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]
How will K5 avoid being crushed by content?

By cafeman in Meta
Sun Mar 17, 2002 at 04:55:59 PM EST
Tags: Kuro5hin.org (all tags)
Kuro5hin.org

Kuro5hin, in growing as a community, faces challenges caused by this growth. One of these challenges is how it will handle the large influx of story submission as the number of participants increases. We need to plan now, less K5 implode under its own success.


This topic crosses over both the development arm of K5 and K5 in the main. However, I believe it should be posted here, as discussion from the users of K5 is core to long-term growth.

The current system works. At the very least, it seems to work. In many cases, this is good enough. Stories that have little content or exploit flaws in the Scoop code get dumped quickly. Stories that have potential for conversation are posted. However, the current model is not scalable.

Kuro5hin currently has population of around 25,000 (plus or minus 5,000). This is a far cry from the supposed 600,000 on slashdot.org. The question is how K5 can avoid the problems slashdot is experiencing. From K5's perspective, one of the most relevant problems is its ability to handle large numbers of submissions.

Slashdot uses the model of a benevolent dictator. All articles must be approved by an editor prior to posting. K5 uses a democratic model, where articles supported by the community get posted. Each has their own advantages. As slashdot becomes bigger, just hire more editors. K5 offers more user participation. While (in my opinion) the approach adopted by K5 offers many more advantages over the slashdot approach (such as clearly representing community interest), it doesn't scale well.

What will happen in the future as the population on K5 doubles? Or triples? When we move from 25k people to 75k people? Over the past year, I've seen the average number of stories in the queue increase from around 2 to 6. The entire section area on the frontpage cycles on a daily basis. If our community population increases four times over, we'll start seeing 24 stories in the queue on a regular basis. If the cyclical rate increases, the chance of decent responses to an individual article decreases. If people don't respond to articles, it becomes less likely that people will submit articles, as there is little reward if no-one see the article. Rather than wait for this to happen, we should start planning a migration path now.

Fundamentally, I believe we want a large community. A larger community means more conversation, it means more articles, and it means we all learn or experience more. So how can we continue to grow without imploding under our own weight?

To see what the future holds, here's a simple scenario. Through slashdot migration and word of mouth, suppose we increase in size from 25k to 200k over the next 2 years (a not unreasonable prediction, I believe). This means we move from 6 article in the queue to 48. Who has time to look through all 48? How accessible are 48 articles for a newcomer? Alternatively, look at it the other way - what happens if all 48 articles are voted on by the majority of regular users? The sections would run of control - articles will drop from the frontpage of K5 in hours. People will never get a chance to comment on them. Either outcome is undesirable.

So what do we do? One option is to further modularise K5. Create multiple, personal homepages, based on individual preferences for sections. Think a collection of tickboxes identifying your topics of interest. Create the equivalent of slashboxes, one for each section. People can then choose what they want to see, based on their interests. Front page articles are still seen by everyone, as if the community deems them important enough to go to the front page, it's highly likely everyone should know about them. The submission rating queue would be tailored to correspond to your section preferences. All submitted articles would be shown, but your nominated sections are clearly separated from the bulk of submissions. These changes no not need to be implemented now, but they need to be planned for. We need to work out where we want to go now, rather than try to respond (and likely fail) in two years.

I'd hate to see K5 collapse under it's own weight. It's important that Rusty et al start developing a strategic plan for K5, rather than provide tactical responses. Financial concerns are core (witness text ads), but K5 needs to continue to grow if it's to remain viable. I, for one, want to see it survive. How can we solve this?

Sponsors

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

Login

Related Links
o Slashdot
o Scoop
o Kuro5hin
o Also by cafeman


Display: Sort:
How will K5 avoid being crushed by content? | 129 comments (114 topical, 15 editorial, 0 hidden)
Hello (3.77 / 9) (#3)
by Should be a Diary Fascist on Sat Mar 16, 2002 at 05:18:37 AM EST

Hopefully the solution will be additional people voting down worthless stories. That is all.

Seriously, I believe Rusty increases the posting threshold as the number of users increase. So as the number of stories submitted increases, there will be relatively fewer accepted and therefore there will be an increase in quality.

--
Hello, I am the Should be a Diary Fascist. That is all.

