This is a cool website. I was wondering about some kind of ranking system like this.
> Speed percentile: 98.82%
While I have to acknowledge that some instinctive/animalistic part of me is excited at being in this percentile, it also prompts me to remember that karma is effectively worthless and I really need to get busy on class work and research...
It's really strange the impact something silly like karma has, because I tell myself I'm above accumulating imaginary points on the internet, but obviously, I don't seem to be.
In some ways, it's curious. My highest rated comments aren't what I would consider my best. Instead, they are ones that appeal to a wide audience with some kind of emotional (rather than technical) insight. My last comment, a quip about about how every programmer thinks they're the best, was worth 64 points. However, a detailed post I wrote on molecular dynamics simulation where I fact-checked and reviewed the literature was only worth 3 points. Go figure.
EDIT: After thinking about it some more, I've decided karma isn't entirely worthless in real life. It is a metric that lets you know how well other people like your written work. Over time, I've slowly come to realize what will get upvoted and what will not. This ability could have interesting repercussions in the way I write about my research, or the way that I craft something like grant or fellowship applications. Every person is different, but posting on HN reveals to me that, in the aggregate, there are certain things you can say that have a certain appeal to others, and the capacity to recognize this could in fact be useful in many life situations.
>My highest rated comments aren't what I would consider my best. Instead, they are ones that appeal to a wide audience with some kind of emotional (rather than technical) insight.
I definitely feel you here. This is a trend I see absolutely everywhere; hell, I was surprised to find out it even extends into video games that implement this kind of voting system.
This problem feels like it could be solvable by changing the "meaning" of the upvote. Right now, the upvote is a tool that allows you to make a comment more visible and to throw a bone to the author in the form of karma. People probably feel inclined to "tip" someone who makes a clever or witty comment with the karma from the upvote. On top of that, people making witty comments who get upvoted are now encouraged to make more witty and unproductive comments because they know that they will get karma for it.
The issue is that the upvote represents general approval of a comment, rather than a nod of the head that a comment is productive or insightful.
Just throwing around ideas - maybe the elimination of the user-karma-tracking system altogether is a reasonable solution?
This is why I was against hiding karma on posts. I suspect that it drives the points reward up for one liners, since there's no way to balance things downwards, ie. not upvote or downvote if you think a post has too many points.
And I think the upvote is pretty much set in stone now. It's essentially "I like this post" or "I don't like this post", there's not much nuance.
I wasn't around at that point, but I disagree with you. Let me explain why.
You shouldn't downvote a comment because it "has too many points." If you think a post deserves a downvote, do so. Whether 25 or five or zero people before you thought so shouldn't play any role in how you perceive the quality of that post.
But in aggregate, it means that a throwaway one liner becomes much more valuable (from a karma/algorithm perspective at least). It removes a large amount of control from the people commenting.
In a previous comment on karma systems I suggested a scheme where there was a voting 'diamond' up, down, left, right, and center. Where up was "good comment/submission" and down was "not-good comment/submission", left was "less like this", and right was "more like this" and center was perfect.
That allows you to establish both what you like the site to have in it, and how you much you liked this instance of it. Perfect would mean simply that "spot on" or "this."
The second part of the system would then adjust the presentation of your karma (to you and others) unscaled vector along the line to the 'perfect' site in the context of the viewer.
To use a simple example, lets say HN was over run with cat pictures, and I thought they were the best so I vote like this and "+1" (up and to the right), You the reader hate cat pictures so you always vote down/less like this.
So we've created virtual root nodes at +loves cat pictures and one at +hates cat pictures. Now I can plot your karma as negative with respect to mine if we assume that the center point between your root nexus and mine is 0. Now do this with enough topics and you get multiple consituencies all in the same discussion space where your view of their karma will inform you how likely it is you'll like to read what they wrote :-)
I had an idea to plot comment karma as a function of number of replies, and upvotes. So the axis would be 'discussion' and 'popularity'. A controversial opinion is probably unpopular but heavily discussed. Likewise, a pithy one-liner is popular but not discussed. Then you can see whether you tend to just parrot public sentiment and spout off-the-cuff remarks, or whether you create interesting conversations.
