Hacker News new | past | comments | ask | show | jobs | submit login
What I've Learned from Hacker News (paulgraham.com)
381 points by mqt on Feb 25, 2009 | hide | past | favorite | 119 comments

I think a lot of the growth problems might have to do with the devaluation of karma. The solution to this might actually raise the quality of comments and front page stories.

When I posted a comment here a year and a half ago, I felt like each vote was a real, actual smart person praising or agreeing with what I had said. Each point had a lot of meaning, and back then 5 points was considered a pretty good score. Now each point means much less for some reason. I'm not sure why. Maybe it's because I don't trust each individual person as much. For example, as I write this, the top comment is about someone using a script to submit this article. It's two sentences, and it has received eleven votes. To me, each of those individual votes mean less than they used to.

The net effect of one vote, for the whole culture of hacker news, has gone down significantly. This reduces the total perception of worth for each user, and, more importantly, for each voting action. Back in the beginning, I felt like I had some amount of power, in that if I voted something up it would literally rise in the list. In fact, I felt like I had some sort of responsibility. Now that active, obvious change in story rank rarely happens, so I feel like my vote is somewhat worthless.

I tend to mostly agree that the quality level of submissions and comments has stayed the same during the growth of this site, but the devaluation of karma has led to other cultural changes. There's something comforting about sharing ideas within a small group of people because you feel like you have real influence, and I think I'm starting to lose that feeling here. There are so many new faces that my perception of quality has gone down, even if the actual net level of quality for each item has stayed the same. Seeing many posts with 100+ karma on the front page recently has kind of solidified that feeling for me. Who are these hundreds of people?

I agree, and I think that the solution is simple: limit people's ability to give out upvotes. Maybe restrict people's weekly upvotes to c * ln(karma) + d or something. (Make d small!) Or have upvotes actually cost you like a tenth of a karma point. (Dunno if karma should be turned into currency, though, since then I could imagine people trying to game the system.)

This solves another problem that really bothers me: an upvote to a really insightful comment (like the parent) has the same weight as an upvote to a comment that's merely useful, or, worse, that merely shares my feelings on some topic. Insight (in my opinion) should always count for more than anything else, and I think limiting upvotes will make it so -- because I think most people actually do care about it most.

- EDIT - Another (even more dire, IMO) problem that might have a similar solution: comment threads are way too long. Only four or five months ago I could make a comment and I'd be absolutely certain that at least a few reasonable people would read it, and so it never felt like luck when my comments moved up; my good comments inevitably got upvotes, and my mediocre comments stayed where they were. Now comments get lost in the wash. I'm tacking this edit onto this comment partially because it's tangentially relevant, but mostly because I don't have faith that anybody will ever see it otherwise.

That's a huge problem, and I suspect that it is in fact the fundamental cause of the degeneration of other social news sites. When people know that their contributions will be noted, they act like they're participating in a conversation. When the success of a comment depends on a metaphorical die roll, I think people adopt the mentality of, "I'll throw shit at the wall and see what sticks."

My solution would be to limit weekly "root" comments (comments that respond directly to a submission, not to other comments) with a formula similar to the one above (c ln(karma) + d), though with the caveat that comments that get more than a few upvotes get "refunded." But frankly I'd endure any kind of unreasonable limitation if it would get rid of the feeling that this community is gradually but inexorably turning into a mob.

I think that's a good idea. When people can give out unlimited upvotes, it's like the government printing money non-stop. Karma loses its value.

I don't know if the ability to upvote should be related to a persons Karma score. That might make one group too powerful and reduce the egalitarian nature of the site. What if everyone was just limited to 10 upvotes per day? Do you really need more than that? The more you read HN, the more carefully you will have choose what to upvote.

I think, rather than a quota, the right to vote should be earned, for example by a history of quality submissions (determined by either of the algorithms described by PG). This is already partly so with the (more simplistic) karma threshold.

In the same vein, addressing the problem of too many comments, the extent of right to comment could somehow be earned. Say an unproven user can only comment in particular circumstances, maybe on threads that aren't already heavily commented on?

Let new users prove themselves in the lesser leagues, so to speak, before allowing them free reign on the site proper.

I don't think the problem is with upvote inflation as much as dilution of individual-to-individual recognition. Limiting upvotes wouldn't address that.

Sounds like a textbook example of Dunbar's number.


Maybe a good solution is to break the community down into neighborhoods of ~150 people.

I've proposed this before: http://news.ycombinator.com/item?id=391292

The big question would be how to propagate submissions and comments between communities. Especially if each community develops its own unique character. I've thought about organizing each community onto branches of a B-tree and letting upvoted submissions move up the tree, though my intuition says that this might not scale, as one community might overpower the rest.

Anyway it would be interesting if someone builds it.

Nice. A more scientifically plausible version of the "monkeysphere."


It felt weird that a satirical website would have such a profound article; it's good to know that more serious consideration into the underlying concept has been made. (The above wiki article once made heavy reference to the term monkeysphere, but had it edited out over time. There's still quite a bit of text about it on the discussion page.)


