Kuro5hin.org: technology and culture, from the trenches
create account | help/FAQ | contact | links | search | IRC | site news
[ Everything | Diaries | Technology | Science | Culture | Politics | Media | News | Internet | Op-Ed | Fiction | Meta | MLP ]
We need your support: buy an ad | premium membership

[P]
Can a Community Based Web Site Scale?

By sethadam1 in Meta
Tue Jan 28, 2003 at 10:37:58 AM EST
Tags: Scoop (all tags)
Scoop

Can K5, as a community, scale?
--
Scalability: An architecture is considered to be scalable if, unchanged, it can handle increasingly complex problems or strain that demand a greater amount of knowledge or resources.
Scoop: Scoop is designed to enable your website to become a community. It empowers your visitors to be the producers of the site, contributing news and discussion, and making sure that the signal remains high.
 


I've been reading kuro5hin.org for about 6 months now.  I kept hearing mention of it on Slashdot.org, or what is affectionately known here as "the other site," and finally I followed a link.  I've become familiar with the culture, complete with inside jokes, recurring themes, popular characters who may or may not actually exist (rusty, Inoshiro...), and even the periodic observation about the site itself in the form of parodies and other miscellaneous humor.  I became enamored with Scoop, and began investigating Scoop hosting and other scoop sites.  

Watching K5 work, you inadvertantly learn a lot about not only the way Scoop works, but also about the readership of K5.  Certain types of articles are consistently favored, while others are frequently frozen out.  Certain types of articles make it to the front page more often, while others are doomed to the section pages, which are less frequently read and therefore, there is less active discussion.  It's interesting to note the psychology behind the gears that turn this site, to watch the way people react.  In essence, I could hypothesize that people watch the way K5 works, and then begin to "predict" how others would vote and vote with the crowd when they don't have emphatic opinions, since all voting is public.  After a few months, I felt confident I understood, mostly by reading and watching, less by participating, how the K5 community works.  

The only thing that I found surprising were how many articles were in the queue and how few seemed to show up on the actual site.  In the K5 RDF feed, it's not uncommon to have only one or two new stories a day.  Other sites that many K5'ers frequent, like the other site, OS News, and the Register are updated with many stories daily, some sites a few times an hour.  Why, then, were there so many stories and so few updates?  

I finally submitted my first article to the K5 queue the other day to see what would happen.  I was surprised, to be honest.  Over 640 people read my submission and voted, and the "score" of my submission continued to hover in the 40s for 36 hours until it magically disappeared never to be seen again.  I wasn't bitter - it was an edgy humor piece and more people than I predicted were not as amused as I had hoped.  More people than I expected were more forthcoming under the anonymity of the internet and were all too anxious to tell me my submission was neither funny nor creative, etc, etc, etc.  This is not a revenge piece: I can certainly live with that.  Not every piece is front page material.  

What occurred to me, though, was that a small section - a few hundred readers - determined the fate of the article.  While the piece was never posted, about 300 people ultimately voted they'd like it to be posted, a little over 50% of the non-abstainers.  It's certainly fair to assume, therefore, that about half of the total K5 readership might have found it, at a minimum, "worth a read," if not worthwhile.  Take into account, though, that many of the 640 aforementioned users are frequent moderators, and, as discussed above, have an idea of what they expect will be published and what will not.  Taboo topics, as you see, can quickly become relegated to the "unseen" part of K5, where unposted articles go to whither and die.    

As someone who is neither a "regular" nor a true K5 insider, it feels a bit futile watching articles go from the queue that do not fit a standard bill.  In fact, lately, it seems the only articles that fly through the queue are either sternly political in nature, well documented research pieces, or K5 self reflections.  I've seen a number of articles in the queue that I've found entertaining, informative, or just plain discussion-inspiring, and they've been shrugged off.  And I rarely smile when reading K5 articles anymore because the overwhelming majority are strictly serious in tone.   As a site like K5 gets larger, more distributed, less centralized, and more diverse, I'm watching the psychology change.  I'd love to see a report that contains a ratio of stories submitted to stories posted.  

My personal preference, and it is purely my personal opinion, is that overposting is okay.  During the course of the day, I'd rather see 5 pieces I don't care about posted than just one I do.  Volume is important to many people on a site like this, because it allows you the choice to not participate in discussions that don't interest you, but still always have fresh discussion available.  At times, it feels as though too much of the bulk of the site is in either the queue and the diaries, and not enough is posted to the general site, if only to generate some fun discussion amongst what are clearly skilled, bright, diverse people.  I've found myself reading the K5 QUEUE for content, not the actual site, because many of the articles submitted, even the bad ones, are often worth the 2 minutes it took me to read them.  This is not a professional news site like CNN, and it's okay to have an article that is less than perfect if it sparks good discussion or gives everyone a laugh.  

Many sites don't seem to restrain themselves like K5 - yet, K5's Scoop posting threshold is the same.  As the site gets bigger, loyal readers, I think, feel a responsibility to the site and want to regulate what is posted more harshly to keep the quality of the site high.  More readers should mean more proportionally more submissions, and I'd anticipate the same overall number of articles accepted.  But it doesn't appear to be the case.  It seems more like the bigger the site, the more the community feels protective of the site, the more exclusive they become, the fewer articles appear to qualify under the more rigid criteria.  Admittedly, few websites where the community is in charge have the readership that K5 does, and so there is an utter lack of evidence to make an informed decision about scalability.   Still, it makes you wonder - can K5, as a community, not as code, scale?