There's still a problem (4.00 / 3) (#4)
by cafeman on Sat Mar 16, 2002 at 05:22:50 AM EST

What you've said is valid for the number of stories actually posted, but not for the number of stories submitted. The queue can still blow out, regardless of what the thresholds are - that's the problem I'm trying to identify and resolve. If the queues blow out, less people will want to wade through large numbers of articles to determine what should be posted. When that happens, out submission queue becomes huge while the front page dies. Further down the track, K5 starts to suffer, as the community stops growing and less articles get posted.



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


[ Parent ]
Further down the track (none / 0) (#6)
by Kwil on Sat Mar 16, 2002 at 05:26:14 AM EST

Doesn't this mean then that it just balances out? It might be interesting to see what the "natural" user level is.

[ Parent ]
Hello (1.75 / 4) (#9)
by Should be a Diary Fascist on Sat Mar 16, 2002 at 05:32:23 AM EST

This should solve the problem of a large queue.

--
Hello, I am the Should be a Diary Fascist. That is all.
[ Parent ]
Scripts (none / 0) (#120)
by Mitheral on Wed Mar 20, 2002 at 05:46:48 PM EST

I don't believe automated scripts are the way to deal with ANY preceived problems with voting or ratings.

[ Parent ]
Threshold (4.33 / 3) (#5)
by Kwil on Sat Mar 16, 2002 at 05:24:30 AM EST

I seem to remember reading a Meta story where Rusty stated that the threshold is actually automatically adjusted to 6% of the total users.

Of course, the other solution to this is we as kuro5hin moderators become more ruthless in moderation. I think this will probably happen just as a reaction to the increased queue.


[ Parent ]
Ruthlessness discourages growth (none / 0) (#8)
by cafeman on Sat Mar 16, 2002 at 05:29:55 AM EST

If you act ruthless under the current system, good articles may get dumped, simply becuase the personal threshold of what a good article is becomes too high. If everything good got posted as the community population increases, we'd end up with a frontpage that changes hourly. Wouldn't it be better to start building the infrastructure now to support more modularity? That way we could encourage more good articles, while still maintaining a high signal to noise ratio.



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


[ Parent ]
increasing threshold (4.50 / 4) (#16)
by dr k on Sat Mar 16, 2002 at 06:05:03 AM EST

There used to be a population proportional threshold, but it led to stories sitting in the queue for weeks, because as new people signed on, this threshold would get higher and higher. So rusty would lower the proportions, and the whole thing would cycle over. Now the threshold is fixed. It is, as I like to point out every now and then, still not the mathematically correct thing to do, but it works.

On a social level, people get unhappy if they don't get a chance to vote on every story. So if a "bad" story jumps onto the front page they feel they have the right to complain about it. But this is meaningless, if it takes less time for a story to get voted up/down, you still have to assume that the vote is valid.


Destroy all trusted users!
[ Parent ]

let the people decide about thresholds! (2.00 / 1) (#121)
by johwsun on Thu Mar 21, 2002 at 08:11:42 AM EST

I told you this a million times, you do not listen to me, you are going to crush!

sudently i feel tired...

[ Parent ]
Not overwhelming content, but better content! (3.83 / 6) (#13)
by xriso on Sat Mar 16, 2002 at 05:46:19 AM EST

The users will know to not post substandard stories. Basically, there will be a quality threshhold that applies for the whole community. The more people interested in writing stories, the higher quality we demand. (since we only want to post a certian number of stories per day). Look at one of the old K5 stories. I don't think very many would make it through as many users as we have today.
--
*** Quits: xriso:#kuro5hin (Forever)
exactly ! (4.00 / 1) (#21)
by fhotg on Sat Mar 16, 2002 at 07:37:49 AM EST

The good side of the growth should be directed towards better quality stories, while keeping the posting frequency similar to now (possible for individuals to keep at least an overview over the whole queue)

Increased supply (of stories) AND static demand (of stories posted) -> better stories posted.

The posting frequency could be held constant by changing the post-criteria. For example an automatic dynamic settig of the post/dump threshold depending on the number of stories submitted .... or ordering stories which reached post threshold and only posting when it's time again ...

The exact policy would have to be very carefully designed and discussed, and starting early to experiment wouldn't be mistake either.

[ Parent ]

Rubbish (4.50 / 4) (#90)
by codemonkey_uk on Mon Mar 18, 2002 at 07:57:10 AM EST

The standard of story submissions, is in my humble opinion at an all time low.

Of course, you'll be able to dig up some shit stories that got posted ages ago, but I challenge you to find a recent tech story as good as one of my, or Carnage4Life's articles.