With a smaller population of users, each interaction becomes an investment in an individual relationship that may pay off later (financially or otherwise). And if someone upvotes your comment, you know they are endorsing you, personally.

With a larger population, this dilutes to something like a branding exercise.

So there's some balance between having lots of diverse perspectives and skill sets (large population) vs. having individualized relationships (small population).

Similar to the idea that companies may have an optimal size, maybe there's an optimal size for social news sites. And for social news sites where utility is part of the draw (HN), this optimal size may be smaller than sites oriented more around entertainment (digg, reddit).

"The most dangerous thing for the frontpage is stuff that's too easy to upvote. If someone proves a new theorem, it takes some work by the reader to decide whether or not to upvote it. An amusing cartoon takes less. A rant with a rallying cry as the title takes zero, because people vote it up without even reading it."

When I read that, I was immedately struck with the idea that story votes have a "velocity in voting" and that a story with a "high voting velocity" is probably fluff, while one that has a "slow voting velocity" may be significant and perhaps some measure of this could be taken into account, kind of like the personal shields in _Dune_.

On the other hand, certain things that have an extremely high voting velocity might be necessary. Emergency posts from pg, news of some incredibly major political event on the order of 9/11 making it on the front page. Perhaps high velocity items should be referred to moderators?

After thinking about the topic for the past year or so I sat down and wrote up the major problem I found with using up/down voting for comments in a rather long blog entry


I would be extremely surprised if HN changed or even tried a different system, but hopefully it can be useful for developers who are creating the next generation social new site. The key point: The number of points a good comment gets is directly related to the number of users that read the comment.

What about borrowing the popular 'like' feature from Friendfeed? After a comment, it lists people that liked something, but it cuts it off at a certain point so it's not overwhelming. Perhaps one could sort the prominence of likes by how recent the like was (so that everyone gets a chance to show they liked something) and maybe the karma of the user, so prominent, active users can express enjoyment of a comment or post without having to add something specific to the conversation.

This would restore something of the personal feeling, too. As pg says, it's better for moral to know that one particular person deeply liked something than many, superficially.

Wow. Maybe karma needs to be more meaningful...perhaps a reflection of who voted for you and what percentage of page views voted for you. It's not that I want to see social stratification edified on HN; but if your comment is voted for by users who have earned more karma through their own comments, shouldn't that predict that your contribution is more meaningful to the community? It might be easier to just say 'a sort of collaborative filter for karma weighting.' Even something as simple as a Pearson correlation might be fun to play with.

I can't agree more about the power loss of the vote. I lost the right to upvote and for a while was disturbed by it. But then I thought "Who cares, it meaningless anymore. Not even worth petitioning for. As long as I get to hear about new ideas and whats going on with interesting tech and science, that's all that matters."

When you have more people voting, naturally each vote is worth less than when fewer people are around to vote.

I find technical news sites to be similar to how fashions work.

That is, a good model of changing fashion are some forerunners that starts doing something -- and then gets copied by lots of people. The ones that want to be fashionable then have to change or look like all others.

Technical news sites seems to be places with interesting discussion that gets overrun with wannabees that lowers the average quality. If this is correct, HN is doomed to be Diggified (the opposite of dignified). Like all non-invitation discussion sites.

(-: I also tried to be responsible for the average quality and only read, but now failed to keep my fingers controlled after a few months. :-)

What a disaster that would be, to attract thousands of smart people to a site that caused them to waste lots of time. I wish I could be 100% sure that's not a description of HN.

The reason HN is so addictive is because it is filled with things we think might be useful someday. We're like pack rats at a flea market.

The question is, what utility do we get out of what we find here?

Idea: maybe there could be a second class of upvote, which could only be given, say, a week, or a month after a story or comment was submitted. The meaning of this upvote would be, "This proved to be useful to me professionally." And so, in addition to HN/best and HN/bestcomments, there could be HN/useful and HN/usefulcomments.

Anway, thanks again for the site, Paul, and please keep experimenting. Watching you experiment is at least as interesting as any particular post.

Despite all the time I have wasted here, I can tell you (with 100% confidence) that it was worth it.

My co-founder is nearly 3000 miles away. Hacker News single-handedly keeps my head in the startup game. Bootstrapping would be impossible without this site to act as a virtual startup town square. The stories and comments I read here motivate me better than anything ever has. I used to start projects and forget about them on a nearly daily basis. Hacker News has focused my efforts and sharpened my mind. Sure, you can fritter away hours here, but hey, you need to break a few eggs to make an omelet.

A big thank you to PG and to each and every one of you smart people!

"Despite all the time I have wasted here....it was worth it"

This sentiment is very common here, but I strongly question it. I also wonder why people don't label their time here a little more honestly: as procrastination. We all justify it as "oh, but I might learn something useful for my future". And some of us might have indeed tangibly improved their lives (though I'm sceptical). But surely whatever benefits you gain can be reaped by visiting say once every 3-4 days, vs 12 times in one day? More generally, out of the people who do claim to have benefited their lives, are they able to point to a specific instance or two where HN tangibly improved their lot?