Sponsors

Voxel dot net
o Managed Hosting
o VoxCAST Content Delivery
o Raw Infrastructure

Login

Poll
What's the Best Course of Action?
o Quit Yer Bitchin, It's Just Fine the Way it is 36%
o Amen! Lower the Posting Threshold 17%
o Exclusivity is OK in Moderation 18%
o We Just Need More Submissions! 22%
o "Upgrade" to Slash 5%

Votes: 58
Results | Other Polls

Related Links
o Slashdot
o Scoop
o Kuro5hin
o Slashdot.o rg
o rusty
o Inoshiro
o parodies
o miscellane ous humor
o Scoop [2]
o other scoop sites
o K5 RDF feed
o the other site
o OS News
o the Register
o Also by sethadam1


Display: Sort:
Can a Community Based Web Site Scale? | 81 comments (71 topical, 10 editorial, 0 hidden)
On the other hand... (4.50 / 2) (#3)
by mstefan on Mon Jan 27, 2003 at 08:16:46 PM EST

If you start casting a wider net, so to speak, you're going to end up with another slashdot where the signal-to-noise ratio is all out of whack.

Personally, there are days where I enjoy just scanning and reading the front page stuff and their related discussions without digging through a ton of muck just to find a single pearl. But when I am in a "muck" sort of mood and want to get down in it, I check out the submissions and I vote. The fact that articles that I comment on may not be around for more than 36 hours, or gets relegated to the backwaters of the section pages, doesn't really bother me. I say my piece, listen to others say theirs, learn a few things and move along.



Yes, it can scale (4.00 / 5) (#4)
by bc on Mon Jan 27, 2003 at 08:22:31 PM EST

I think this has been demonstrated countless times already.

Kuro5hin, a medium sized blogging site, is hardly a pioneer and is hardly entering uncharted territory.

♥, bc.

methinks you didn't read the article (4.33 / 3) (#9)
by muddd on Mon Jan 27, 2003 at 08:52:19 PM EST

it's not about code scalability, it's about the community.  

none of the links you provided are community based the way the author describes it.

muddd
[ Parent ]

Sure they are (4.66 / 3) (#10)
by bc on Mon Jan 27, 2003 at 08:57:20 PM EST

Of course they are community based, they demonstrate what happens to any internet "community" when it gets larger - they split up into smaller communities within the site structure. This will happen with k5 just as it has on many other sites. it is human nature.

♥, bc.
[ Parent ]
hence the fiction section... [nt] (3.50 / 2) (#12)
by kpaul on Mon Jan 27, 2003 at 08:59:48 PM EST


2014 Halloween Costumes
[ Parent ]
Excellent point, bc (4.83 / 6) (#21)
by leviramsey on Mon Jan 27, 2003 at 09:50:54 PM EST

What I'd like to see on K5 is a huge increase in the number of sections. Maybe have sections be created by simple a simple majority of active users. The section bar at the top would be generated every hour with links to the 11 section pages with the most recent articles posted. Sections that no one ever posts to would thus only be accessible through a page that lists all sections.



[ Parent ]
Interesting. (4.50 / 2) (#24)
by Greyjack on Mon Jan 27, 2003 at 10:20:43 PM EST

Thinking ahead...  at some point then, perhaps, separate moderation queues for each section?  That's a very interesting thought -- would be a hassle (although far from impossible) to watch all of them ('specially if "moderate all sections" wasn't a choice); instead, people would most likely moderate more heavily in the areas that interested them the most.

Could make for a higher level of quality for stories in each section, assuming people gravitated to where there areas of expertise typically lay.

Could also make for chaos, as it would be easier to muster up a group to blitz a specific section.  There's absolutely no precedence for such antics, of course.

Mind you, I'm not even remotely advocating such a shift--I'm just toying with the idea of alternative mechanisms (and ramifications) as things get larger.  ('tis the meta section, after all)

--
Here is my philosophy: Everything changes (the word "everything" has just changed as the word "change" has: it now means "no change") --Ron Padgett


[ Parent ]
the 'News' section for example... (4.00 / 1) (#32)
by kpaul on Mon Jan 27, 2003 at 11:19:58 PM EST

I wish there was a way to speed the process on a breaking news type story. It's kinda like a print newspaper. By the time it's reached the front page, it's yesterdays news. Unless, of course, the story is so good it slides through, but I haven't been around long enough to know how often that happens with a News type story.

Maybe if a News story got a certain amount of votes within, say, 6 hours, the author could decide to release it to section only?

Just some thoughts...


2014 Halloween Costumes
[ Parent ]

Actually . . . (5.00 / 2) (#37)
by Greyjack on Tue Jan 28, 2003 at 01:03:38 AM EST

On the whole, K5 is just fine the way it is.  If the users weren't happy with things, they'd either a) make some noise and, er, encourage Rusty to make some changes (cf. the Fiction section), or b) move on to another community.

b) isn't necessarily a bad thing -- if absolutely everyone was comfortable here, you'd wind up with articles that were the written equivalent of the superbowl halftime show.

I wish there was a way to speed the process on a breaking news type story.

I think it's plenty fast overall, just not here.  Fast-breaking news just doesn't fit the K5 zeitgeist.  If there really was demand for it, someone would post a meta article, and if reaction was favorable, Rusty would eventually give in, post a poll, and we'd all vote.  But, that hasn't happened.  As it is, news items that make it through the queue make it through 'cause enough people want to argue about 'em.

Which brings me to this quote from the original article:

Volume is important to many people on a site like this...

...but not important to enough people 'round here who actually bother to vote.  Otherwise, more stuff would make it through the queue.

--
Here is my philosophy: Everything changes (the word "everything" has just changed as the word "change" has: it now means "no change") --Ron Padgett


[ Parent ]
Separate Communities (4.00 / 1) (#35)
by webmaestro on Mon Jan 27, 2003 at 11:38:37 PM EST

While K5 is like a community now, one thing I was thinking is maybe several seperate communities within K5, not necessarily now, but if K5 were to get big enough. A person could submit a story to a specific community, or to an all community queue. Then people from the seperate communities would vote if they want to allow the stories from both queues combined into their community page. The story could interest several different groups of people, and would then be posted to several communities. It would kind of be like an independant movie that any of the communities could pick up and show in their community.

