Poll: Which S/N improvements do you think would most improve HN?

34 points by tptacek ↗ HN
Special bonus round: think of something else (something simple! extra-special bonus points if you read news.arc before suggesting it!) and I'll add it to the poll.

Super-bonus extra special bonus round: if you've been here for more than 500 days and think HN's S/N hasn't degraded over the last 6 months, tell us all why! My perception is that most veterans think it has.

103 comments

[ 5.6 ms ] story [ 189 ms ] thread
I wonder if there would be any benefit in adding a small cost to voting (1/10th of a point? different costs for up and downvoting?). If nothing else, it could be an incentive against drive-by downvoting.

Also, why isn't flagging for comments more visible? I suspect many are unaware that it's even possible. I routinely flag "obligatory xkcd", "looks like your database isn't scalable enough to host your blog post about database scalability", "correlation does not imply causation", and other recurring noise.

How would that second feature work?

To be honest, I don't even know how comment flagging works. I know there's a button there (it's buried and shouldn't be), but I have no concept of what happens when I hit it.

To whatever extent it does work, I think an automatic N-week hellban after getting more than one comment flagged off the site makes sense.

More awareness that you can flag comments, such as showing it next to 'link' on thread pages. I rephrased my comment to make it clearer.
I am aware of the flag link myself and where it is located but I have no idea what it really does so I never touch it.

(Not that I would touch it if I knew, I am averse to punishment as a form of socialization as I think it's less effective than encouragement and gentle correction.)

It doesn't punish, it draws moderator attention (though too many flags on an article before N upvotes kills it, IIRC).
I would prefer no karma for story submissions.
If you look at the people who submit off-topic stories and the people who drag comment threads down, (a) is there overlap, and (b) is there enough concentration among the submitters so that depriving them of an occassional bucket of karma points would matter?

Remember, all the threads that germinate from these stories tend to generate more karma in the aggregate than the submission itself.

Preface: I don't know if this actually happens or not, but...

My guess is that people submit stories from certain sources, no matter what the content is, to try and be the first submissions and get the karma. If ~15 people do this, the submission is going to rise rapidly, not because of it's content, but because of people trying to cash in on a well-known name. For instance, I think an Arrington article is going to get a lot of votes right away because of people trying to be the first submission, not because of the content article. If there was no karma associated with submissions, people wouldn't race to get these types of articles submitted.

Again, just a guess.

> My guess is that people submit stories from certain sources, no matter what the content is, to try and be the first submissions and get the karma.

Looking at the submission history for some users easily confirms this.

Along with other adjustments to incentives for submitting content, some sort of check for whether most of the user's submissions come from the same domain name could potentially be useful - especially for domain names with the top X% of all submissions site-wide (e.g. techcrunch).

Whether eliminating submission karma or just tuning it, that's not a bad idea - that could get rid of the incentive to be the first to post techcrunch etc. stuff that may be voted up.

So much good stuff spills off the front page just because it didn't get several votes within the first half hour (and/or wasn't submitted at prime time for EST or PST readers).

1) Separate user karma for their submissions and comments

2) Weight upvotes on comments from people with high comment karma more than people with low comment karma

3) Weight upvotes on submissions from people with high submission karma more than people with low submission karma.

4) Remove upvote counts from stories and comments, or display a function of the weighted upvote count.

5) Display user's comment and submission karma on their public profile.

Probably too complex of a solution, but it's an idea.

I took (1) and added it to the poll. 10+ extra-special bonus points for you.

(4) is already up there (hide some or all scores). This is my most wanted feature for the site because holy hell do we all waste a lot of otherwise valuable cognitive resources thinking about the little numbers next to comments.

(2) and (3) worry me, because HN is an echo-chamber site. I get way more karma for the same unit of comment quality than you do, and it's not because my opinions are any more valuable than yours.

I like the idea of hiding scores, with one caveat: Ask HN. Those posts often end up taking on a StackExchange-like manner in which upvoting for agreement makes sense and you really do want to know which responses got the most votes.
During the discussion that sparked the creation of this poll, I too suggested weighting upvotes by the karma of the upvoter. Thinking about it further, I still like the idea, but it would have to be done properly or it could be easily gamed. Example: a small group with relatively low karma could simply upvote every post and comment made by others within that group (a complete directed graph of upvoting). It would not take a particularly large group to be able to artificially inflate its karma at a greater than linear rate.