I'll be honest: I resent visiting here often. I'm a worse person for it. Every minute I (or you, for that matter) spend here, someone else is spending that minute doing great work (for some definition of 'great'). In fact I am suspicious of anyone who has a ginormous amount of comments posted frequently; I do not see it as something to be proud of, but quite the opposite.

I'm sure there are people who buck the trend (pg seems to frequently post here, yet still manages to do great work. I am convinced that he has cloned himself without telling anyone).

I realise that what I'm saying might be heresy (see 'What You Can't Say' etc) given the constant optimism among this site's visitors, but I do not mean to denigrate anyone when I say that most of us (me included of course) are timewasters.

I am less concerned by the implication of reduced productivity. Though reading Hacker News may negatively impact present productivity, it may improve future productivity by exposing readers to new concepts, approaches, etc.

It reminds me of an article about Pixar (though a cursory search doesn't seem to find it...) which described their workspaces as having only a few central bathrooms in order to facilitate people running into each other, discussing ideas, and being exposed to different people's work. I think Hacker News provides a similar service without having to get up and walk at all.

Hacker News, unlike bathrooms, is instantly available and you can get to it in a few clicks. This is easily the site I spend my most time, and I'm not entirely certain that's a good thing.

An important difference is that people actually need to use the bathroom anyway, so such chance encounters are not creating an interruption where one did not exist before. They are also self-limiting in a way that browsing the web (or RSS feeds, etc.) is not.

I have definitely found utility in HN, in direct feedback/advice that I've solicited, in examples from what other people are doing, in learning about sites that I could integrate with, and from principles that helped me prioritize.

If you look at the feedback I initially got here for OurDoings, and then look at what OurDoings is like today, there's no question I've learned a lot.

Oh, and from JavaScript code I copied from the HN. The Image() voting code looks like a hack, but if Paul Buchheit did it it must be OK. :-)

I live in the middle of the Tirolean Alps and don't speak the local language. HN has been my outlet for 'geek talk'. Hopefully when we're back in Italy (next week! yay!) I'll have a bit more contact with various friends and contacts.

I think it's important that a site that kills submissions provide a way for users to see what got killed if they want to. That keeps editors honest, and just as importantly, makes users confident they'd know if the editors stopped being honest.

I think on HN, editor transparency is less important than story Quality.

Reddit and Digg were built explicitly around some democratic idea of deciding what stories should be displayed on the front page. While both have their problems living up to this, it is their stated objective.

Hacker News has a different objective. It's goal is to be interesting to Hackers. I really could care less about democracy and I think others would agree. Submissions and voting are just a means to the "hacker interesting" end.

Absolutely agreed - and this goes for the voting system as well. I don't necessarily care about up-modding a submission or comment; I do so to try and help out and keep discussions clean and at least somewhat intelligent. If there was some intelligent way to filter submissions or comments without user input, then I wouldn't mourn the loss of "democracy" in the system at all.

Oh, and from the post - snarky comments and memes are definitely becoming a problem on HN. They add nothing to a discussion, are not insightful in any way, and do not express anything except the author's ability to regurgitate old jokes. They need to stop.

A lot of startups have some kind of secret about the subterfuges they had to resort to in the early days, and I suspect Digg's is the extent to which the top stories were de facto chosen by human editors.

These also make for excellent reads. What was Viaweb's? :)

Ours was that initially we couldn't give it away. The first couple merchants would only agree to use the service if we gave it to them for free and did all the work of building their stores ourselves.

There was also the fact that many of our most successful merchants were selling something dubious. But I wouldn't call it a secret because we didn't know it ourselves. We had no idea why the stores selling Japanese comics or "dietary supplements" did so well.

Web 2.0 runs on dubiousness.

Completely minor factual amendment:

It was Ed Koch (and his police commissioner) who initiated the broken windows based policing. Giuliani took it to the next level and as you correctly phrased it, is the one with whom it became most associated.

Source: http://books.google.com/books?id=vHKBXA8nXvQC&pg=PA63...

Glad to hear that the orange names won't be back. I think people do more than enough judging based on the poster name as it stands without any highlighting.

Me too, but I'm glad it happened. It was a neat experiment.

Indeed - the true test is not whether you try something new, but whether you have the balls to admit it was a mistake and pull it back.

I wouldn't bet on that, since the usernames are gray you never really recognize them. i.e. based on your username, I never heard of you...but looking at your profile, I remember reading that before.

Don't worry vaksel, I recognize your username.

I would be interested to know how much time PG and others spend moderating the site -- not improving the code but killing stories and comments, banning users, etc. HN always feels to me like a self-policing community, but this and other remarks from PG point to non-trivial amounts of intervention by higher authorities, which makes me wonder how much work it takes to keep things on track.

I would guess about 1-3 hours a day between all the editors combined. But it's not too painful, because the editor UI is designed to make it easy to kill crap as you come across it while reading the site.

Does the submitter have a script that automatically posts new essays from paulgraham.com? You've submitted the last three in a row almost instantly. Not that I'm criticising; just curious :-).