The problem with your proposal is that most people think they're objective most of the time. The less/more like this distinction relies on people knowing and admitting that sometimes the stuff they like is not better than the stuff they don't like.
slashdot does let you choose the meaning of your "upvote". While its comment system is far from perfect, it does have many good ideas. For some reason, I felt like karma whoring was less of an issue back when I was over there. Maybe because you could only be upvoted 5 times per comment, and you couldn't directly submit stories.
This is the one thing I have always missed about slashdot discussions. I rarely value the "humorous" posts on sites like this. It is not that I do not find the jokes funny, I do, its just that I'd rather increase the SNR even at the cost of a few chuckles. I think the big impediment to widespread adoption is the difficulty in creating an useful interface for tracking "kind of upvote" and then filtering based on type of upvote. I always had my slashdot prefs set up so that I never saw the "+5 funny comments."
I can't speak for PG but being in the middle of designing a karma system for our application, karma is an incredibly powerful tool to encourage engagement.
Karma's value to the casual visitor is in communicating quality content that's been flagged by the community worker bees - the karma "waggle dance" :) From that perspective you want to be critical and only add weight to the "best things".
However karma also plays a very significant role in guiding people as to how to modify their behaviour such that the community appreciates them. We all want to be liked by communities we identify with and karma is a powerful tool to see how we can "fit in".
Karma is an incredibly powerful tool to communicate in a very discrete way to new community members how they can establish themselves.
As the karma-moderator I find myself looking to give the community ways award karma wherever I can. I'm not worried about how insignificant the effort that it was rewarding; all I want to do is give people encouragement when they do something (anything) the community appreciates.
Karma is a tool to promote social cohesiveness in a group. By allowing peers to reward and punish others, it provides a "majority rules" code of conduct that is as fluid as the whim of the mob.
This is good for building a sense of community where in-person interaction is infeasible, but it also brings with it the bad qualities (group think, punishment of new ideas that contradict the norm, cliques, personality cults, etc).
The good news is that there are oddballs in every group who ignore the social rules (karma) and present ideas that are novel (but perhaps unpalatable) to the group despite the meting out of punishments (which are ignored). They are the antibodies that fight off the cancerous growth of tribalism.
The obvious answer here would be Slashdot's system- not just upvotes, but specific categories of upvote. It seems to work well, but it does add an overhead to the action of upvoting an item.
I believe that your "Edit" hits the entire purpose of karma - encourage better writing. Karma for good posts. Karma for good submissions.
What is sometimes a struggle is that better writing is not always what we wish it would be. Sometimes is well reasoned explanations. Sometimes it is passionate argument. And often, highly rated comments and articles become highly rated because of timing. Sometimes it is just showing up (80% of the time, Woody Allen may have said).
It's satisfying to make a pitch perfect comment - even more so I find when it isn't snark or mean. I know what my best comments have been, and I know these are not the same as my highest rated. My best comments have been ones that I've cared about and some of those have even gone negative.
Do I really deserve that 69 points of karma? Is that 20 or 30 or 69 times more valuable than other posts where I've spent a lot more effort?
Perhaps this is where some of the 'snark' is coming from. In the race to get more karma, people are looking for the cheap throwaway one liner that everyone will snicker at and upvote.
Karma is probably more closely related to how popular your opinion is on a given subject in the hacker news community. As you said it does not really encourage better writing, while there is sometimes a correlation between a good post and many upvotes, especially on very technical subjects.
Also, Karma on news submissions has become a race to be the first to publish the latest techcrunch news or related sites, so it's not all about comments or contents.
> My last comment, a quip about about how every programmer thinks they're the best, was worth 64 points. However, a detailed post I wrote on molecular dynamics simulation where I fact-checked and reviewed the literature was only worth 3 points. Go figure.
It doesn't seem surprising to me. It takes effort to recognize quality, most people aren't going to bother. It makes sense that quality is hard to measure, because if it weren't it would most likely be easier to produce. Perhaps the score should be corrected for length and how deep in the thread a comment is, but getting it right is probably more trouble than it's worth.