The problem is, as I recently pointed out in a diary that the level of dicussion has gone to pot, and its just not worth the effort required to write a good article, if the feeback is crappy.
---
Thad
"The most savage controversies are those about matters as to which there is no good evidence either way." - Bertrand Russell
[ Parent ]

Non-problem (4.14 / 7) (#14)
by Skwirl on Sat Mar 16, 2002 at 05:49:32 AM EST

If people don't respond to articles, it becomes less likely that people will submit articles
Won't this solve your overpopulation problem? If you ask me, K5 is bound to scale better than Slashdot, because, as your userbase grows, your editorbase grows. So what if an individual doesn't vote on every story in the queue? They'll vote on the ones that catch their eye and the ones with bad/vague titles (or summaries if those ever enter the queue) will die the deaths they probably deserve. Now, there is a case to be made for the end of a community feeling on K5 and a worse signal/noise ratio, but I'm sure that by the time that happens there'll be a new place for the 1337 to hang out.

--
"Nothing in the world is more distasteful to a man than to take the path that leads to himself." -- Herman Hesse
The queue will have to be restricted in some way. (3.00 / 4) (#15)
by la princesa on Sat Mar 16, 2002 at 06:00:18 AM EST

Either working in on the fly editing or reducing the number of people who can submit votable articles or both will likely have to happen in order to get anything resembling quality articles voted up regularly if the queue balloons any further.

___
<qpt> Disprove people? <qpt> What happens when you disprove them? Do they disappear in a flash of logic?
I am hesitant to vote this up... (4.55 / 9) (#17)
by Talez on Sat Mar 16, 2002 at 06:15:01 AM EST

You raise an interesting point, yet to be frank, it is a point where most K5ers will say "more users = more editors" and take the attitude of "she'll be right, mate" which is a view I tend to agree with.

We already have "special users" in the form of Trusted Users. I don't know whether you have noticed but if a user posts enough high rated comments, they will be invited to join the ranks of the K5's trusted elite. Whether this trusted user status could be transformed over to a preliminary mod queue before the main queue, I don't know. On one hand, you have a bunch of well known, knowledgable people picking out the best stories to be submitted. On the other hand, we're all a bunch of cynical bastards.

As for the customisable home pages, I've always believed that this is a bad idea simply because it gives each person their own individual view on the world. While this may seem convenient, it does nothing to promote intelligent conversation as people will often only know the side of the story that they want to know. IMHO, a huge pool of conversation moderated by the community would be much better than lots of little pools of conversation each talking about their own thing.

Si in Googlis non est, ergo non est
I agree, because... (+1 needs the discussion) (4.00 / 1) (#22)
by nobby on Sat Mar 16, 2002 at 08:24:20 AM EST

I am new to kuro5hin, I am not ex-slash dot, I am completly new to all of this...

I was involved in the 'dear abby' story quite heavily and I was very clearly discussing the story with a bias toward law and order. It took some heavy comment from other users to get me to see the points of view being presented.

I'm sure that the diverse views presented to me there were because people glancing at my post's saw an arogant point of view that required a little humility. I totally agree with them.

I also voted down an article during that 'red mist' and got slated for it... Absolutly right. If the user base is split into smaller groups then I might not have learnt so quickly to be more open to others views, and a little less cynical about the motives of other posters.

To summarise, I think the broader readership of the articals benefits people like me. I also think that the diversity of the replies educates me more.

So I think that it would be a shame to reduce peoples exposure to others views.
Quis Custodiet Ipsos Custodes?
[ Parent ]
Take a leaf from other democracies (4.22 / 9) (#20)
by SlickMickTrick on Sat Mar 16, 2002 at 06:44:24 AM EST

You compare the K5 article system to a democracy, so why not adapt some of the ideas seen in real democracies?

Not everyone actually has time to sit through and vote on every entry in the scoop (though I do try). What if people could allocate proxy votes. That is, people can say as they don't have time to vote for, that someone else can do it for them. Much like an elected parliment, only without the definitive status. A person who as allocated their proxy may still vote, and thus for that story the value of the proxy is reduced by one.

Now, the idea has a couple of flaws. What stops someone creating a hundred new users and allocating their proxies to themselves? There needs to be checks and balances to a certain extent. You could start by saying a user can be a proxy for no more than 5 people, making it damn hard to screw with the system. It's not perfect, but you could easily abuse the current system by creating enough users and voting on every story.