Years ago, I figured that I could get lots of karma by creating an auto-submit script to submit stories from techcrunch and paulgraham.com. Best of all, if another auto-submit script submitted the same story a bit later, they would actually be voting up your story, so the more auto-submitters existed, the better. Sure, more competition from other scripts and users sounds like it's bad. However, I calculated that the scripter still wins because when they're the first to submit, there are even more votes than before (due to other scripters submitting a story a few seconds later and therefore voting up the original). It's better to get beaten by other scripts most of the time and make it to the front page with a bunch of upvotes in just a few minutes, than to always be the first and only submitter/upvote, most users never seeing the story.

And when an article has 10 upvotes in the first few minutes, it's almost guaranteed to make the top spot on the front page, and therefore gain even more votes from real users due to the publicity and feel that others have found the article interesting.

I was top 15 and wanted to get invited to Startup School again, so I didn't make the script. But since then, users have come out of nowhere to reach the top ten, accumulating karma at a dizzying pace.

One of the top users has said he submits the best stories from his rss reader every day. I believe a more likely explanation is that all stories from sites that are commonly referenced on the front page are auto-submitted by a script that checks for new articles every few minutes.

This was also my explanation on why so many techcrunch stories have been upvoted on news.yc--they make an easy candidate for auto-submitting: not because of story content, unfortunately, but because auto-submitting stories from sites that already appear often and that other auto-submitters might submit gives one a free set of 4-10 votes each time.

I guess my point is that Hacker News is being gamed the way reddit and digg are.

I know most of the highest karma users and I'm pretty sure they're not using scripts. I also have lots of code watching voting patterns (e.g. for voting rings) and though it's common for a user to submit many stories from one site, the evidence suggests nearly all of them are doing it to drive traffic, not game karma.

though it's common for a user to submit many stories from one site...

Hmm, looks like 15 of my 21 submissions have been from my blog.

... the evidence suggests nearly all of them are doing it to drive traffic...

Ouch. Any chance that I can convince you that I honestly think that the articles I submit are of interest to hackers? After all, I don't submit all of my blog posts here... just almost all.

Don't worry, there's nothing wrong with submitting your own stuff. It's only a problem if it's offtopic.

I submit my own posts from time to time, and I've only had one killed (to my knowledge). And that one was easily explained: somebody else had beaten me to submitting my own post with a slightly different URL!

> most of the highest karma users Ones who post articles from PG.

I'd like to offer my perspective as a relatively new HN user.

First, I stumbled upon Hacker News by searching for information related to angel and vc investing. I found Y Combinator naturally. As I was reading the application (just to see what the process entailed), I read one line about how good applicants come from HN. My impression at the time was that applications were screened for high karma points--even though I had no idea how a karma point was earned.

So to be perfectly honest, for the first few weeks I was obsessed with Hacker News. I spent A LOT of time on this site, before I came to the conclusion that Mr. Graham noted. It took a while for me to realize the TRUE value of this site: a civilized discussion forum where I could meet other people just like me and learn from their experiences.

But I also noticed competition in the comments section that has left me a bit turned off. Sometimes I feel like people are trying to prove something, by unnecessarily down-voting a good comment or arguing just for the sake of arguing.

I think the sum of my newbie experience has been good. Nothing is perfect. I have certainly learned a TON about the start-up scene, hacking, and just this forum in general.

I wouldn't read too much into how other people behave on here.

I'm argumentative, rambling and usually off-topic, and most likely to satisfy my requirement for cognitive dissonance, I've carefully ignored my karma so that my personality on HN isn't beholden to my karma score. Obvious results, I might add!

I guess my point is that Hacker News is being gamed the way reddit and digg are.

What a sad thought. "Gaming" hacker news to get invited to Startup School.

Here's an idea:

Why not just be yourself. Contribute what you can. Take what you need. Make a few friends along the way. And learn a thing or two that may change your life. I can't think of a better place to do it.

Karma is a very minor byproduct. What you and this community become is the real payoff. But like a planted seed, you may just have to wait a while before you see that result.

I think I read that sentence as being the opposite as you did, i.e. He didn't game HN because he wanted to get invited again, even though he almost did.

Correct. While I was working actively on my software, my brain focused on the best ways to do X or Y in the fastest time using a software application. In other words, writing something to make me more productive or more time efficient, even if it wasn't something that would be sellable. If it saved me time, then I could therefore use that time to code something I could sell. So I let my brain wander around the karma concept and that's what I came up with (a long time ago). But no, I decided not to do it.

One corollary to that is by not having done it, all I can do is write a post on news.yc. But that does not mean that it should be confused with the act of actually doing it.

Another corollary is that maybe doing it wouldn't have been a bad idea. If karma is the measure of value, and users enjoy the stories, maybe developing an algorithm that automatically submits stories to news.yc (and judges how users vote on it to then submit better stories, the way a human might) would actually be something Paul loves, and the first release of such an app, of course, could basically submit everything indiscriminately to start with.