I think that it is hacker etiquette to only pass judgement when you are qualified to do so. This is why funny comments that don't require specialized knowledge are upvoted more than insightful comments.
The Slashdot comment system tries to correct this by awarding karma to "interesting" or "insightful" comments and not giving karma points to users with "funny" comments. Of course, the downside to this is that the poster of a comment that many find to be funny can end up with bad karma because one "flamebait" mod counts for more than six "funny" mods.
The problem with Slashdot is comments that are flat out wrong are often moderated as insightful. This probably happens because moderators feel obligated to upvote what they perceive to be "intelligent" comments even when they are not qualified to judge them as such.
I don't think it's etiquette, but rather disinterest. As soon as a comment is specialized, it has a smaller audience and hence draws from a smaller pool of potential upvotes.
The problem with Slashdot you cite is interesting. It is related to more general problems with crowdsourcing. Wikipedia is successful but sometimes you really notice that an article is written by different people with different viewpoints. Similarly, a "democratic" model for science, where instead of peer review papers are upvoted, does not seem very attractive.
Perhaps this all boils down to the notion that neither truth nor quality are democratic; i.e., the average opinion is not a good estimate of them.
On the last point, one thing I've taken away from the HN karma system is that your votes are most highly correlated with the general traffic around a discussion, more than with anything specific to the comment. Posting almost anything decent on a story that sits on the front page for a day will get you double-digit upvotes, especially if it's an early comment. But posting something in an article that only briefly hits the front page will not get much attention.
Come to think of it, that does transfer to both business and academia to some extent.
Are you suggesting that a knack for timing and an ability to deliver something that others value might be individual qualities which YC might wish to develop in potential applicants who participate on HN?
I don't see it as very HN-specific. It's the exact same pattern you find at Reddit and Slashdot, so if YC is mainly attempting to cultivate that property, looking for users with the highest Reddit karma might be a good strategy (and they've proven themselves to a larger audience, probably more representative of average consumers). The negative way of looking at it is that it's most important just to be in the general vicinity of activity and say something with either broad appeal, or provocation.
In academia this is becoming a common strategy with the rise of metrics. Find a hot area, publish a paper with a linkbaity title and prominent keyword placement, and wait for the citations to roll in. In business, it's the strategy of choosing your field of endeavor by picking the places where the most money is flowing around, and then just doing something average in that space.
It depends on the topic, lots of popular legal based threads have had a very in depth reply from grellas with a ton of upvotes. I think in that situation although not everyone would fully grasp the content they are upvoting in appreciation that someone who is an expert in the field has taken the time to explain and simplify the complexities of a legal case or argument.
This also depends on what your interests are. For example, my primary interests are in programming languages, functional programming, type theory and so on. My highest voted comments have mostly been fairly long ones in these topics.
Moreover, most of them were pretty constructive: things like explaining how to generalize pattern-matching, how typeclasses are useful, what FRP is or even just how your editor can display Haskell syntax with Unicode. (It's like an awesome version of syntax highlighting.)
Admittedly, my single highest-scoring post was a bit less constructive: a rant about Go. However, in my defense, it was a long rant with substantial details and examples. And, honestly, I'd be extremely happy if it saved some people from learning Go and got them to consider Haskell or OCaml instead.
Basically, be as mediocre and stereotypic as possible, conform to the crowd, please everyone, this is what you are talking about? Maybe, it's an inherent problem with the "karma" system. It encourages you to avoid saying what you really think unless you are a Mr. or Ms. Mediocrity. This gets even nastier considering that most people are narrow-minded and aggressive.
Is it easy to separate karma aquired from submissions and karma aquired from comments? My average comment karma is below 3 and I've only had a few comments over 20 but most of my karma has come from 2 or 3 submissions of interesting links, which seems almost like cheating.
I think the most karma I ever got was from being the first to post about Sublime Text 2. This was interesting but I only got the karma because I beat everyone else by a few seconds. I feel that's far less valuable than the odd thing I've spotted which no one else would have ever come across.
In the same way, I wonder how many people have high karma just because they are one of the first to comment on popular submissions.