The quick (and probably flawed) solution I came up with was to have a tiered base amount that is determined by the karma percentile of the upvoter (eg, 75th percentile might get you an additional 2 points, 25-75 1 extra). Furthermore, there is an additional potential bonus based on the karma delta between voter and votee (ie, upvotes from top HN'ers are more valuable to newbs than to other top HN'ers). As an example, we'll say that we add one point for every 25% tier between the voter and votee (no bonus if they are the same). Using this example, a top-tier vote would be worth 5 points to a newb but 3 to a veteran.

This would allow high-quality posters with more recent accounts to quickly gain the necessary karma for voting, polls, etc without causing such massive karma inflation on the high end.

I disagree with the "500+ karma required to submit "Ask HN's" " suggestion. Sometimes people need help and they can't ask with their usual profile. A recent example: http://news.ycombinator.com/item?id=2213417
So allow an anonymous "Ask HN" from a URL not linked to the site. /newpoll is the same way, isn't it? You have to go look for it (I had to, just now; I've never done a poll before).
Heh, you could allow crowd-funding of karma for Ask HN posts :)
Re "Make the "flag" button more powerful" - meta-discussion is for the most part off-topic; rightly so, it usually only generates noise. Yet I wonder whether downvoting or flagging is all that useful as feedback for (relatively) new users. I don't want to add to the meta noise by replying publicly, but they won't see the fact that I've flagged it. Worse, I've recently seen quite a lot of irrelevant, ill-informed, no-content, off-topic or otherwise useless comments upvoted a few times. Downvoting a noise comment from 5 to 4 isn't going to cause the author of the comment to think about posting something similar again.

I wonder if attaching some kind of reason to a downvote would be useful, or even being able to reply privately (visible to the original author, admins/mods and maybe other downvoters?).

Of course, it's possible that it's my judgment that's off, but I get the feeling I'm not the only one.

Oh, and: I've had an account here for ~3 years and lurked for longer, and I'm still not sure I'm using the "flag" feature correctly. I also don't know what happens when I press it, exactly. This sort of thing really ought to be more obvious.

It's not clear to me how any of these would improve things. Not saying they wouldn't, just saying I don't see it right off.

What does flagging do? Remove a post? What would more of it accomplish? What are some sample posts on which it would be used? Are the dangers that such a mechanism would be used to punish unpopular opinions, or even punish correct statements when the consensus is wrong?

If you're saying you don't see an S/N problem on HN right now, that's a totally valid opinion, but you should express it directly. As it is, I can't tease apart from this comment whether you're saying "things are fine" or "these features won't help with the problem I, too, see".
I'm not the poster you asked, but I think I agree with him and interpreted his post as a rejection of the proposed solutions, not a rejection of the assertion of a problem.

Having only been here about 500 days, I'm newish, but it seems like quality is degrading. I don't see why your solutions would fix that. The only thing I can think of that might be interesting to examine is advogato's model.

In the end, it's a hard problem. People want to hang out and discuss news items with all their friends, not just the ones we wish they'd bring around to show up.

I buy Joel Spolsky's take on this, which is that there are software features that improve communities and software features that degrade communities. They may not be the sole determinants of community quality, but they exist. Do you think Hacker News has identified all of the low-hanging fruit already, or do you disagree with Joel Spolsky on this point? (Or is there a third option I'm missing?)
I agree that better features field better community, I only disagree with all the features listed in the poll. I don't have a better concise idea right now, but I will think about it sometime I am not waiting in a foreign airport and see if there's something I can suggest.
I did not say there was not a S/N problem and that was not the focus of my comment. The comment "if you're saying" when I wasn't feels a lot like an attempt to discredit a comment by reframing the debate by assigning statements of intent to a post that clearly are not there. I hope that is not the case as such tactics are not productive.

My comment is that I read the proposals in the form of a poll and do not see what they are expected to do to solve problems. Perhaps more explanation of their expected virtues is in order. I asked about one in particular as an example, the flag.