The trouble is a lot of good stuff slips through the new page unnoticed unless you post at the right time or get a friend to bump it. I find just as many tasty links from new as I do the homepage. Perhaps the homepage could shuffle in a few random new unvoted posts too?

Isn't it pretty neat, though, that Paul Graham doesn't have to write an rss scraper to get content, he can just incentivize users with karma and the crowd will do it for him?

Wait - if you do better at Hacker News you get invited to Startup School?

I know nothing about Startup School, but I didn't think it had anything to do whatsoever with things on Hacker News.

As I remember, it's not your karma score that really matters, but being recognized from here helps when they read applications.

The same effect is obviously true everywhere. Being known by someone on the inside gets you a job more often than not, etc.

If I recall correctly, when I went in 2005, the application requested my reddit username. I had hardly any karma, but a few people, including Steve and Alexis knew me from my comments.

"I guess my point is that Hacker News is being gamed the way reddit and digg are."

Only if you think the "game" is to accumulate karma. For me, the submissions are merely seeds around which insightful discussions grow.

There is a tangible reward for people who have the most karma though.

Sort of. Consistently making smart comments carries a lot of weight in getting an invitation to YC interviews. It was the deciding factor for at least 3 people we funded: danielha, alexwg, and one other I probably can't disclose because they're not launched yet.

It's not simply a function of karma, though, but whether we recognize the username as someone who seems smart.

No there isn't.

I remember reading that top karma earners get invited to startup school...I can't find that anywhere.

But here is something close: http://news.ycombinator.com/item?id=136837

Karma is taken into account when you apply. That is incentive enough to attempt to game HN.

There is? What is it?

I'm sure someone's created a bot by now that posts all the top stories in reoccurring patterns amongst all the social news sites. With that plus your idea I'd call it the bot of infinite karma.

I've been toying with the idea of adapting the Elo rating system from chess to a social site's user score. The basic idea from chess is that you assume someone has an intrinsic skill in some arbitrary units (I think you start out around 1,000 and a 2400+ is senior master) and that we can only infer that skill from how they perform in tournaments. Furthermore, your skill in any particular game is assumed to be normally distributed about that intrinsic skill value. It is also assumed that your skill will only change slowly over time (we don't have to pin down a moving target).

So the mechanics of this system are basically that when you compete with someone that is much better than you and win your score should be adjusted upwards. Conversely, if you lose to someone worse than you, your score will be adjusted downward. Adjustments are done in batches, say after a player has finished a tournament (I use five voting events between readjustment).

My application to a user karma system would be analogous in that a "match" or "competition" would happen every time someone directly rates someone else (stories may be exempt). A high-karma user upvoting a low-karma user would increase the low-karma user's karma and decrease the high-karma user's karma. If there are two users with equal karma and one up or downvotes the other. the number of points exchanged will be minimal (the gain/loss scales with the difference in karma).

The idea is that your ideas are competing with the opinions of the high-ranked users. The assumption is that new users will have bad ideas and that high-karma users will downvote them. A downvote from a high-karma user is expected and doesn't cost the new user very much (they are expected to "lose" to them often).

I've sketched out some of the functions here: http://paste.lisp.org/display/76193

EDIT: grammar fix, typed this before coffee

That doesn't seem like a very nice assumption, that new users will have bad ideas, and if old users lose karma every time they vote up a new user, how can you ever know if the new user genuinely has nothing to offer, or if old users are just wary of decreasing their score, and therefore, influence? (i.e., I have lots of karma, and I know if I upvote this comment that I like, it will increase their score quite a bit - I have influence now - but it will also lower my karma, and I'll have less influence over the next comment I see which I may like more...)

This is what happens in chess: highly-ranked players just simply won't play new or lower-ranked players at all. It's not a bad thing there, but then again, it is a competition. A social site isn't, though.

Oh I agree completely that it isn't a nice assumption; it is just a simplifying one (like the proverbial spherical cow) that's mostly wrong. But I don't think it is entirely wrong, and the ways that it is right may be useful, as an experiment.

Maybe I shouldn't have used the word bad. But new users may not know the established traditions of a group, maybe they haven't even bothered to read the FAQ. The articles that got me thinking along these lines were: http://www.shirky.com/writings/group_enemy.html and http://www.cc.gatech.edu/classes/AY2001/cs6470_fall/LTAND.ht... . The gist is that an important group function is to defend itself against (new) users.

But I think that what it'll do is to make votes matter because they cost something. Mostly, I'm interested to see what the implications would be.

Okay, but why should votes cost more for a user with more karma? I would think the opposite should be true, that if karma is an indicator of a user's value to the site, someone with more karma should pay less to vote up a newer user.

The way that I'm thinking about it is this: that it is relatively unusual, from the standpoint of the community, for a high-karma user to upvote a low-karma user.

Karma is a measure of the norms of the community, parceled out to users.

What a disaster that would be, to attract thousands of smart people to a site that caused them to waste lots of time.