Edit: typo (I wonder how proof reading my comment more would have affect my karma)
This was interesting initially, when I just "ego-surfed" it when the link was first submitted, because of the richness of the data displayed. It became more interesting, as the discussion here in this thread developed, when I realized I could look up any arbitrarily chosen user, for example the user who kindly submitted this link, freework, to see the same rich data for that user.
The current user interface on Hacker News proper shows a leader board
But this interface offers more, and prompts me to look up some fellow users to see what their detailed statistics look like. As I expected, cwan, who gains almost all of his karma from submissions rather than comments, has a higher percentile than anyone else who began participating here in the same month.
AFTER EDIT: Link to an old discussion I tried to stir up, on a topic which other users are discussing in this thread, "Ask HN: What Kinds of Comments Should Be Upvoted?"
I'm always interested in how to emphasize the positive by upvoting early and often when I see good stuff here, so I like to hear opinions from all of you about what is good stuff.
P.S. By following the links from the submitted site, I am reminded that patio11 and I seem to have the same join date. I'm honored to be behind him in any ranking that includes the both of us.
I think you're missing an important metric: karma per submission. People with high KPS tend to be those who keep their mouths shut most of the time but when they do have something to say it's generally worth paying attention. I think this is the kind of behavior that really ought to be encouraged more than contributions per unit time.
Not sure if you strictly mean karma per submission, or karma per comment, or both. As for karma per comment at least, reaching a critical mass of absolute karma points seems to have a big impact on your KPC, regardless of the comment's content, thus partly invalidating the value of KPS/KPC.
I just anecdotally see karma peasants post comments that are functionally identical to comments from the karma kings, and the peasants just get buried at the bottom. It would interesting to have some kind of NLP-based crawler go through HN and see to what extent that actually happens though...maybe it's just my imagination.
The problem with karma implementations is that they don't always encode for the best outcome. I am not sure what karma means on HN. There are some topics where you could very easily fish for karma by going with the prevailing thoughts in any given community. Post against the grain and loose karma. It's fairly predictable in most communities.
Then there are effects as I've seen in some of the Stack Exchange sites. One that comes to mind is ServerFault. Moderators and high-karma members with elevated privileges have turned that particular site into a nearly unusable and hostile environment for anyone outside a very narrow definition of what the "gang" wants.
This is an interesting effect because it encodes what a few privileged users want as opposed to what --conjecture-- a larger majority might be looking for out of the site. And, from a strict business view of the problem it reduces that audience and the applicability of the site/s by a potentially significant factor.
It's like having a mechanical engineering site where high-karma users have decided they only want to discuss hex bolts used by aircraft mechanics. You show-up to ask about hex bolts used on motorcycles and watch out!
Perhaps the ultimate questions --on a site-per-site basis-- might be:
- What's the design intent behind karma or user reputation?
- What is it you want to encode?
- How do you ensure that this is, in fact, encoded into the rating?
- How do you prevent users from polluting or de-valuing karma through
misinterpretation, petty/spiteful voting, "gang-like" behavior or other
degenerations?
- If karma is used to elevate user rights, how does it also affect
responsibilities? Does it impose a level of accountability?
- How do you keep karma "wars" from affecting your community? By this I
mean gang-like behavior by regulars who'll attack outside their pack
with the only tool they have, karma.
- How to you keep karma from affecting site business?
- Short of the benevolent dictator model, is there a better approach to
encoding user value, participation or relevance?
Not an easy problem to solve as most communities have discovered. For the most part HN seems to regulate rather nicely. We are all guilty of veering off here and there but eventually things come back to a reasonable center. Maybe it's because posts age very quickly and it takes a lot to stay in the first page or two. Other types of sites don't have that advantage.
It's just a number in a database. What you say and how you say it is more important than what other people think of it. If your words are important to you and you said them in earnest, why does it matter?
Side note: You may even be unconsciously censoring yourself in fear of losing karma. It's a similar effect to suddenly having a lot of followers on Twitter.
Sure, but karma is an important measurement of the (perceived) quality of your posts. If a comment gets upvoted that's not only because people agree with you but also because they think your comment has value. Likewise, if a comment gets downvoted, that's a fair indicator it did not have a lot of merit.