I know flagging from youtube, it is used much for erotic and violent content which after being reviewed manually by employees of youtube (which undoubtedly has significant labor costs) has a warning label assigned and age sign-in or click through added. Perhaps it means the same thing here. If so it's not clear what sorts of posts it would be used on since I have not seen violent or erotic posts here, or even spam. So that is why I asked how does it work and please give some examples of actual posts is should be used on. Also discuss how introducing new punishment features will not have unintended consequences. A discussion of these sorts of things should precede any vote on what actions to take, that's how they do it in various congresses and parliaments.

(Regarding the other topic, I don't know if there is a S/N problem or not. There are different kinds of posts than there were a year ago. There are more people here and different people. This is not what my comment was addressing at all though. I assume as a starting point that there is a problem with wrong posts and here are some proposals to solve it. I don't know if that is the case but I am assuming it to move on and discuss proposed changes to the board. The question I had was how will these solve the problem. The mechanisms of how these would work to improve things is no doubt clear to the OP, but is not obvious to me. Then again, since we are being asked to rank them, perhaps it is not so obvious even to the OP. If it is very obvious and clear to all we would just be told these should be implemented, and not asked which ones we have a gut feeling might work.)

My HN account says that I have been around 983 days. Probably much longer before I signed up for an account. I do think the quality of comments has gone down. More people commenting without putting much thought into it.

I am against putting Karma restrictions to upvote or post Ask HN. It insulates us from attracting newer and possibly better members.

I think it's an incredibly valuable feature of the site that it draws in people like Joshua Schacter and Jason Fried†. So I'm absolutely with you in the spirit of this comment. However, not showing Jason Fried the "upvote" button and not showing him the "Text" box on submissions isn't going to keep him from contributing. Candidly, I'm guessing he could give a shit about those features.

Two examples of people who weren't here at the beginning of the site, both of whom are notable company operators.

There's a reason people comment without putting much thought into it: it's because there is an incentive to do so. The earlier you post a comment that simply points out something obvious, the more upvotes you get, so people do that.
None of these address the real problems, some of which are:

* infographic spam still has an appalling attraction to many HN users who should know better * the 'new' queue is thoroughly full of junk * hardly anybody looks at the 'new' queue so that good things don't get promoted out of it * it seems very likely to me that almost all successful posts involve a voting ring of some sort * all too often, three or more stories about the same topic dominate the home page * certain people are running "made for HN" blogs where they quickly pound out a quick (but banal) response to a controversial article and then get voted up.

If you really wanted to improve matters, you'd want some A.I. filtering against infographic and 'me too' articles, protection against voting rings, and some mechanism to force people to moderate the 'new' queue in order to accomplish what they want to accomplish.

The 500+ karma stuff is a joke; that keeps new people out, except for the ones who don't have a life. That kind of thinking is a big reason I don't participate on StackExchange.

A "joke"? Not being able to upvote or to "Ask HN" would keep you off the site entirely? You know there are already karma-threshold features on HN, right?
Karma is an absolutely terrible metric to use - jaybol has 3,316 karma and his job is to make infographic spam and submit stuff to social sites, quite obviously including HN.
I read this, thought, "whoah, really? That's damning!" and took at a look at 'jaybol (who I'd never heard of). This is an unfair summary of that user. His top submissions on SearchYC aren't infographics. Infographic spam may be his job, but he submits diverse and interesting articles to the site and that's where his karma comes from.
He's paid to make those submissions whether it's an infographic his company produced or a blog post they didn't. If that's not spam then what is it?

From one angle HN is just a traffic source for user submitted content and that makes it a target for people/sites/companies who aren't invested in the community that is the value.

Again: I am not arguing with you about whether he's a spammer or not. I have no idea if he is. I'm saying, when I look at his top submissions, they aren't low-quality.
You would probably enjoy submissions that are approved by voters who voted similar to you in the past. You would probably want to penalize submissions that are approved by voters who you voted inversely in the past. So, the idea is to weight every vote according to what your past voting pattern. For initial implementation we can use PG's voting pattern: http://news.ycombinator.com/item?id=2242525
Can we have a 'None of the above' and an 'Other' option?

Great poll, though we may be missing some information here.

The guidelines should be simple, clear, and should just pop out right next to the box whenever you're writing a comment.

Hiding score is also a great idea since it reduces upvote-downvote wars.