Its just a thought.
--
Check out Worldofun.com. It's a world of fun.
[ Parent ]

There are no 'insiders'! (3.33 / 3) (#6)
by StephenThompson on Mon Jan 27, 2003 at 08:43:53 PM EST

Did you read the FAQ? There is no Cabal! :P

Seriously, what makes k5 different from 'that other site' is that there is no controlling authority.  There are no 'insiders' who make the editorial decisions.  It's pure democracy.

There are some bugs tho hehe (3.00 / 1) (#7)
by StephenThompson on Mon Jan 27, 2003 at 08:44:45 PM EST

that was supposed to be a response to your comment below...

[ Parent ]
is it REALLY? (4.00 / 3) (#8)
by sethadam1 on Mon Jan 27, 2003 at 08:50:58 PM EST

It's a democracy, but only a democracy of the people who choose to moderate!  Do you believe that the moderating community is a perfect cross section of the entire community? Of course not.  

When I say "insiders," I mean the loyal readers who read everything, and probably, the people who have been visiting K5 the longest.  I betcha they are NOT very representative of the rest of the K5 community.  

I could be wrong.  

[ Parent ]

A choice. (4.00 / 4) (#14)
by ffrinch on Mon Jan 27, 2003 at 09:05:00 PM EST

The other readers choose not to vote.

As with most democracies, K5 members have a choice -- they can visit the edit queue and vote, or not. Voting isn't compulsory in countries like the USA, and even in countries where it is compulsory, it's very easy for someone to slip in an empty ballot or make a donkey vote. A person only votes properly if they care about the outcome.

At K5, stories are voted on by the readers who care enough to vote. It might not be a true cross-section, but it's not going to change.

-◊-
"I learned the hard way that rock music ... is a powerful demonic force controlled by Satan." — Jack Chick
[ Parent ]

Yes. (none / 0) (#25)
by NFW on Mon Jan 27, 2003 at 10:49:08 PM EST

It all depends how you define community. At K5, the moderators are the community.

Your story wasn't voted up. That doesn't mean that K5's community doesn't scale. That just means that your story was not the sort of thing that K5's community wants very much. And K5 is set up so that the only stories that succeed are those that are wanted very much.


--
Got birds?


[ Parent ]

And another thing... (4.00 / 1) (#31)
by NFW on Mon Jan 27, 2003 at 11:13:40 PM EST

When I say "insiders," I mean the loyal readers who read everything, and probably, the people who have been visiting K5 the longest. I betcha they are NOT very representative of the rest of the K5 community.

I argue that loyal readers are, almost by definition, well served by the moderators. And as I mentioned in my previous message, the moderators ARE the community that K5 exists to serve.

If a reader isn't represented by the moderators, the read wouldn't bother being loyal to K5. They'd be reading stuff somewhere else, where they ARE well served by whoever (if anyone) moderates.

Why would someone not represented by the moderators bother to stick around? There are many communities to choose from, it isn't that hard to find a place that feels like home.

The fact that not everyone feels at home at a particular site does not mean that the site does not scale. It just means that the community isn't right for everyone. Communities are like that. The least scalable form of community is the form that puts a bunch of people together at random. Communities are defined by who they include and exclude, who they cater to and who they shrug off. Finding yourself a member of the "shrug" set doesn't meant the community doesn't scale.

It might means that the community doesn't include you.

It might mean that you don't understand what the community expects of its members.

It might mean something else. But I don't think it has anything to do with scalability.


--
Got birds?


[ Parent ]

I respectfully disagree (3.00 / 1) (#33)
by sethadam1 on Mon Jan 27, 2003 at 11:24:29 PM EST

I don't know why you're so easger to cast this as a piece where I'm mouthing off because my last submission didn't make it.  That's not the point, and if you thought it was, you didn't get what I was saying.  

There are plenty of people who read K5 and don't a) understand Scoop or b) give a crap.  That doesn't mean that they are not represented or looking to leave, they just don't take advantage of moderation or the other features.  I truly don't believe that a frequently moderator is the same type of reader as the majority of K5 readers.  

Judging by uids, there are over 40000 registered K5 users.  It's rare that 1000 vote on a story, even over the course of 36 hours.  That means usually, roughly 2% (or fewer) of users read the stories in the queue.

I bet a good portion of the other 98% don't even know what else is there.

[ Parent ]

misunderstandings (4.00 / 1) (#38)
by NFW on Tue Jan 28, 2003 at 01:57:58 AM EST

I know it's not about your submission, but it is about what you infer from the numbers surrounding your submission. I make very different inferences from them.

I'd wager that very nearly every K5 reader, registered, loyal, or otherwise, knows about Google and knows how to use it to find interesting stuff.

If the readers didn't enjoy the site, they wouldn't be here. There is no reward for being here other than enjoyment of the stories and the discussion they generate, nobody's holding a gun to their head, paying them, encouraging them... nobody's doing anything to keep them here, except by providing reading material. The only people who stick around are those who find said reading material interesting.

And of those 40000 registered users, I'd wager that at least half - maybe 3/4 - no longer read K5 at all. They looked, they read for a while, they signed up, said a few things, hung around a little while, found something more interesting.

Rusty, if you're reading this, can you divine how many unique userids logged into K5 in the last week? Can you guesstimate the ratio of logged-in users to not-logged-in readers over the last week?


--
Got birds?


[ Parent ]

Guesses (5.00 / 2) (#42)
by rusty on Tue Jan 28, 2003 at 02:35:19 AM EST

Rusty, if you're reading this, can you divine how many unique userids logged into K5 in the last week? Can you guesstimate the ratio of logged-in users to not-logged-in readers over the last week?

Unfortunately, I don't really know how many accounts are active. I can tell you that 2492 distinct accounts have voted on at least one story this month, and 2617 last month. Based mainly on these numbers, I'd guess there are around 25,000 "active" accounts, because increasing steps of commitment seem to cut out about 90% of people. That is, one in ten readers has an account, one in ten people with an account has voted on a story, and so on.

So basically, chances are good that ten percent of users are selecting all of the site's content. And also bear in mind that that ten percent is probably more like one percent of regular readers, counting people without accounts. But you have to figure that if they weren't doing a good job, they'd be 100% of the site's users, right?