Of course, HN distorts this somewhat by just having one score for all the possible reasons a user might have to vote on a comment, and inevitably superstar posters get upvoted reflexively often drowning out better comments. But overall, as both a tool for measuring quality and a reward system it works pretty well.
By voting, we give each other points of a credibility currency that transcends individual discussions.
I just got a measurement of my post, simply because you replied. ;)
Just by talking to me, you've let me know what I said was worth at least a reply.
This is just opinion, so take it for what its worth... I'd much rather someone reply to me, email, chat whatever, and let me know what exactly they think rather than assign my words a value via a mouse click. Wouldn't you?
You're confusing a single conversation with the ongoing presence on the forum. Karma is an imperfect but effective way of assessing a user's position within the community based on her/his previous actions and reactions. It can be looked at as a substitute to an auditory/visual first impression in the "real" world.
> I just got a measurement of my post, simply because you replied. ;)
Yes, but I also upvoted you ;)
I essentially agree with you, it's just that I think karma has its place, too. And usually when I reply to a comment, I make it a point to give it a vote as well.
> Side note: You may even be unconsciously censoring yourself in fear of losing karma.
I've thought about this, actually, and based on my comment/karma history, it appears that I'm too dumb to keep from blurting out things people downvote.
I'm free of the tyranny of the hivemind by virtue of ignorance and stupidity!
You may even be unconsciously censoring yourself in fear of losing karma.
I consciously say things I wouldn't otherwise because I know I have karma to burn. It's when you have < 100 karma you have to watch out otherwise you could end up in negative territory quickly and your posts be dead by default (the idea of voting people down for disagreement rather than stupidity is still going strong, it seems).
I have always wondered what the top karma list would look like if it was broken down into two lists: comment karma and submission karma. Has anyone done this?
Show how many people / what percentage have 500 karma (which presumably is what it takes to downvote.)
Show how much karma a person would currently need to be at the 25th, 50th, 66th, 75th, 90th, 95th, and 99th percentiles (or whatever percentiles make sense).
Most interesting thing to me: this month is my fifth anniversary with an HN account. I actually started lurking before that, in the Startup News days, but I didn't really have anything to contribute until it broadened scope.
Not sure if it's a bug or my browser is just too old, but on Safari 5.1.7 (OS X 10.6.8), the "See your Data" button doesn't do anything. Works fine in Chrome.
I've got a user CSS file that just colors that text the same color as the header. It's not perfect - for instance right now I'm on my iPad (no browser extensions) and I can see it - but it sure helps.
This is a very good tool to track patterns in the community. Realize that each of the top 200 users represent a given demographic. If you see the change over time for each user, you can pinpoint those patterns and adjust to it (if you are selling something to that market). You can also quickly learn how to create your own organic circle of upvoters, so that your company posts hit the front page without needing to email everyone in your network so they can come in and upvote (which is do not approve of and not do). Its good to create your own niche audience inside such a powerful community like this one.
michaelapproved is not a valid username when searching but MichaelApproved is. Does HN allow for different case usernames or is it something that was overlooked during development of the search feature?
Looking at the news.arc code, user profiles are stored as files in the directory arc/news/profile/, so the names there are case sensitive because of the underlying file system. The users are loaded into a map using the name as a key, and map lookups are case sensitive in Arc. Thus, both the on-disk and in-memory storage of users is case sensitive. Of course names could be converted to a case-insensitive form, but the "natural" implementation results in case-sensitivity.
The code can be downloaded from http://ycombinator.com/arc/arc3.tar
Note the code is from 2009 and didn't exactly match the running HN code at the time, so any conclusions from the published code aren't guaranteed. I did an analysis of the HN ranking formula back in 2009 - http://www.righto.com/2009/06/how-does-newsyc-ranking-work.h... - but the live formula is surely very different from the published code by now. And ironically, that article got about 0 upvotes so it didn't rank well :-)
Usernames could still be normalized to lower (or upper) case before creating/reading/writing the file. (Admittedly, the simplest method is without doing this.)