Wow, it's like university. :-)

Ok, now that I've made that moronic joke, being HN I know I'll need to add some more substance to counter balance the negative effect. This, in essence is one of the reasons I love coming here - For users that care it's self policing, in other words, what Paul says regarding the broken windows theory is absolution correct.

The whole broken windows theory (and it critiques) are fascinating reads. I'd urge anyone who only knows of the summary version to dig deeper into the ideas. Certainly up most hacker's alleys.

For users that care it's self policing

FWIW, there are a lot of users that don't care. I know some of my highest-rated comments here have been one-liner jokes. This means that the community doesn't really consciously think about "policing", they just click things that they agree with or that made them giggle.

(As an aside, have you ever read a really well-written post that you completely disagreed with, and didn't upmod it? I have. It's hard to promote people that are wrong. Human nature trumps "self policing", it seems.)

I've enjoyed HN as an anonymous reader and learned a variety of lessons too.

This essay finally compelled me to register an acount, upvote some links and to make my first comment.

Purpose defeated? May be I'm starting to ruin the news site I like more.

Thanks Paul.

I know many of us have probably wondered what you've been thinking about your experiment. I know I have. As usual, you've managed to write the essay before people are even asking the question.

The addictive/interesting balance is a fascinating subject. I'll be looking forward to hearing from you about that in the future.

How about requiring submitters to tag stories that are submitted?

If you have a set of blessed tags, then the submitter must carefully think which tags apply to their submission. This also forces the submitter to consider if the story fits with the content guidelines of the site.

Tags would also be great if you want to find information on a particular subject later.

Maybe that would be a job for the guys that run searchyc (www.searchyc.com)

HN has reached critical mass. No need to worry about a quick death anymore. Add friction to keep the randoms out and to help make those that stay feel like they are part of an exclusive club. Friction = cost. Some micro payment, say .99/yr, that adds enough friction to slow the adoption by non-core types. Those that value this exclusive "club," like myself, would gladly render unto Caesar.

Too much navel gazing is really bad for a site. This isn't to say the essay lacks insight, just that Hacker News is diminished by frequent debates.

PG, have you considered invite only schemes?

What about explicit 'people rank', where friends can rate eachother in a reputation system? That might really help with the quality of comments.

It would also help conversation is you could follow people you like. Twitter and Tumblr both do a really good job of 'pick your peers'. I often track these two links: http://news.ycombinator.com/threads?id=pg http://news.ycombinator.com/threads?id=ivankirigin

It would be nice if there was a thread stream of selected people. It would make me engage in conversation more often. Too many comments don't have an easy response, making them closer to declarations than discussions.

Also, have you considered adding an explicit cost to votes? You could weight those above a certain karma threshold higher if each vote were $0.10.

What effect has it had on you as far as going from just posting an essay to posting an essay and having it deconstructed? Has instant criticism made you more reticent in what you say, or even annoyed?

Everything I've posted online has been examined pretty thoroughly since about 2001, first on Slashdot, then Reddit, and now here. So I take it for granted now.

I think it's good actually. Now when I'm about to publish something, it's second nature to go through and look at each sentence and ask "Is this false? Is this the sentence that someone is going to tear apart and make me look stupid?"

That's why when someone is dissing me in vague terms in a comment thread, I'll occasionally ask them to quote a specific sentence or passage from the essay they believe is false. Because I've already checked.

  Dilution is a hard problem. But probably soluble;
Best. Pun. Evar.

Hacker News is one of the places I go to "spin plates." In some ways it's a productivity enhancement inasmuchas I keep my mind active when otherwise needing a quick break.

HN is also my primary tech portal. While I still visit slashdot, I've found HN to be faster on stories that appear there, and I've also found more relevant items here.

For instance, there was recently a thread here regarding list copying in Python. Though I've used Python for years, I'd never known the information that was in the linked article. It's now a part of my personal Python knowledge base.

I like HN. I read it regularly, whether at work, at home on the computers, or on the go through my iPhone. It's that good. Thank you!

"HN users can do this by flipping a switch called showdead in their profile"

I wish you could still see the URL though. If an article has been deaded, there may be some comments about it, but you can't see the url, or click on it...

You can still see it by using edit?id=whatever

Ah thanks for that :)

One way to avoid meanness is to apply a principle known colloquially as "Miller's Law." http://en.wikipedia.org/wiki/George_Armitage_Miller

To understand what another person is saying, you must assume that it is true and try to imagine what it could be true _of_.

That is, when you see someone wrong on the internet, instead of trying to force them to agree with you, identify what they know that makes them believe their assertion.

I found absolutely spot on in your article to submit an article with different headers on whatever “community” you target (e.g. Reddit or HN or even Digg)

However it’s not only the “tag line” which need to be rewritten but also the content.

Which mean one version of the article for HN users, another for Reddit and yet another for Digg.

Which can translate into three levels:

- advanced (readers able to understand abstractions),

- intermediate (readers who want to understand “how to control”)

- and beginners (readers who just want to find a solution to their problem through a simple step by step process).

This force onself to write with a specific audience in mind. And I think it’s the key point of Paul Graham article: understanding your users.