____
Not the real rusty
[ Parent ]

Figuring (none / 0) (#43)
by NFW on Tue Jan 28, 2003 at 02:47:17 AM EST

But you have to figure that if they weren't doing a good job, they'd be 100% of the site's users, right?

Yep, exactly. Readers are here only because they like the site.


--
Got birds?


[ Parent ]

Yes, BUT... (none / 0) (#60)
by sethadam1 on Tue Jan 28, 2003 at 12:15:02 PM EST

I'm not implying the site is bad.  I'm saying the site is GOOD and I'd like to see MORE of it.  The problem is that getting out of the queue is hard if you don't meet what I consider pretty rigid standards.  

I stay because I like K5.  I'd come more frequently if there were more, because I'm not interested in EVERY story.  

You see?

[ Parent ]

I see! I see! (none / 0) (#70)
by NFW on Thu Jan 30, 2003 at 01:35:06 AM EST

I think. You tell me.


--
Got birds?


[ Parent ]

Like democracy in the US (4.00 / 2) (#34)
by webmaestro on Mon Jan 27, 2003 at 11:33:40 PM EST

This is much like democracy in the United States. Not everyone takes the time and effort to vote, but everybody has the right to. The difference is the opportunity to vote and the publicity of what is being voted on. If your going for the democratic approach there really isn't a better way that I can think of.
--
Check out Worldofun.com. It's a world of fun.
[ Parent ]
there are insiders! (none / 0) (#74)
by johwsun on Sat Feb 01, 2003 at 07:49:19 AM EST

..a story of mine voted down with -14 score, and also some angel at my permanent ballot dissapeard without traces. Either god, or rusty, or an insider can reject stories in -14 score.

[ Parent ]
and its NOT pure democracy.. (none / 0) (#75)
by johwsun on Sat Feb 01, 2003 at 07:51:40 AM EST

There is no knowledge tree to be voted by the users.

[ Parent ]
but is close to Democracy.. (none / 0) (#76)
by johwsun on Sat Feb 01, 2003 at 07:53:43 AM EST

..k5 is the closier thing to Democracy appeared ever on earth, after the Athenian Democracy.

[ Parent ]
Timing (4.44 / 9) (#15)
by kpaul on Mon Jan 27, 2003 at 09:09:12 PM EST

I've often wondered if certain stories (slanted either towards the US or 'against' the US) have different chances of making it out of the queue depending on what time they're posted.

For instance, if I knew the story was geared towards a US audience, would posting it during the day in the US give it a greater chance of making it than if I posted in the middle of the night in the US while it was daylight elsewhere in the world.

Thinking on it more, the 36 hour window probably hampers this a lot, but what if a story got voted down quickly when being read by members in UK/US/China before the other half of the world saw it?

In any case, Scoop's the best I've seen yet for being able to control the signal/noise ratio w/out needing to be a dictator or spend hours and hours moderating a site.


2014 Halloween Costumes

Signal/Noise and half-below-average (5.00 / 9) (#16)
by Seth Finkelstein on Mon Jan 27, 2003 at 09:14:21 PM EST

A critical problem is that in any group, half the people are below average. So as the size of the group increases, so does the number of below-average contributions. This then leads to the problem of sorting signal from noise, and it turns out to be far more subtle than one might imagine at first impression. There are deep recursive effects, since as the number of below-average items rises, the sorting system itself also has to deal with more below-average determinations.

Slashdot is architected somewhat like broadcast talk-radio, with a more ad-oriented model, stories selected by managers for controversy or pure audience-pull, and lots and lots of ranting.

K5 is more like public broadcasting system radio, with listener-supported funding (even had a pledge drive!), stories selected for intellectual interest, and a certain elitism in commentary (which I mean in a good way).

Maintaining that elitism, however, requires some fairly harsh gatekeeping. But too much of that, and it become insular, provincial, and self-defeating.

What happens in the future, I won't venture to predict.

-- Seth Finkelstein

And (none / 0) (#54)
by medham on Tue Jan 28, 2003 at 10:50:05 AM EST

Where would you rate yourself, intellectual gatekeeper?

K5 reflects perfectly the ideological constraints of a peculiar brand of American technofetishist posers. As such, it's completely predicatable but of great sociological utility.

The real 'medham' has userid 6831.
[ Parent ]

posers? (none / 0) (#61)
by NFW on Tue Jan 28, 2003 at 02:52:03 PM EST

You mean my fellow K5ers are only pretending to be technofetishists?


--
Got birds?


[ Parent ]

everyone but trhurler (none / 0) (#69)
by dr k on Wed Jan 29, 2003 at 11:10:48 PM EST

He's pretending to be an idiot.


Destroy all trusted users!
[ Parent ]

even worse. (none / 0) (#67)
by pb on Wed Jan 29, 2003 at 01:16:02 AM EST

I have a feeling that the distribution we're dealing with is exponential (i.e., more of the early adopters were already higher on the average, and as time goes by the site acquires more below-average people); therefore, the average has also gone down over time.

Also, the tastes and the guidelines of K5's readership have demonstrably changed over time.  One thing that has not changed since the beginning, however, is the annoying K5 policy of not letting (almost) anyone see a story once it has been "dumped".  I've been against this for a long time and spoken out about it, but the majority in this case does not want anyone to read what they have censored.

Otherwise, I'd point you to three really good stories that have gone through the queue recently and generated great comments (in 36 hours no less); the most recent one had over 300 comments, gobs of moderations, and some interesting discussions.

But that's not nearly as interesting as more K5 meta-wankery, it seems.  Can't get enough of that!  More meta-wankery and less interesting discussion, the majority has spoken!