Thats because someone typed in that username with a trailing space character ("mtgx ") and it made a new entry for that username int eh database. Nothing a little .strip won't fix...
For an account only 290 days old, mtgx has a ton of karma. Any idea who that is or where the karma comes from? Don't see many highly rated posts or comments in their feed.
Submitting stories. Whoever submitted Bellard's jsLinux story received as much karma (3000+) for that one URL and title as I had from several years of comment writing.
pg (129961)
As of: Mon Feb 04 2013 07:34:16 GMT+0200 (IST)
Among all active users
Total users: 13431
Total users with less karma: 13430
Your rank: 1st
Overall karma percentile: 100.00%
Average karma increase per day: 56.260
Total users with slower increasing karma: 13427
Your speed rank: 4th
Speed percentile: 99.98%
Among active users from your registration month (October 2006)
Total users: 9
Total users with less karma than you: 8
Your rank: 1st
Percentile: 100.00%
I know this is going to sound preachy, but the best way to improve your karma is probably to stop thinking about how to improve your karma.
That is, just post your thoughts on a topic, written as best you can. At least, that's worked well for me and probably most other high percentile-ers. Don't start pandering to what you think will attract a lot of upvotes- karma is meaningless after all.
Cynically: many stories are reposted. Discover what those are, and what the popular opinions on them are. Then monitor new, and make the popular comment early to reap most karma.
Commenting at the right time seems to help.
Having a single thought per post also helps; people tend to upvote if they agree with the entire post.
The fact that this is not using Twitter Bootstrap makes me so happy. It's delightfully refreshing to see something that uses even just the barebones rails scaffolding styles.
Neat. One thing I noticed though... my karma isn't correct even though it states it was last updated today. I haven't commented in almost a week prior to this comment.
it lists total users as 11777 that looks like a very small number for hn which sends 10-20k visits if a site hits top spot. or are there a lot of lurkers who never signup?
Since HN doesn't publish a userlist, this site gets its data by crawling the "latest posts" HNSearch API every ten minutes. That 11,000 users includes all users who have made a post on HN since January 25th. Thats right, 11,000 unique users have means post on Hacker News in the past 2 weeks.
It's not entirely accurate because of this. HNSearch API allows you to list users ordered by karma. According to that, I'm ~120. According to this site, I'm 94. Heh.
ah good to know. in that case 11k are the total active users in last 2 weeks. so going forward would your service retain data from 25 Jan onwards or only for the past 2weeks.
I'd think that if it retains data from 25th within a few months it would be able to show a very good estimate of total active users
Would be even cooler if I could get stats of my most popular threads and comments. It would be also interesting to see pg's most popular thread and comment.
But it would be nice if all these "small" (not criticizing in any way here) websites allowed either query parameters in the URL or directly "short URLs" (like bit.ly etc.) so we could share links.
Well, this site may not have it in short URL form, but if you do need to shorten something else without too much of a hassle, there are lots of bookmarklets out there :
> Speed percentile: 98.82%
While I have to acknowledge that some instinctive/animalistic part of me is excited at being in this percentile, it also prompts me to remember that karma is effectively worthless and I really need to get busy on class work and research...
It's really strange the impact something silly like karma has, because I tell myself I'm above accumulating imaginary points on the internet, but obviously, I don't seem to be.
In some ways, it's curious. My highest rated comments aren't what I would consider my best. Instead, they are ones that appeal to a wide audience with some kind of emotional (rather than technical) insight. My last comment, a quip about about how every programmer thinks they're the best, was worth 64 points. However, a detailed post I wrote on molecular dynamics simulation where I fact-checked and reviewed the literature was only worth 3 points. Go figure.
EDIT: After thinking about it some more, I've decided karma isn't entirely worthless in real life. It is a metric that lets you know how well other people like your written work. Over time, I've slowly come to realize what will get upvoted and what will not. This ability could have interesting repercussions in the way I write about my research, or the way that I craft something like grant or fellowship applications. Every person is different, but posting on HN reveals to me that, in the aggregate, there are certain things you can say that have a certain appeal to others, and the capacity to recognize this could in fact be useful in many life situations.