Rather than adding new functionality I'd prefer some time be spent on making the site readable on mobile devices. Trying to read HN comments is ridiculous on the iPad let alone a smartphone screen.
I would just like to say that Hacker news reads very well on the Kindle.

It is quick to load which is key and I use it as almost a newspaper each morning going to college.

I think stricter guidelines regarding what's off topic, and stricter enforcement of them would be much appreciated.

I have my own sense of what's off topic, but lots of people seem to like to play the "7 degrees of hacker news game", exemplified by this post:

http://news.ycombinator.com/item?id=2241455

In a few hops, you can make anything hacker news!

My own opinion is that politics and other topics that give people that "riled up" feeling ought not to ever be here, whereas the occasional random story about something really unique and interesting might pass muster. But ultimately my opinion counts for little: is PG going to leave the site to 'mob rule', or come down on those kinds of stories? I guess a decision either way would be good in that we'd avoid some of the meta-discussion.

On Twitter, tptacek specifically suggested a hard ban on politics as an example of more restrictive guideslines. What makes me uneasy about this is that there's a huge gray area surrounding what qualifies as politics. For example, I don't regard tptacek's recent comments about the market for talent as politics, but a lot of people probably would. If we ban politics, then petty squabling over politics will be replaced by petty squabbling over censorship each time the mods kill a thread which is only borderline political.
I can't decide on my opinion of the politics (especially economic policy) submissions. I appreciate some of them, but I can't help but feel most of them are the same regurgitated crap that just happens to get a lot of votes. Presumably these (votes and submissions) are from people who agree with the article, not necessarily because they think it's a valuable new contribution. The best I can think of is to tag them as such and weight them differently. How? I don't know. We don't really have enough information about who votes for what to make guesses like that. (And I'm fully aware that weighting some votes higher than others could easily be seen as elitism)
I like the current guidelines which I'll summarize as "interesting things of interest to hackers."

I agree that "7 degrees of reddit" or whatever is bending the rules too much.

Something like a hard ban on politics might lose things that are of interest. For example, what role are technical tools playing in the arabic revolutions right now? Obviously Wikileaks, reddit, youtube, twitter and facebook are playing major roles. And what role are hackers playing? Assange, Anonymous, and the Egyptian Google guy are all using technical skills to create worldwide revolution. That's pretty exciting, using technology to make political change. But should discussing it be banned? Is it necessary to ban that to get rid of the various latest cop shooting incidents reposted from reddit where technology is at most a minor role.

Not only do I think a lot of the "how technology is changing the Middle East" comments shouldn't be on the front page, but so does Paul Graham; he manually buried a bunch of them last week, despite hugely high ratings.
Now THAT'S politics.
That's a semantic game. We all know what the word "politics" in the site guidelines means.
So the articles are interesting and have huge ratings so they are of interest to hackers and they are about technology, but one person with power decides he doesn't want to see the topic discussed, so he squashes all discussion with the thumb of a dictator who must not be questioned.

How is this not politics.

Just so everyone's clear, the mighty-thumbed dictator we're talking about here is Paul Graham.
How do you know that he manually buried them?
(a) Because he did it before with TSA stories.

(b) Because one minute there was a 200+ story at the top of the page less than a day old, and the next minute there was no such story at the top of the new page, and yet the story hadn't been flagged.

According to this comment, http://news.ycombinator.com/item?id=1936734, any admin can "kill [an article], mark it as political or whatever, or do nothing". That implies that pg may not have buried it, and that he didn't make some special exception for that one topic.
All the Egypt stories vanished from the front page at the same time, but I'm not an admin and concede that I could be incorrect about this. He said specifically he was burying TSA stories a few months ago.
Do you remember which stories?

BTW, tagging a story as political doesn't cause it to disappear; it just ranks lower, but would probably still be on the frontpage if it had a lot of points.

The "Mubarak resigns" one is the one I remember.
(comment deleted)
Is tagging an explicit feature of the site? Since I have not seen it, I assume there's a karma threshold for it?
None of us can "tag" things except Graham and his minions.
I like the poll option of making the flag button more powerful, but it wouldn't do enough on its own. In addition to that, I'd like to dock karma when a submission gets flagged: -1 karma for each user who upvoted, -10 karma for the submitter.