So, yes, I guess I'm of the opinion that K5 became insular, provincial, and self-defeating, and has been for a while.  Majority rule isn't democratic, and it never has been; it also produces bizarre compromises.
---
"See what the drooling, ravening, flesh-eating hordes^W^W^W^WKuro5hin.org readers have to say."
-- pwhysall
[ Parent ]

Hmmm (none / 0) (#71)
by Ubiq on Thu Jan 30, 2003 at 12:03:49 PM EST

Perhaps what we need is two different minus options, just like we have two different plus options. Something like "Dump it (-1)" and "Don't post (-1)".



[ Parent ]
i kinda like K5 and its articles.. (3.80 / 5) (#18)
by postindustrialist on Mon Jan 27, 2003 at 09:24:20 PM EST

but i agree with you, if it doesn'tfall into thos three categories nobody seems to like to discuss it.. personally i feel one of the better solutions is to see that if an article survives its 36 hour period, and DOESN"T get the 20 under to vote it off, then it should go up.. k5 is large enough that a topic doesn't go by without getting at least 100 votes.. not only that, but a slight widening of the gap so that at maybe -35 instead of -20 would alow it to stay.. a bit more variety IS needed here, but don't complain about the system itself.. K% is purely democratic and i love it for that, but it sets limitations that are unable to handle its huge number of voters. if within thew first 25 people who see your article (and this is a rather reasonable number for your first hour up,) only 4 people like your article and the rest vote against it, you're fucked.. even if down the road 100 people would have voted for it..
oooh.. looks likes somebody has anger problems.
question everything.
this sig is only one hundred and fifty characters long and it's still not eno
Other sites (2.00 / 2) (#19)
by CaptainSuperBoy on Mon Jan 27, 2003 at 09:40:51 PM EST

I don't read OS News or Teh Registar, and I only have one purpose at the other site. It exists for my amusement.

--
jimmysquid.com - I take pictures.
Try scaling up your example - (4.93 / 15) (#20)
by gbroiles on Mon Jan 27, 2003 at 09:46:57 PM EST

You said you'd rather see 5 uninteresting articles rather than just one interesting one - but your article is about scaling. What if you add an order of magnitude to those numbers? Now you're choosing between 50 uninteresting articles and 10 good ones - is your preference the same? What if you go to 500 versus 100?

It's no big deal to ignore a few uninteresting things - but when the sheer volume of them makes finding the good stuff difficult, the entire enterprise may be abandoned.

One way that communities scale is by splitting up into small ones - maybe the average K5 voter didn't like your article, but that doesn't mean nobody does. If your response to that experience is to go find a more receptive community, maybe that's not a failure .. maybe that's K5 tending towards maintaining a manageable scale (by not trying to be everything to everyone), and an opportunity for another community (more sympatico with your sense of humor/interest) to grow.

I don't want there to be one big Borg of "community" where every topic and viewpoint is equally likely to be represented - that makes it difficult for me to choose when and how I participate in different communities. If I feel like reading (or writing) some K5, here it is. If I want something else, those communities are there, too (or waiting to be grown & maintained.)



I like the lower number of stories per day (4.62 / 8) (#22)
by michaelp on Mon Jan 27, 2003 at 10:10:02 PM EST

as it gives more time to read, reflect, and comment.

Things seem to come and go too fast on /, some interesting discussions were there when I first visited, but now I seem to hardly be done reading a story before it's way down the page and on the way to gone.

At a news site I'm working on, we hacked a 'story priority' engine into PostNuke (a great OSS news warez, btw)'s News mod, so editors can keep a story 'on top' as long as they want.

Perhaps something like that could work on k5, where stories could rise to (or remain on) the top of the front or topic page based on the discussion activity they are generating rather than their date?


"Every gun that is made, every warship launched, every rocket fired, signifies in the final sense a theft from those who hunger and are not fed, those who are cold and are not clothed."

I, too, like it slow (5.00 / 1) (#29)
by Idioteque on Mon Jan 27, 2003 at 11:08:47 PM EST

At least for me, k5 is not the site I quickly check when I have a free minute or want the cutting edge news. It's the site I can sit on the couch and read with my laptop in the evening. My homepage is the Everything section, and I like the fact that if I miss a day or so, I can easily catch up. I don't think I would like the site if it moved too fast. To me, it wouldn't feel right.


I have seen too much; I haven't seen enough - Radiohead
[ Parent ]
I'm with you (5.00 / 1) (#63)
by spottedkangaroo on Tue Jan 28, 2003 at 03:57:22 PM EST

Sometimes the stress of checking the other site often enough to catch every story gets to me ... to the point I quit reading it.

I actually like k5 a LOT more lately because what I _do_ read is worth it. I like to take the time to actually read many of these stories and/or click through the MLPs.

Even when I do read the other site (which is still farily regular), I usually just flip through the blurbs and ignore the rest.

[ Parent ]

Obviously not. (3.00 / 1) (#23)
by qpt on Mon Jan 27, 2003 at 10:10:15 PM EST

Sometimes I just want to go where everybody knows my name, and what good is a large site for that?

Domine Deus, creator coeli et terrae respice humilitatem nostram.

Who the hell are you? (5.00 / 1) (#26)
by NFW on Mon Jan 27, 2003 at 10:59:50 PM EST

K5 is bigger than you think.


--
Got birds?


[ Parent ]

"NFW!" (5.00 / 1) (#27)
by kpaul on Mon Jan 27, 2003 at 11:06:10 PM EST

The other K5 patrons yelled as he lumbered onto a thread...

Oh yeah, K5 fiction. It's coming. ;)


2014 Halloween Costumes
[ Parent ]

I am qpt. (3.00 / 2) (#30)
by qpt on Mon Jan 27, 2003 at 11:11:48 PM EST

That you would pretend not to know who I am is slightly amusing.

Domine Deus, creator coeli et terrae respice humilitatem nostram.
[ Parent ]

Slightly, yes... (none / 0) (#39)
by NFW on Tue Jan 28, 2003 at 02:02:08 AM EST

Almost half as amusing as your belief in your own fame.


--
Got birds?


[ Parent ]

My fame? (5.00 / 1) (#41)
by qpt on Tue Jan 28, 2003 at 02:09:35 AM EST

For heaven's sake, man, my name is attached to every single one of my posts, including the one to which you originally replied.

