Comment by aristus | Hacker News Reader

Comment by aristus | Hacker News Reader

McKinsey suggests and practices a rule about growth: don't grow more than 25% per year if you want to preserve your culture. I think once HN started having self-reflective posts it passed into "Shirky Completeness", a stage in communities where the old hats become preoccupied with preserving the culture and cannot keep up with training new users to fit their mould. Sub groups fork off to the point that the several concurrent streams stop interacting. My friend and I realized we were reading completely different aspects of HN.

Karma is (or should be) a proxy for quality. But there are no mechanisms except social ones.

I also suspect that there are certain commercial "reputation" companies that have HN in their sights now. Not spammers per se, but father along the continuum to PR. There have been quite a lot of activity geared toward "whipping the froth" and post-storms around specific topics.

It may be time to fork HN. I actually also suspect PG has already forked HN for private YCombinator use. I haven't seen a post from RTM or TLB for over a year.

I'm by any standard an absolute rookie here and I realize that gives me 0 standing on how things 'used to be', but as compared to how things are elsewhere HN is a very large improvement.

That might be reflective of how small the site still is but at the same time it seems that the quality of the discussion is well above most tech oriented subcultures out there.

More users will mean more submissions, eventually the 'new' queue will go by so fast it will be harder to pick out the good stuff, hopefully enough eyeballs on the new submissions will keep the spam and the astroturfing down to a minimum.

And keep in mind that even a user with huge 'karma' can have a bad day, it's an average, not an instantaneous measurement of quality.

There is clear evidence of what you classify as 'whipping the froth' (beautiful metaphor):

For instance there have been many attempts to get the google friend tracking feature listed, at some point I saw three submissions on the 'new' page at the same time and a whole bunch of them that had already passed by, clearly that is not helping at all but it does not seem like anybody cares enough to do something about it. The more that gets a foothold the harder it will be to get rid of it.

Outright spam could be easily handled by making it impossible to submit links until you've crossed some karma threshold, linkfilter.net has used that method fairly sucessfully for years.

Good point that karma as applied to users is an average. I've done some flaming here I'm not proud of. I meant that points applied to posts and comments are supposed to be a measure of quality. If we start going solely on reputation of the poster we've merely set up a pantheon and kicked away the ladder.

You're right that the velocity of the /newest page is the biggest problem. There may be a UI way of encouraging more upvotes. But I think it's still down to culture if we want to maintain the overall quality.

People started worrying about preserving HN culture about 6 months in. It's one of the constants of the site. Which is not to say there's nothing to worry about. Just that people worrying is not necessarily a sign there is.

There's no separate fork of HN. Rtm never posted, and you can easily test your statement about Tlb:

http://news.ycombinator.com/threads?id=tlb

I wonder if calculating a user's karma as a weighted average of the scores of their last N comments, rather than a sum, would help. Inasmuch as karma scores affect anything, it might gently encourage fewer, more insightful comments.

Also, some comments stimulate interesting sub-threads. In cases where something is voted way, way down but has positive responses, it's probably trolling; positive with a lot of positive responses could mean "thought-provoking".

Of course, the real trick is to encourage interesting discussion and not just an echo chamber...

I've been working on something very much like this for the last couple days. Stay tuned.

The weighted average of recent posts approach to karma sounds like a good idea. I imagine that might change my own posting behavior in a way that I hope is liked by all.

Hey Paul, just putting down a quick observation as why this could potentially penalise some users based on geography (That is if you have it so features are enabled/disabled based on your average karma instead of total)

One of the things I've noticed is that us on the other side of the world have trouble generating the same kind of karma that other's can generate which is really a function of being too late to discussions.

For example, east coast USA is often around 14 hours behind my timezone, so I typically get to conversations about 6-8 hours after they've started and when people are usually heading to bed.

Generally I find my comments down the bottom of threads with 1 or 2 karma.

However, if I post in the middle of the night (usually as I'm heading to bed) or really early in the morning (essentially as I get up) I am in the peak hours of HN, my comments on average do very well for Karma, simply because more people are available to read them and as such vote them up or down accordingly.

Just a thought - you might want to have thresholds based on a person's "prime time" instead of a predefined average.

Yeah, I know about this problem. I hope to cook up some kind of solution eventually.

ok cool - just highlighting it ;)

how about something like this:

http://news.ycombinator.com/item?id=467739

Have you thought about publishing a subset of HN's data so others can experiment with it?

There are a couple archives of all collected posts, though in quick searching I can't seem to find the right keywords. (Haven't had my tea yet though...)

That's a very clever idea. Although with any calculation like that - there is still the bobcat issue: http://xkcd.com/325/ I really like the concept though - A running average of karma forces those who post lots to post quality and rewards those who are new but post well.

Past attempts to fork it with newmogul.com (nickb) and nonhackernews.com (mattmaroon) didn't work:

http://siteanalytics.compete.com/ycombinator.com+newmogul.co...

Ah yes, but those were attempt to fork the non-HN crowd away from HN.

No one should be surprised that doesn't work. it's the old timers that are dissatisfied with the new crowd not the other way around.

Forking away the hackers works, reddit did it to slashdot, proggit to reddit, HN to proggit and reddit.

I didn't say forking inherently doesn't work. Both nickb and mattmaroon are old timers who wanted to create valuable communities themselves, not just to create distractions and attract a lesser crowd or something.

I like New Mogul well enough, and am learning which submissions to prefer to send there and which to post to HN. I learned about nonhackernews.com from your post, so thanks for the information, and now I'll see if that site will fit some of my desire for online discussion with smart people who discuss topics that don't fit well here on HN.

I suspect that part of the problem is in the execution--our collective social will simply isn't sufficient to maintain an awesome community as time (and users) go to infinity.

Along that line, I've been experimenting with some more direct controls for community control: scaling voting power with user intelligence as measured by an objective standard. I posted about it here: http://news.ycombinator.com/item?id=464700

http://news.ycombinator.com/threads?id=tlb

You are right -- tlb is active again though rtm is not. I meant that I hadn't noticed them in the mix in quite a while.

I don't think forking the current concept would work. The Digg/Reddit model seems to have had its best time. Some new concept should be created. Twitter seems like the right direction to me because you can follow only the people that you think are interesting.

FriendFeed seems to be actively building on that model, with their commenting features. What worries me is that this inhibits meeting new people far away from you in the social graph: I'd like to hear PG's or Paul Buchheit's opinions, even though I don't know them in real life.

Reply on news.ycombinator.com