This is sort of interesting to look at, but not too reliable: it only shows users whose profiles happen to be swapped in to the in-memory Arc data structures whenever the list is regenerated. So from day to day, depending on post and caching and HN restart activity, different people are included or omitted. For example, at the moment raganwald isn't on the leaderboard, even though he has >40k karma, presumably because he hasn't posted in a bit, so his profile is swapped out of memory. I filed a bug about that but it got a WONTFIX [1].
I ran into that while putting together a profile of prolific HN posters [2]. I put it together a bit at a time over several days, and noticed that people were appearing and disappearing from the list, which seemed odd. I ended up finding a third-party list on Heroku that was more stable day-to-day [3]. However it seems not to be updating any more– the numbers are now a few months old.
This reminds me of those lists middle school boys and girls would make within their social circles. "Cutest Boy" or "Nicest Clothes" ... all pointless when looked upon as an older person, but at the time it was all that mattered.
What's interesting to note here is that the points totals have been removed for the top 10. It used to show vast differences between the "leading pack" and the rest... of the order of 20k karma points or so!
I guess the rationale for this change is to stop the silly competition (if there was any) once at the top of the pack. I'm not sure there ever was such a competition, mind you. I was #7 once upon a time and it never affected my posting habits.
It's sort of helpful to be able to click a profile and see their karma, to get a sense of whether they're, I don't know, invested in the site? But it's not worth the creepiness.
Getting rid of scores on the top 10 was a step in the right direction, but ultimately the long-term goal should be to get rid of scores for the top 2000.
Slashdot used to have a thing where your karma would eventually just say something like, "Lots".
Interestingly, within the top 100 there is a weak positive correlation between the total karma, and the average karma per comment, implying that the people at the top of the leaderboard are actually posting quality content, and not just making up for it with quantity.
The closest power law fit is
predicted average score = 0.27 * (total score) ^ 0.25
which has a correlation of about 20% with the actual average scores.
Either that or people with higher average karma per comment are posting their comments earlier so they get more upvotes. Or HN's comment sorting algorithm is privileging users with higher total karma, so those users have their comments appear above other comments that have the same number of upvotes. Or users with better name recognition are more likely to get upvoted.
I find this slightly embarrassing. It's also broken. For example, I switched from my old account 6 months ago and it's still in the top 30.
The main reason we haven't done anything about it is that it's not clear what to do. For example, we could make it use a shorter time window than "forever"—but do we really want to gamify karma further? I doubt it.
Just think what we can do with all those imaginary points.
I'd like see some kind of solution where submissions don't disappear so quicky.
Maybe using tags that become sub-forums of sorts and the front page is just an aggregate.
Like tags for SSL, ddos, security, centos, payment-solutions etc.
Tags are not hard to implement, there could even be a limited set to choose from, that only certain users can add to, if there is any concern about new tag creation abuse.
And then a submission can span multiple tags, it may get buried under one tag but rank better under other tags.
https://lobste.rs/ does that, but the idea has failed to find any support on HN over the years.
In terms of "doing" something with the points, I quite like Stack Overflow's approach of letting users with higher reputations do more things. There is a little of this on HN (e.g. flagging), but on SO it includes things like editing posts, deleting comments, review flags, etc. But that would lead to a broader, less centralized system than YC is likely to prefer, and it's their site and decision ultimately.
I'd like see some kind of solution where submissions don't disappear so quicky.
I think it would be great if there was an "old" button next to "new" in the header, in which submissions of the past month (or so) show up, ordered by a score that allows for old stories to be bumped back to the top, like
points * age * comment rate.
It would probably also help if new comments would be highlighted.
Not really what you describe, but it's along the lines of "give me an alternative to the frontpage or /new." I don't know precisely how it works. I think it gives more weight to votes from older accounts, but that's just a guess. It does seem to be a slightly higher quality frontpage, but that might just be confirmation bias. After all, if it were truly better, then /classic would be the actual frontpage.
There's also http://hckrnews.com/ which shows a time-ordered list of all submissions that reach the frontpage each day. I prefer it to HN's frontpage, both because you can scroll back over the past few days and because there aren't that many submissions that reach the front page each day, so it's easy to quickly scroll through a list of all of them to pick out anything of interest.
My thought was that the comment rate factor would allow for old stories to be bumped back to the top. Highlighting new comments would allow to continue old threads. I clarified my comment above in that regard.
I really like the average points tracking in Hacker news.
Its interesting to note that some people got on the leader board via only commenting when their posts were relevant, the avg. 5+ points people and some got on their through just making sure their voice was heard, even if they didn't have anything relevant to say, the people with an avg below 3 points.
I've always treated HN's a bit differently from other message boards in that I only comment when I think I have something relevant to say, which is probably driven from the average point tracking system.
I'm actually surprised at the number of leaders with avg scores under 2. I'm guessing these people treat HN's like a regular message board and just comment on everything.
The average of the top 100 seems to be 4.3 so it looks like
I do not think this is a fair characterization. Lots of valuable comments don't get a lot of votes.
Giving feedback on a less popular "Show HN" is a good example. In most cases, you'll just get one vote, from the person who posted the story. Sometimes they don't even bother giving a vote, so you'll get nothing. If they ask a follow-up question and you reply, you probably won't get any votes at all because your comment won't be generally useful, just useful to the person with the question. For example, this series of comments, for which I received not a single vote: https://news.ycombinator.com/item?id=8009176
It also goes the other way... here's a comment where I received 13 points for doing arithmetic: https://news.ycombinator.com/item?id=7720950 (Actually, I didn't even do the arithmetic, Wolfram Alpha did.)