Relying on word of mouth, Entertainment Weekly, or what not would be absurd, considering.

Domine Deus, creator coeli et terrae respice humilitatem nostram.
[ Parent ]

hey norm! (none / 0) (#28)
by eudas on Mon Jan 27, 2003 at 11:06:59 PM EST

maybe www.cheers.com would be what you want.

heh.

eudas
"Nothing is on fire, but the day is still young" -- Phil the Canuck
[ Parent ]

!WARNING - pop-psychology below! (4.00 / 1) (#36)
by lurker4hire on Tue Jan 28, 2003 at 12:49:09 AM EST

First off, interesting discussion material so +1 section (just to piss you off :) ...

However, you complain (and complain it is, regardless of your protestations otherwise) that articles get self censored to reflect pre-conceived notions of what is appropriate for k5 but fail to clearly state what exactly those categories are.

Here's a thought, I have been lurking on this site for quite awhile now... and go through various phases of just reading, commenting, rating, and abandonment... for many of the reasons you mention in your piece.

However, for me I get sick of k5 because I feel it leans too heavily toward right-wing neo-liberal, mainstream consumer techno-elite (while still hanging onto some sort of notion of a downtrodden existence) point of view. I suspect that there is a sizable number of people who feel the exact opposite regarding the content posted (ie: they see too much tree-hugging femi-nazi commie bullshit).

My point is, to engage in the pop-psychology you brought into this discussion, is that people tend to notice what irks them... and silently nod along with what they agree with. k5 is a community, and as such tolerates (with varying degrees of success) a variety of opinions.

Personally, I think the k5 system is pretty good. Of course it suffers from ballot stuffing duplicate accounts, but I tend to think it accurately reflects what the people who visit it find interesting and discussion worthy. If that happens to be a lot of politics, it's only because politics affects everybody, so everybody has an opinion.

Word. (none / 0) (#79)
by dikaiopolis on Sat Feb 08, 2003 at 02:59:39 PM EST

From the point of view of a tree-hugging femi-nazi comminis' though, I have to say that it's the vocal people that shape what political view a place has, so rather than abandoning it, write about it as I'm sure you do. I really like your analysis of the hypocracy of trying to maintain the downtrodden feel while rusty's off selling adds and memberships. It obviously aint downtrodden no mo'.
gnoske seauton
[ Parent ]
On evidence, I'd say yes (4.00 / 3) (#40)
by Delirium on Tue Jan 28, 2003 at 02:09:26 AM EST

I've been here for maybe two years or so, and it's definitely a much larger site than it used to be. It hasn't scaled perfectly, and it's definitely changed, but while it's no longer a good small community, it's still a pretty decent medium-sized community. I think it's likely to stay somewhere in that range for the forseeable future. Could it scale to a large community? Perhaps not; I'm not sure any large communities really work well at all. But it seems like it should be able to keep doing what it's doing.

On the topic of stories, on the other hand, I'd actually like to see more stringent standards. Don't come up with an idea for a story, write a first draft in 30 minutes off the top of your head, and post it. Plan it out so it's coherent and readable, do a bit of research, and maybe proofread and revise once or twice. After all, as we're scaling, we can afford to be more picky. If before we had to take the top 50% of stories to get a decent number, now we can afford to take maybe only the top 20% of stories. Not that we can't take more if they're all good, but we don't have to if they're not, 'cause we have enough good ones.

Story quality (4.50 / 2) (#59)
by rusty on Tue Jan 28, 2003 at 12:13:07 PM EST

It's been my observation that overall story quality has improved with the community growing larger, and I think it's for exactly the reasons you describe. With 50 stories to choose from in a day, the 4 or 5 best are likely to be better than if we had 10 stories to choose from. It's interesting that the number of stories posted per day hasn't really changed in a very long time, while the number of stories submitted has kept increasing. It seems like the voters have generally converged to a sense of how fast things should cycle, and seem to be maintaining that by being increasingly selective.

____
Not the real rusty
[ Parent ]
-1, dosn't understand statistics. (3.42 / 7) (#44)
by delmoi on Tue Jan 28, 2003 at 03:05:19 AM EST

You may or may not belive it, but according to modern statistics, the margin of error in a statistical survey is only related to the sample size, not the whole population. So, a survey of 640 people will only have a margin of error of around 5% wether we are talking about the population of kuro5hin or the whole world.

Of course, queue voters are not a truly random sampling (maybe for a larger site it should be?), but in general it's the 'regulars' who decide, and who better? Why should someone who only visits once a week get to chose what apears on the front page?
--
"'argumentation' is not a word, idiot." -- thelizman
That's bullshit (5.00 / 1) (#52)
by twistedfirestarter on Tue Jan 28, 2003 at 09:41:56 AM EST

A survey of 640 people out of 640 people will be completely accurate. Idiot.

[ Parent ]
Exactly (none / 0) (#57)
by vrt3 on Tue Jan 28, 2003 at 11:51:57 AM EST

And a survey of 640 people out of 300 people will be complete bullshit.
When a man wants to murder a tiger, it's called sport; when the tiger wants to murder him it's called ferocity. -- George Bernard Shaw
[ Parent ]
OK idiots and twits, (none / 0) (#78)
by dikaiopolis on Sat Feb 08, 2003 at 02:52:46 PM EST

1) Stop the name-calling it's obnoxious, especially since his point is valid, and you are fools for not seeing it.
2) Delmoi's statement is in fact, accurate, for, as a physicist (I'm a physicist) would say "Values greatly less than 100%". Indeed, accuracy grows linearly with sampling rate for small values, in fact, for values up until close to 100% accuracy. This is because the number of samples required to get a certain accuracy of measurement is determined only by the amount of data contained in each sample, and not by the number of total samples. Consider the fact that you can look at a couple clocks and get a pretty good idea what time it is, and after that, looking at more clocks doesn't help too much. Furthermore, the amount of data in a K5 voting sample is finite and describable. It is an integer from 1 to 4: +1, +1fp, -1, and A. This indicates that the level at which the accuracy to sample size becomes nonlinear is very low, ie you only have to take a few measurements to have near-100% accuracy. The curve is like this: 10 people, 30% accuracy, 20 people, 59% accuracy, 30 -> 74%, 40 -> 86%, 50 -> 90%, 100 -> 95%, 1000 -> 99%, 10000 -> 99.9%, et cetera. For K5, I think 600 people even is plenty.
gnoske seauton
[ Parent ]
Hm (none / 0) (#80)
by vrt3 on Tue Feb 11, 2003 at 09:44:41 AM EST

Are you saying a survey of 9 samples out of a population of 10 is only as accurate as a survey of 9 samples out of a population of 200 million?