Last but not least: having the comments section on where you found the link but not on the page where is displayed the article.

I found HN comments with Stackoverflow are so far the only sites where proper discussions can really happen with very inspiring thoughts.

However those Karma thingies, I absolutely don’t care. Also the triangle button (upvote), I use it only as a bookmark button.

Also what’s really rewarding is when you check the “New” page and see few hours after it hits the home page. You feel you’re not only following.

Have a good day and thanks for your work. Thibault

"stupidity more often takes the form of having few ideas than wrong ones."

best line of the essay.

This is probably sophomoric, but... I was wondering why there is no semantic engines that can identify the major themes in the comments and present them like a table of contents or color code them like checking a cached page on google? I'm imagining something like Yelp's engine that picks out the best dishes.


Hear, hear: if you have a solution for detecting stupid comments, let Paul know.

I liked the two ideas Paul mentioned: length of comment is related to stupidity, and not using the same words again. Just make those fall toward the bottom of the page and we'll all be happy.

While any algorithm can be gamed, length is still pretty good. If I have to go through the extra effort of adding two paragraphs each time I write something, well... it takes time to type. So I have an incentive to think and improve my comment.

Really, I like this. It's really the same problem as Google search: out of all the junk out there, how can you tell interesting content? Because we all want to read only amazing stuff and engage in discussions with brilliant people. In practice, it doesn't work that way yet, but what if technology made it possible?

I have no problem with comments as short as just one word, for instance: sometimes just the mention of a counter-example is far more effective than a longwinded riposte.

What I do have a serious problem with on HN is 1000+ word diatribes. They just don't fit the weighted threaded format, and tend to be rants anyway. If you can't fit it into a few normal paragraphs, it should really be broken up into several comments. If it can't be broken up, it's not suitable for HN.

Having length as an indicator of quality intuitively (i.e. I don't have statistically significant counter-examples) seems wrong to me. It goes directly against what hacking stands for (quick, dirty and yet elegant solutions).


Is anyone else finding this absolutely hilarious? PG was giving a "statistically significant counterexample", which is now being voted down because in its literal sense it is anti-HN spirit.

It does prove my point I suppose.

If it's a similar problem to Google, then you wonder whether the reputation of the commenters and the comment's responders could factor in.

I'd like a system that gave credit to people who start high-ranked threads (in the aggregate).

I would be interested in seeing the total aggregate comment points for a thread. I think some of the most interesting threads take off with unexpected tangents. For example, the recent one on marriage/dating.

A lot of really dumb, risible comments inspire thoughtful, valuable takedowns.

Maybe http://news.ycombinator.com/active is what you are looking for?

In general I enjoy Paul Graham's essays. However, they all have the footnote problem when reading them. The footnote problem is that I can click on the footnote's number to read the footnote, buf I can't click to go back.

I am only commenting here because I can't find a more direct access to Mr. Graham.

>> The stories on the frontpage now are still roughly the ones that would have been there when HN started.

Subjectively, the frontpage and discussions have grown quite bland here; without necessarily corresponding to a reduction in the quality of each article, it feels as if the range of content linked has become homogenized.

Objectively, however, it's quite clear that the range of content hasn't strayed much at all (witness archive.org). This failure to change lends a certain banality of tone to the site that is almost unbelievable.

Wow...Great! I truly enjoyed reading this most enlightening article. Many points you touched on, I say Amen to that.

Once you get in,addiction is non-reversible,the digging goes deeper and deeper ..into the web's bosom.It never stops. The NET is nursery rhyme's..."Come into my parlor" "Will you walk into my parlor?" said the spider to the fly; ... and went into his den, For well he knew the silly fly would soon come back again: ..."

Summed up in you words,"You may be wasting your time, but your not idle." Haha Cheers!

Part of the reason Hacker News has maintained its character is the people it attracts. The people it attracts, hackers, are drawn in by the subject matter of hacking, which hardly appeals to the lowest common denominator of the internet. I'd like to see the same approach taken with sport or politics. Can an internet community be created that discusses politics or even religion in an intelligent and respectful way? It might be possible, likely way harder than hacker news to pull off.

HN users can do this by flipping a switch called showdead in their profile

I have never had the urge to flip that switch. The sheer quantity of the quality links are enough to keep me engaged.

I'm pleased to hear that you're interested in the problem of attractive/addictive distractions. Important, interesting problem. I'm looking forward to reading your thoughts.

Here's a relevant thread from Overcoming Bias. It is more about modeling the problem than looking for fixes, but you might find it interesting:


Might be a minor point, but - pg, could you elaborate on the distinction between "bad people" and "people behaving badly"? I read most of the "Dilution" section as saying that bad behavior is situational (someone who flames on reddit posts intelligent comments here), except for the following blurb in the middle:

"forbidding bad behavior does tend to keep away bad people".

Is the implication that these "bad people" can't possibly behave any other way except badly?

Paul, I really hope you don't decide to shut down or limit access to HN. While I'm not at the programming skill level that HN is targeting, I'm learning a great deal from the articles I find here and I think I'm becoming a much better programmer as a result. HN is my first read every day and I appreciate being able to access it. Thank you very much.