Like anybody, I have a reflexive tendency to feel pride when a number goes up, but pride in my average-point ranking is tempered by the knowledge that it's a statistic-of-convenience. It's not necessarily there because it's worth tracking or optimizing. It's there because the laziest possible stat, "total points," which tends to scale linearly with tenure and with volume of posts, became boring. So now we also have the second-laziest possible stat, "total points divided by total posts".
I have nothing against the simple stats -- why waste time on complex pseudoscience when basic pseudoscience will do? -- but the risk is that we'll design a stats page by adding numbers one at a time, in laziest-first order, until the screen starts to look cluttered, at which point we'll pick the most promising pattern we've spotted so far and start trying to optimize it. This is a great way to generate Powerpoint slides with lines that move up and to the right, and a lousy way to measure value. The micromanagement of statistical placebos is the occupational hazard of our age, and we should try to push back.
It's easy, too easy, to game my average-points ranking. I should contribute only popular thoughts to popular posts. I should assiduously avoid commenting on anything that's not on the homepage, or replying to people who are not near the top of a thread. I should make sure to comment within the newly-arrived-on-the-home-page time window, which is presumably now measured in milliseconds. I should certainly avoid giving advice to people on Ask HN.
I'm fairly certain they changed the ranking algorithm for comments around April/May. My average used to be 10 (it's 4.5 now), and this meant that I could comment on a thread and it would essentially always be the top comment, even if it only had 1 point and the comments below had more.
This created a positive feedback loop - because the top comments get read more, they also get voted up more (as long as they're not blatant trolls/spam). So once you had an average of about 6 or more, it was very easy to maintain and rack up points.
They seem to have changed the algorithm them - I remember when they did it, because I noticed the effects almost immediately (not just on my own account, but on others' accounts, like tptacek and tokenadult).
Overall, this is probably a change for the better, since it mitigates the "rich get richer" effect of commenting frequently while still rewarding people who consistently contribute valuable comments.
I tallied locations ~9 months ago [1]. Of the top 100 posters, I could determine the location of 81. Those broke down as 54 US, 27 non-US. Mostly from English-speaking countries: US/Canada/UK/Australia together accounted for 69/81 (85%).
Actually, pg was on the leaderboard for years. But, yes, his name was eventually removed, and while it was on the leaderboard, it was typically the top username by a very wide margin.
Oh, sorry, I was thinking of nickb. :) Thank you for checking the Wayback Machine. I must have been aware of pg's karma totals by looking at his individual profile.
What's the motivation behind posting these? Is there a way to say "these are the kinds of posts that are good for the community" without tacking a number on it?
I would guess it correlates highly with number of posts, so I just think of it as a proxy for "highly active long-time users". Another measure could just be postcount. Karma is like a slightly "corrected" postcount that is supposed to boost people who post highly regarded posts, and downrank those who post a lot but mostly junk. But I would guess just raw number of posts is still the main explanatory factor in the ranking (I don't have the data to test this guess, unfortunately).
Trivia: It used to be one of the main links at the top of the page about 2+ years ago.
I set a goal several years back of getting into the top 100, then top 50, and so on, and was commenting and posting loads. I guess they didn't like that too much and it disappeared onto the Lists page ;-) I don't post as much now but that's probably a good thing for me.
I set a goal several years back of getting into the top 100, then top 50
Why? Oh, why? I look at my own HN posting as a personal weakness that does nothing to advance my career, general happiness, or sense of personal development. The really successful people I've known in my life don't get caught up in tussling in Internet forums. They stay focused like a laser on goals that are much more recognizable as "success".
And that reminds me to close my browser and think about what I really want to accomplish today!
My time on HN has not been as negative as I infer yours has been though. It benefits me and I enjoy it. That back-handed airy-fairy malarky about what "really successful" people do sounds like a load of preachy cobblers to me though.
I've always kind of found the total points to be a bit irrelevant. But I do take karma into account when posting which probably would prevent me from ever breaking into the top 100 of total points but does prevent me from posting every silly thing that comes into my head[1]. It seems karma's a more relevant metric for value then total.
Not to speak for Peter, but some people like to set goals because the effort to achieve them might reveal interesting bits about the mechanics or psychologies involved, including their own.
Some of us tend toward the product side, some the process side, and some the people side. The people side is the most visible and most likely to arouse suspicion among geeks, but it's not inherently worthless.
And, of course, self-branding and promotion is a thing that people do.
If given the choice to target users based on GMail data or targeting users based on Facebook data, I would take the former. GMail data would provide you with a much clearer and nuanced profile of someone, even if GMail has only 1/3 - 1/2 the user base. Facebook should do well with consumer brands though.
I ran into that while putting together a profile of prolific HN posters [2]. I put it together a bit at a time over several days, and noticed that people were appearing and disappearing from the list, which seemed odd. I ended up finding a third-party list on Heroku that was more stable day-to-day [3]. However it seems not to be updating any more– the numbers are now a few months old.
[1] https://github.com/HackerNews/HN/issues/49
[2] http://www.kmjn.org/notes/hacker_news_posters.html
[3] http://hn-karma-tracker.herokuapp.com/overall