Also, remember that accuracy is not the same thing as precision.
When a man wants to murder a tiger, it's called sport; when the tiger wants to murder him it's called ferocity. -- George Bernard Shaw
[ Parent ]

No, I'm not saying that... (none / 0) (#81)
by dikaiopolis on Wed Feb 12, 2003 at 12:58:08 PM EST

I'm saying that the accuracy to number of samples thing is a non-linear function of the number of total samples and the precision necessary to encode each sample, and making the further point that for LARGE sample sets, which 10 is not, the effects of the number of total samples can be neglected, since its effects are minimal at small percentages.
gnoske seauton
[ Parent ]
twit (3.75 / 4) (#55)
by hypno on Tue Jan 28, 2003 at 10:55:13 AM EST

You voted to dump the story, and then you go on to discuss it. If it's worth you expounding your opinions about it, isn't it good enough to go on K5?

Perhaps you are the sort of person the article refers to?

[ Parent ]

Uh.. (5.00 / 1) (#56)
by awgsilyari on Tue Jan 28, 2003 at 11:29:58 AM EST

You may or may not belive it, but according to modern statistics, the margin of error in a statistical survey is only related to the sample size, not the whole population. So, a survey of 640 people will only have a margin of error of around 5% wether we are talking about the population of kuro5hin or the whole world.

If the sample is randomly selected, you're right. However this sample is obviously not randomly selected. People choose whether to vote. You can't draw any statistical conclusion about the whole population from this at all.

Suppose that, of the people who would have voted A, 20% choose to vote.
Also suppose, of the people who would have voted B, 10% choose to vote.

Now, suppose that the number of people who would have voted A is equal to the number that would have voted B. A is probably going to win, because B individuals are less likely to vote -- even though the number of A individuals and B individuals is the same. The point is, there could be a correlation between the whether a person votes and how they vote. This correlation can't be exposed by the self-selected sample. So you can't make any meaningful conclusion at all.

--------
Please direct SPAM to john@neuralnw.com
[ Parent ]

Yeah, so what. (none / 0) (#77)
by dikaiopolis on Sat Feb 08, 2003 at 02:13:42 PM EST

Yes, that is true, but I don't see where it has anything to do with delmoi's point, which I think is quite valid. Why slow things down by making 5% of the readership read something before posting it when it doesn't decrease your error magin significantly from making 600 people read it.


gnoske seauton
[ Parent ]

Use the diary section (4.00 / 1) (#46)
by TheophileEscargot on Tue Jan 28, 2003 at 04:47:48 AM EST

We already have what you're looking for, the diary section. It's faster moving than the main page, but still gets quite a few comments.

Examples: my latest article got 30 comments in the story section, 17 in its diary draft; the one before that got 89 comments on the front page, and 38 in its diary draft. OK, you only get about half as many comments, but that's pretty inevitable if you speed up the story flow.

Of course, I suspect that what you really want is something fine-tuned to exactly the precise speed, quantity and quality that your personal surfing time allows. Good luck with that.

There's a big internet out there, with room for a lot of little niches. If you really think there's a demand for a new niche, you could always try building it yourself; but moving K5 out of its current niche doesn't seem to make much sense.
----
Support the nascent Mad Open Science movement... when we talk about "hundreds of eyeballs," we really mean it. Lagged2Death

Slightly misleading (none / 0) (#49)
by Scrymarch on Tue Jan 28, 2003 at 07:40:44 AM EST

Since your diary has a lot of subscribers; it's a mini-channel in itself.  That's why it routinely gets more posts than RFT.  But an attractive one-off diary will still bring in the punters, no matter who writes it.  A cool link or a controversial and half-decently argued view definitely helps.  

[ Parent ]
Not really misleading (none / 0) (#50)
by TheophileEscargot on Tue Jan 28, 2003 at 07:49:38 AM EST

If you look at diaries like this or this, people can get loads of comments without being regular diaries. There aren't that many subscribers around, and most of the traffic is in the diary section these days.
----
Support the nascent Mad Open Science movement... when we talk about "hundreds of eyeballs," we really mean it. Lagged2Death
[ Parent ]
Subscribers (none / 0) (#51)
by Scrymarch on Tue Jan 28, 2003 at 09:14:52 AM EST

The experience pre-subscription suggests that people will watch out for diaries whether there's a technical mechanism for it or not.  They expect "Brand Username" to be entertaining.  And I did mention that one-offs can get a lot of comments ...

[ Parent ]
a lone(?) gun speaks (4.00 / 1) (#48)
by three-pipe on Tue Jan 28, 2003 at 07:26:51 AM EST

alright, i know this is not the heart of your article, but i think its an interesting discussion point nonetheless...
In essence, I could hypothesize that people watch the way K5 works, and then begin to "predict" how others would vote and vote with the crowd when they don't have emphatic opinions, since all voting is public.
yes, public voting is great...if you bother to look. but how is a "k5 identity" supposed to emerge naturally if, when voting, we go with the flow? i, for one, (pardon the self-righteousness) tend to just read the story before i vote. perhaps it helps that i'm already on a site on which most of the users have similar viewpoints to my own... but all the same. cellular life emerged as a set of self-assembling, self-replicating chemical systems... can't culture do the same? consider the file-sharing movement.

\\this comment cut short by a 9:00AM class.


-chad \\ warfordium.org \\
Some community building ideas (1.00 / 1) (#62)
by snowmoon on Tue Jan 28, 2003 at 03:04:28 PM EST

One I think the site needs some way of democratically punnishing users that are abusive of the system.  Some way of taking the down a notch so they are easier to control.  Penalties could include no moderation rights, no story poiting right, and or no diary posting rights.  I think that posting comments should still be allowed, the penalties would fade over time.  

New users should have some "probation period" to prevent the most atrocious ballot stuffing problems.

Some of my less well though out idea include...

Drop the voting for FP.  Just promote existing stories based on voting and feedback.  So you could change the voting to "yeah, yawn, and boo".

Possibly some sort of PKI web of trust so it's easy to spot new and "untrusted" users in a crowd.

Just my .02USD before inflation.

Hmm (5.00 / 3) (#64)
by cafeman on Tue Jan 28, 2003 at 05:26:47 PM EST

Semi deja-vu.

The thing I realised after I wrote that was that I think K5 has an (intentional? unintentional? a case of emergence? who knows) inbuilt story limiter. The relationship between community population and story submissions is far from linear - most people don't have the patience or willingness to submit multiple articles that get voted down (as I think people are more likely to vote a story down as the number of stories in the queue increases). So, most give up after one or two attempts.

I don't think K5 stories will scale - I think there's an upper limit on the number of stories being posted. My guess is that it follows this chain:

Site population increases. More people means more story submissions. As the number of submissions increases, the proportion of rejected stories also increases. People get discouraged and stop submitting. Things balance out with some slight (but highly disproportionate) growth in the number of stories posted.

Personally, I think the upper limit of simultaneous submissions seems to be around 12 at the same time, maybe as high as 15. I don't think it'll go much higher than that under the current setup. I think there are inbuilt dampeners which discourage people from submitting stories beyond that. As the number of submissions increases, the number of posted stoies seems to remain pretty close to constant (or show marginal growth). I don't we'll ever see 20 articles being posted on the same day - instead, we'll see the quality of stories increasing, but at the same quantity as always.

The number of comments per story, on the other hand, will probably continue to grow. Minimal effort, immediate returns, and the chance to smear your opinions on the walls - great fun! Ratings will become more and more important though, as on a popular story there's getting to be quite a bit of crap. I think the cliques and factions on K5 also limit population - certain types of stories tend to get posted more than others. As the cliques grow, the site becomes more and more focussed, which in turn limits general population growth.


--------------------
"No Silicon heaven? But where would all the calculators go?"


related thought (none / 0) (#65)
by NFW on Tue Jan 28, 2003 at 07:57:08 PM EST

Would it help to handle story voting in the following fashion?

Rather than a race to get 95 +1s before getting 20 -1s, just tally X votes (X ~= 300? 500?) and then base approval on a percentage?

I doubt it would actually change the results much, but there is an element of arbitrariness in race-based nature the current system. (That's race as in speed, not heritable traits, thankyouverymuch). Then again, perhaps this is just exchanging one element of artibrariness for another. But at least this proposal guarantees a larger group of voters and thus a (theoretically) more representative election.

It guess it would slow things down, especially in the case of rejections, which would lead to a new variation of queue abuse. But it might assuage this author's concern, at least a bit.

Or would it? You tell me, Sethadam...


--
Got birds?


Interesting... (none / 0) (#72)
by sethadam1 on Thu Jan 30, 2003 at 11:30:55 PM EST

Yes, I definitely think that entertaining other options is always good.  

The current method is not flawed, but a percentage might be interesting.  Of course, there would have to be a time limit in that case, and nothing could leave the queue quickly - everything would be in the queue for the full amount.  Another thought would be:

Post reaches 95: post.
Post reaches -20: dump
After 36 hours, if no answer, if there are more yeses than abstains or dumps, post it.  

I wonder how many articles would qualify.  

[ Parent ]

good conversations that suffer from poor articles (none / 0) (#66)
by seanw on Wed Jan 29, 2003 at 12:47:24 AM EST

the only thing that bothers me about the Scoop queue right now is that interesting conversations can be cut short if a story goes under.  especially the ones that languish for 36 hours and then are dumped--they can really take a lot of interesting comments with them.  even the obvious losers that race down to -20 can be really entertaining.

I think that if you have voted a story up or commented on it, even if it gets dumped off the queue you (and the other people who've commented/voted/indicated their desire) should still have access to it, as though it were on your Hot List.  The K5 community at large may have voted it down, so there's no reason it should be "published" onto the site, but...

Basically, I come here FOR the conversation.  Just becuase an article is crap doesn't mean it can't start a great conversation, and the great conversation shouldn't die with a crap article.

Write in vote (3.00 / 2) (#68)
by S1ack3rThanThou on Wed Jan 29, 2003 at 06:06:42 AM EST

I think we need to raise the number of votes required for a decision on a piece. Perhaps % of regular readership.

"Remember what the dormouse said, feed your head..."
in order for a community to scale... (none / 0) (#73)
by johwsun on Sat Feb 01, 2003 at 07:45:44 AM EST

..you may use user-mode-linux package, and create virtual linux on top of a linux. I think its nice, because you can move you linux file from one host to another.

Can a Community Based Web Site Scale? | 81 comments (71 topical, 10 editorial, 0 hidden)
Display: Sort:

kuro5hin.org

[XML]
All trademarks and copyrights on this page are owned by their respective companies. The Rest 2000 - Present Kuro5hin.org Inc.
See our legalese page for copyright policies. Please also read our Privacy Policy.
Kuro5hin.org is powered by Free Software, including Apache, Perl, and Linux, The Scoop Engine that runs this site is freely available, under the terms of the GPL.
Need some help? Email help@kuro5hin.org.
My heart's the long stairs.

Powered by Scoop create account | help/FAQ | mission | links | search | IRC | YOU choose the stories!