There are no repercussions for upvoting or submitting questionable content at the moment. While my proposal is only a small penalty, hopefully it gets people to at least think before actions that potentially increase noise.

Edit: I meant "when a submission gets successfully* flagged", ie killed. Thanks davidw for pointing out the ambiguity.

> There are no repercussions for upvoting or submitting questionable content at the moment.

This is a huge factor, IMHO - a much higher signal/noise ratio in the articles on the front page would probably set up positive feedback, helping with other issues. Unfortunately, there are positive incentives for being the first to submit anything that could potentially be voted up, drowning out interesting content.

I like the flag penalty a lot, and would only suggest that the penalty be sharp. 10? How about 100?
Of course... you'd want to look and see how that could be abused as well. A pisses off B, so B looks for A's recent submissions and flags a few.

One way around that would be that if a story is killed, then and only then does the penalty occur, and it's sharp.

Yeah, I'm saying, when a story gets flagged off. It's a good point.
Ok. The problem then becomes: the 'mob' seems to like lots of these stories, thus making them difficult to flag into oblivion. Once something has hit the front page and is rising rapidly, only PG and the moderators can well and truly nuke it.

I think it really comes down to a human solution to a human problem: tech only goes so far to fix problems like these.

Docking karma from everybody who upvoted a flagged->killed submission is harsh, but brilliant. I really wonder how that would play out!
Some scheme that at least partially counterbalances the large comment karma advantage garnered by early commenters on a story.

As it works now, this advantage is an incentive for users to jump in early with low-quality comments that just get votes by pointing out the obvious.

An example of a partial fix for this: show upvotes for comments as normal, but when calculating a comment's contribution to the user's karma, give lower weight to early comments, to reflect the advantage they had in getting upvotes.

I vote for making the flag button more powerful. Most often, the worst discussions are about flamebait from TC about something Apple or Google has done(I am as guilty of this as anyone). If, with enough flags, those stories could be moved off, then we could go back to having saner, less shrill discussions, and improve the overall quality of HN as well.
I've been here close to 1000 days now. I've seen a lot of online communities get popular and lose what made them special, but so far, HN has been remarkably resilient in the face of serious growth.

HN has a lot going for it. The sparse look and scary name keep out most of the undesirables. It's hard to turn a place called "Hacker News" into yet another funny cat pictures board.

Looking at the front page right now, I'd say things are still very much on topic: Collision detection, hash tables, Amazon S3, C++ style guide, etc, etc... Sure there's a couple lame Techcrunch articles, and maybe a bit of current events, but if you go back and look, those things have always made the front page in limited numbers.

If there's one thing I'd do, it's make the "new" page time-based, instead of strictly limited to 30 items. Everything should get an hour to percolate before being pushed to page 2.

Collapsible comment threads and complete hiding of comments below -2. Doesn't directly cure the root problem, but why not take the edge off the symptoms in the meantime?
Over the last 6 months I've seen comments that deservedly made it over the +10 mark hit -2 before getting there. I don't think you can just disappear comments that cross -2; it takes time for the score to settle.
Tagging. Some people are interested in news about famous startups, others are interested in technical articles.
flag-like button for duplicate/repost.
Submission quality is secondary to the drastic decline in discussion quality.

I've always believed that best way to improve the quality of contributions is to make the opportunity to contribute seem finite. It'd be interesting to see what happens if you cap submissions at 2 per user per day and max comment thread participation out at 5 or 10 per day. With a limit in place, people may do a lot more thinking before speaking or submitting.

I can't argue about how effective this feature might be. It seems like a very, very good idea. But GOD do I hate it. :)
Maybe bonus submissions/comments could be awarded for some sort of karma threshold on individual items (e.g. One more submission if you hit 50 karma on an item or one more comment if you hit 15 on a comment.)
Maybe give extra or different points to those upvoting and thereby curating the newly posted links that haven't reached the frontpage yet.
Count every user's up-votes and down-votes.

Change voter weight depending on how much it correlated with PG's votes in the past. Vote can even change direction if inverse correlation with PG's votes was observed.

Next step would be to create personalized page for every user when new submission is sorted according to correlation with that user (instead of PG).

Just kill political stories more quickly.