My solution to the news-site addictiveness problem was to install LeechBlock:


and put up barriers to visiting reddit or hacker news after 11:00 in the morning.

Thanks for this experiment project Paul. I think Hacker News naturally attracts the right people which creates a very good vibe. Most if not all of the comments I read here are constructive and helpful at times. This is unlike any other sites out there.

I don't post any comment on HN anymore and I don't read comments often. It's time consuming. I visit HN every day and I pick 2 to 3 interesting links on average. That's it. If I need more time to waiste I go to reddit it's more fun.

pg's essays give a lot of insight into practical things in life. The corresponding comments are also good, but sometimes not good.I am a regular reader of these essays. So who posts the comments can be restricted, but sustaining HN will be good for the people like me, who learn a lot from it.But it is quite natural that one wants to just post some comments, just for the sake of being a part of the discussion.Once you keep reading good things for quite some time, your fingers itch to type in some comments. Yes, I agree that this is not good and one should observe some control.But please keep up HN, it is very good.

One thing I really like about HN is that you have chance to, once in a bad light (low karma), redeem yourself with future posts that are more carefully written.

Out of curiosity, what language is HN programmed in?

It’s programmed using ARC

When you download ARC you get a HN example inside.



"but also because I don't want to spend all my time dealing with scaling"

I'd be glad to have that problem in my new venture.

Thanks for sharing Paul. Very insightful comments about the challenges of running an online community!

This is not nested because it is only partly a reply to Jessealdridge.

I am new to hackernews, so I can't claim any idea of how to apply these ideas to that site. However, I find the notion of a "town center" a very intriguing one for the web, especially apropos of footnote 7 of the article. Most social sites act like huge suburbs, and many of their users are invariably suburbanites trying to escape the suburbs. In this way, the problem of a lack of place gets abstracted, and suburbanite communities emerge around topics that the real people can't discuss in their physical environments for lack of a place to do so. Their place to discuss things they would explore in the real world itself becomes amorphous and isolationist. But this is too broad; enter Dunbar's number.

I don't think Dunbar was the first to suggest that communities need a finite size, but his observation that that size is proportional to the volume of a brain region is interesting in that it provides an example of a willingness to assign a metric for community size to which human beings would be subject. This raises the question I'm trying to ask and the point of this long-winded comment.

How can we create fulfilling communities on-line that will enhance the essentially human aspects of such communities? Jessealdridge suggested breaking large communities up into smaller numbers (perhaps something like www.fluther.com allows). But if any metric like Dunbar's number is to be believed, this number should be a reflection of a deeper determinant of identity in a community, which itself requires an obvious representation in the allowable forms of interaction on the site. In the high-quality sites in the physical world, human communities are delineated by spatial relationships according to the organizing principles of architecture. But such places are rare in the physical world, and are mostly confined to very old cultures or small liberal arts colleges.

Perhaps one of the reasons that such frameworks of fulfilling human interaction are so rare is that they are very difficult to understand and very expensive to make. The web can do better at creating frameworks for human interaction because it uses information explicitly, potentially clarifying the underlying principles generating the framework, and because it is much less expensive for people to come together and create something on the web than it is for them to come together and make something architectural, in the physical world.

Unfortunately, this seems to mean (as you pointed out in your article) that sites much less often impart a sense of a change in place, or of a place at all. I wonder how the problems you've outlined (trolls, stupid comments, mean people) map onto the architectural representations of these problems, and how the solutions you've come up with could map onto architectural solutions to the analogues of those problems. In short, the article makes me wonder, however tangentially, how the web could benefit from being more than a surrogate, and perhaps more of a support structure, for physical human communities.

Could hackernews benefit from having some relationship to physical places?

happy with HN now. please don't change the formula. i say that knowing it will change...but pleading with you to try to leave it the way it is.

I was living in New York when Giuliani introduced the reforms that made the broken windows theory famous, and the transformation was miraculous.

I would like to remind you here of the analysis made in the book "Freakonomics" regarding the "miraculous" reduction in crime at that time. Namely, lots of individuals that would have been in the social situation leading to crime were not born. The suspected cause being an abortion case won some years before in court.

I love Freakonomics, it's a large part of the reason I'm studying economics but that chapter is a catastrophic pile of wrong. Two articles picking it to pieces

http://www.bos.frb.org/economic/wp/wp2005/wp0515.pdf http://islandia.law.yale.edu/donohue/Joyce%20%282004%29.pdf)

Interesting, had no idea their point had been put to such a scrutiny. Thanks for the very interesting links.

I was interested as well, I discovered this:


Down the rabbit-hole we go!

I have a little idea too. May be it will be useful for users of this site to view little icons (pictures) of country flags, based on geoip (like in most torrent clients), OS based on browser version - the major ones - osx, win, linux, bsd, iphone, symbian. And may be browser icon - safary, ff, chrome, etc.

If those ideas does not confront with privacy policy, they would be simple but useful tricks.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact