groupthink and more industry buzz (since that stuff always gets a gazillion upvotes) instead of interesting articles.
Certainly, people think about link karma already, but I think this would make it worse. Creating a cost will cause even people who don't think about karma to think about it, even for just a moment -- which is probably not what you want.
You want them to think "is this interesting and on-topic for HN?"
My earlier proposal (https://hackernews.hn/item?id=2242453 ): deduct karma from submitter and _anyone who upvoted_ on submissions that were successfully flagged. That should get people to think the way you mention not only when submitting, but when upvoting as well.
Improve the submissions and comments will naturally improve -- submissions that are one-sided/gossipy/demagogic tend to attract similar comments. Whereas submissions full of technical content are often rich with links to further reading, contrasting viewpoints with cited evidence, etc. (Disclaimer: observations in this paragraph may be subject to fundamental attribution error).
I think there would need to be a little more to associate the action with the punishment. Since the point deduction would be pretty delayed from the voting, voters might not recognize which votes were poorly placed, and wouldn't learn to place better votes.
I wonder about that, too, but think it'd be worth experimenting. The cost should probably be small.
I find myself wondering, "if I submit this now, will enough people be awake/etc. to notice it before it spills off the page?" That has deterred me from submitting more interesting content than costing 2 or 3 points ever would.
Also, a lot of those upvotes come from duplicate submissions. I don't have detailed info, but probably a significant percent, likely enough to lift them out of the new feed and onto the front page.
Or, maybe duplicate submissions shouldn't give the original an automatic upvote? I don't know.
Or, maybe duplicate submissions shouldn't give the original an automatic upvote?
Why shouldn't they? The fact that someone else indepedently thought the article was good to submit seems to be at least as strong a sign of interest as a click of an arror.
The problem here is that it doesn't distinguish between these two cases:
1. Multiple people independently submitted a content they found, because they found it interesting. This is (generally) good. Maybe it's not the community's cup of tea, and gradually fades away, but that's fine.
2. Somebody notices some fresh, overtly linkbait-ish article. They post it ASAP, hoping that fifty other people will be slower on the draw, netting them fifty upvotes. (Those people may be doing it for the same reason.) Either way: fluff gets submitted. Maybe it has fifty upvotes, maybe there are fifty others that just crowded out something really fascinating. This sets up a feedback loop that rewards posting things because they're linkbait, not thought-provoking.
I suspect that rewarding duplicate submissions too much tips the scales towards rewarding submitting linkbait. Looking at the new page basically ever backs this up. I don't know if adding a cost to submissions, reducing duplicate-submission karma, penalizing everybody involved for flagged->deleted articles, or what will help, but that seems like the root of the problem, so let's talk about it!
If you don't think a story will get 2-5 up votes, then you probably shouldn't submit it.
Unless a link is linkbait attention-whoring or breaking news, it's basically luck whether it gets that first upvote in its first few minutes to cause it to hit the front page. If you patrol "new" often, you'll see this constantly: tons of things in "new" are relevant articles, but still have no upvotes.
This problem only gets worse as HN grows in size.
Now, obviously, I have >4k karma and I shouldn't care at all about such a small karma cost. But the psychological effect exists regardless.
Timing can play such a critical role in an article gaining traction, that a lack of votes is not necessarily indicative of a lack of quality - e.g. right now, there is probably a worthy article on the new page that someone actively using no-procrast won't see because it is in competition with this announcement from PG.
Crowdbooster is a YC company right? They should totally implement a feature for the best time to submit an HN article like they do currently do with the best time to tweet.
If you just polled new submissions and cataloged the number of upvotes over time, you could establish the best time to submit an article. If you want to get fancy, you could add some sort of categorization to the metric and break it down by category (like: "articles about node.js seem to get the most attention when submitted Tuesdays at 3:00pm")
>Aside, I like not showing the points on articles.
I'm actually surprised at how much I like not showing points at all. I generally pride myself on carefully considering people's arguments and voting accordingly (if at all), but I've found myself reading comments much more carefully. It's too soon to tell what the effect will be in aggregate, but I my initial reaction is to hope the change is permanent.
The role of timing is interesting. It makes me wonder if a time-of-day-and-week based weighting could help out to reduce the need for posters to be so concerned about when they are posting.
What if the front page had a few random picks from the "new" page to give a little random extra exposure to undiscovered stories? They'd be different for each user...
Adding to your comment, it could also encourage group-think: "I'll only submit what I know others find cool." Note: I'm saying this as someone who submits stuff I think is very cool but doesn't get voted up very often.
I'm pretty surprised by the amount of fundamental attribution error displayed in this thread.
If each of us is so confident of our own abilities to rate articles impartially, be objective, etc., why are we putting so much thought into how to ensure that we can keep each other objective? If I believe that I can do it, why can't I believe that you can do it too?
I like the way this discussion is going. It is strange how karma could encourage people to act in a counter productive way.
1. How about not displaying the karma of new users for the first year?
2. The second insight is multiple postings from a news site. HN was good at recommending "hard to find" articles or esoterica. Where there are multiple people submitting the same link, it is probably not going to be hard to find or esoteric in the first place.
I don't get why a cumulative karma score even exists.
I like to see what other people think of my opinions and if I've touched a nerve. And I guess average comment score could conceivably be useful. But Total Karma just seems pointless. If the idea is to identify the oldtimers, then show their join date and be done with it.
The point to cumulative karma is to feed the rodent-like portion of our brain which likes getting slow, semi-random rewards. There's a tiny amount of satisfaction you get every time you look at your karma score and see that it's gone up since last time you looked at it... and a tiny amount of disappointment every time you see it go down.
Exactly. It's the same basic principle that Farmville uses for more nefarious purposes. It's a method of encouraging participation and engagement, which makes it very effective for cultivating a community of like-minded individuals.
It's probably an experiment in reputation. There was time when people would try to optimize their site's PageRank. The moment you start measuring things, magical emergent behaviors are created.
Even though it might not be the perfect way to estimate topicality for a comment/submission, karma is the best thing we have. As things are at the moment, groupthink is a huge problem no matter what rules you place on the system. You can't really stop it.
Everyone thinks about karma to some extent as it is, given that you can only do certain things with a certain amount of the stuff (downvoting, flagging, etc).
Even though it might not be the perfect way to estimate topicality for a comment/submission, karma is the best thing we have.
I don't think so. "Topicality" matches "lack of flagging", not "many upvotes". Or at least, it should -- there should be some standardized way, used by most HN members, of marking that they don't think a submission is good HN content.
This actually brings us to a very common problem on HN caused by the fact that you can't downvote links. It works like this:
Suppose you have two links, Link A and Link B. Link A is stupid TechCrunch linkbait, and Link B is an interesting scientific article. Link A is something everyone on the site understands, and 100 people upvote it. But in reality, suppose 2/3 of the people on HN actually would have downvoted it for being a terrible article -- but couldn't -- because there's no downvote.
Link B doesn't have as wide an appeal and gets only 20 upvotes, but nobody thinks it's a bad link. Yet despite 2/3 of the people on HN thinking that the TechCrunch article is terrible, Link A rates way higher than Link B.
While downvotes have their downsides, the fact that HN has no link downvotes basically guarantees that stupid linkbait industry buzz fills the top stories constantly. This is basically the HN equivalent of the Bikeshed problem.
It also gives people an incentive to submit the most linkbaity, overdramatized links possible, because the goal is not to submit good links: it's to submit links that get lots of upvotes. And when there's no downvotes, the best way to get upvotes is to get as much attention as humanly possible.
There are two types of articles that belong here on HN. The first type includes articles that speak for themselves. Everyone learns from them, but fell little need to comment unless they have something small to chime in. This type should have a low ratio of comments to upvotes.
However, the other type includes articles that provoke people and cause discussion. It doesn't just include contributing comments, but rather it contains debates and opposing opinions. These types of articles need a balance between a low and high ratio. However, the religious arguments you speak of are usually easy to find. These arguments are usually found in a deeply nested comment thread. Once you discard of fourth level comments and all the noise below, you should find that the ratio of comments is a pretty good signifier of the quality of the article.
It's very rare that there is a huge volume (>100 comments) of insight provoked by a single article, just because there's not usually that much insight to be had in the first place on that topic, and you can be sure that 100 separate comments didn't have it all on the same thread.
You usually can only get that much noise on a topic by poking people such that they NEED to correct some injustice done by the article. Sometimes good articles poke people like this, but in either case the comment thread ends up sucking.
I second this. It baffles me when I see an article with 100+ karma and 5 comments. How can something be so interesting and yet spawn so little discussion? It makes no sense.
As pg says, those are usually very good articles. I'd say the first impetus to comment is to correct, the second to add relevant detail. When the article doesn't need either, it's good.
There's a reverse problem too with complaining about karma farming eroding the SNR. I care about learning interesting things and couldn't care less about my karma on a website. I've only ever submitted two articles, and those were only after I read something that I found profoundly interesting that had just been posted (I read them right after getting an email alert that they were posted). For both, I also took the time to search HN to make sure they hadn't been submitted yet, before I hit the button. But the second one still drew complaints of being a duplicate based on prior submissions of a related previous Techchrunch post that hadn't shown up in my search and that I'd never noticed despite hitting HN several times a day, because they hadn't been upvoted at all and had quickly plummeted off the first few pages. I hope a solution can be found that cuts down on both redundant submissions and occupying comment space with annoying complaints about redundant submissions (and that obviates the need for annoying complaints about the annoying complaints).
I'd suggest trying just eliminating awarding karma to users, at least for submissions - just use the submissions' points for the ordering on the page. It takes way more effort to write a valuable comment than to hit submit for a link, so let comments become the sole determinant of users' karma. I think users would still submit pages they find particularly interesting for HN out of desire to share interesting things.
Would it really help with spam though? The rewards for a spammer to make it to the first, second, or even third page are probably worth the risk of negative karma.
As for duplicate articles, I'm not even really sure what the problem is. Is it duplicate articles (same url, slightly different) that clog up the system? Or is it the same subject being covered from many different sources? There have also been a few times within the last month where two stories about the same topic will be right next to each other on the front page--amazon/apple/other big industry announcements like to do this.
I expected that new accounts wouldn't be able to submit anything until they have a couple points of karma (from upvoted comments).
The problem is that the main page of the new feed frequently has a max age under an hour (44 min., currently) - if something doesn't get any upvotes by then, it falls off the page and becomes drastically less likely to get any further upvotes. It also can't be resubmitted for several months. Slowing this down would almost certainly be good - quite a few interesting posts get submitted at the wrong time and never get noticed.
We might take this concept further and make it a "wager" of karma with a minimum bet, and you get gains at some rate proportionate to the size of the bet. But extending it in this way, it would probably also have to have some effect on the front page algorithm for a large bet to ever make sense. Which in turn could lead to a stagnation where elite posters are constantly sinking thousands of karma points into "sure things" and nobody else can compete.
I think this would exacerbate the "comment quality" problem by encouraging people to make lots of sub-par comments so they can gain enough karma to submit their latest blog post.
Having more articles on the front page that aren't based on a scramble for duplicate-submission karma would probably improve the overall discussion threads, too; those posts tend to draw a lot of shallow comments.
It would also help with spam. Win win win.
I also wonder what percentage of upvotes for submissions comes just from duplicate submissions - maybe those should be counted differently (or not at all)? If the front page is already full of threads about some news about Apple (or whatever), being first to submit a redundant (but distinct) post is disproportionately rewarded, yet reduces the signal/noise ratio even further.