Perhaps there could be a reputation system of sorts, based off whether a person has had any stories published. It does however, reflect the karma model used on Slashdot, which many would blame for the one sided commentry in the upper ratings on that other site.


proxy votes is fine if... (none / 0) (#122)
by johwsun on Thu Mar 21, 2002 at 08:16:02 AM EST

we keep our right to tune the votes our representative casts on our behalf..

[ Parent ]
I don't see this becoming a problem (4.64 / 17) (#23)
by localroger on Sat Mar 16, 2002 at 08:31:53 AM EST

Let's suppose the population quadruples. Right now the dump and post thresholds are set at -20 and 95. Since they don't seem to need to be scaled linearly with userbase, they might be bumped to, say, -50 and 300.

You will need the higher dump threshold to keep a small clique from killing a good story before it gets a showing before a good fraction of the users; really bad stuff tends to get flushed quickly, and that will still be the case.

Similarly, really good stuff will probably reach the post threshold just as quickly as it does now, with more people to vote toward the higher setpoint.

That leaves the in-between stuff -- some of which is just a bit lame, badly written but interesting, or otherewise marginal; and some of which is really good but controversial. Right now we have a solution for that too, in the form of the timeout. The jury is still out on how that will work, but I tend to think it will keep the queue size down.

The one refinement I think might be needed as the population grows is a slight change to the timeout algorithm. In addition to the maximum time-in-queue, we might need a much shorter minimum to keep a small clique from voting something out of existence before the bulk of the readership has a chance to see it. (I'm thinking a couple of hours.) So far this doesn't seem necessary but since the vote thresholds aren't likely to scale linearly with population, it will become a bigger problem as the population grows toward the 6-figure range.

As far as the FP cycling, much of that can be attributed to a preponderance of MLP's. Now MLP's which stimulate discussion are good, but they aren't as much work as a fully formed article, and we might want to think about putting them in a separate category (enforced, of course, through the queue) which cycles them faster. Maybe we should have "FP articles" and "FP MLP" each with their own separate lifespans. Or, since MLP's are mostly summary anyway, FP MLP's could be displayed as titles only, as the top section articles are now but a bit more prominently. This could even be a user preference. There are numerous ways of handling it on an aesthetic basis, none of which is probably very hard to implement in Scoop.

Meanwhile, I am frankly amazed at how well the system works as implemented -- and I say that having had a few of my own bon mots kicked off under blood-pressure elevating circumstances. No system is perfect, but K5's moderation works amazingly well.

A bigger problem might be S/N ratio in the comments, but that is a whole other kettle of fish which isn't really addressed here.

I can haz blog!

Change the Front Page threshold. (4.33 / 9) (#24)
by haflinger on Sat Mar 16, 2002 at 09:14:48 AM EST

If the Front Page starts to cycle too quickly, increase the percentage of votes required to get Front Page.

That's it. It's just math. Increased numbers is a mathematical problem, and suggests a mathematical solution.

Did people from the future send George Carlin back in time to save rusty and K5? - leviramsey

Percentage based (2.00 / 2) (#30)
by miah on Sat Mar 16, 2002 at 12:28:46 PM EST

You could have it scale based on percentages. One good metric to use would be the number of logged in users moderating that day. If a decent percentage said FP+1 then it goes there.


Religion is not the opiate of the masses. It is the biker grade crystal meth of the masses.
SLAVEWAGE
[ Parent ]
I don't vote on all the stories anyway (4.00 / 5) (#25)
by maroberts on Sat Mar 16, 2002 at 09:21:38 AM EST

You do not have a Sacred Duty to vote on every story in your In-Queue - I only vote on the ones in which I have an interest. The result of this is it doesn't matter if your sumbission queue doubles or even triples in size, you just vote on stories of interest to you.

I get 2-300 emails a day from a subscriber list and there is no way I read all those - just those about subjects that interet me. If I can handle that many emails I can handle twice the number of K5 stories as currently submitted with no problem.
~~~
The greatest trick the Devil pulled was to convince the world he didn't exist -- Verbil Kint, The Usual Suspects
Biggest problem (3.40 / 5) (#26)
by G hoti on Sat Mar 16, 2002 at 10:02:36 AM EST

I can see the biggest problem not as article numbers but comments. Have you ever tried browsing /. at -1

The real question is... (4.52 / 25) (#27)
by DesiredUsername on Sat Mar 16, 2002 at 10:12:30 AM EST

...how will K5 avoid being crushed by meta content? If the newbie flood of "I know how K5 should be run" doesn't end soon, I'm gonna scream.

Play 囲碁
Newbie? (3.00 / 1) (#37)
by cafeman on Sat Mar 16, 2002 at 03:15:27 PM EST

I'm not quite a newbie. I agree that there have been lots of meta articles recently. I found this interesting though, because it's a question I haven't really seen resolved. Does rusty want K5 to stay niche, or does he want K5 to grow? The current system works with niche, but it won't work as K5 grows.

There's no problem with being niche, and the current system will actually discourage growth in the longer run. I just like the idea of a larger K5.



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


[ Parent ]
I Hate To Be Skeptical... (3.50 / 2) (#45)
by Canar on Sat Mar 16, 2002 at 05:16:31 PM EST

..but the question needs to be asked. At present, the concept that the site won't work as K5 grows is pure speculation. It may have logical concepts behind it (I see them), but we have no proof that this will be the case. What we do have proof of is that K5 is evolving to meet a userbase of 25000 from a one-time userbase of only a few thousand. If that sort of scalability is possible, dealing with an increase on an order of magnitude of users, why would it not be feasible that an increase of another order of magnitude will scale equally?

In my opinion, K5 is becoming higher-quality, at least in terms of the discussion, and to be completely honest, this higher standard is making me somewhat insecure in my postings. This is what I think the next problem in scaling may be that a K5 comment cabal forms, formed of only the people who are really clued in, and the rest of us are so cowed by the brilliant arguments of these people that we fear to post. =)

[ Parent ]
You may be right (none / 0) (#46)
by cafeman on Sat Mar 16, 2002 at 05:51:59 PM EST

I think it has become higher quality, but I still think there are some scalability problems. The comment rating system seems to work, no argument there. But I've seen the submission queue go from 1 article a day to 9. If growth continues at the same rate, soon we'll have 18 in the queue. And so on. That's one area that I'm concerned about. Then, assume 25% of stories make it out of the queue. If we've got 18 per day, and the sections area of the front page only holds 7 articles, every story will only sit on the front page for 1.5 days. The number of comments per story will decrease because people just won't see the articles. That's what I'm concerned about - I want to maximise the signal picked up for a particular story. There's a balance between how long a story should should be seen to maximise relevant comments and how long is too long. I'm worried that in the future articles will be seen for too short a time.



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


[ Parent ]
Articles to the front page (none / 0) (#119)
by Mitheral on Wed Mar 20, 2002 at 05:40:06 PM EST

The number of articles to the front page is already a problem for me. I only have access at work. A long weekend in Canada that isn't matched in the States means articles are going to be pushed off the front page before I can read them. I've just come to grips with the fact I can't read everything and try not to let it bother mean. My advice for article writers is to make the head line and first few sentences as compact a summary of your point as possible. If you wait until the fourth paragraph to come to the point I'm not going to read it.

[ Parent ]
Something you might be interested in (4.00 / 1) (#124)
by cafeman on Thu Mar 21, 2002 at 11:13:56 PM EST

For what it's worth, you can customise the number of articles displayed on your front page in your display preferences on the right. That might help you a bit, at least in the short term.

I only just found out you can also choose what news feeds you want. The only problem is that the feeds aren't categorised, making it a little hard to find out what information each site gives.

Once I get my linux box set up again, I'm going to have a look at contributing to scoop to fix that. See how hard it is, more than anything else. If it's easy enough, I'll give it a go and submit it.



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


[ Parent ]
Hasn't been a problem in practice... (4.00 / 2) (#57)
by localroger on Sat Mar 16, 2002 at 09:31:07 PM EST

This is what I think the next problem in scaling may be that a K5 comment cabal forms, formed of only the people who are really clued in, and the rest of us are so cowed by the brilliant arguments of these people that we fear to post. =)

Hmmm, that particular problem doesn't seem to have plagued Slashdot, with their 45 readers and 599,955 trolls, karma whores, and FirstusPostus reloaders.

I can haz blog!
[ Parent ]

Large community bad (4.66 / 6) (#28)
by jasonab on Sat Mar 16, 2002 at 11:30:56 AM EST

Fundamentally, I believe we want a large community. A larger community means more conversation, it means more articles, and it means we all learn or experience more. So how can we continue to grow without imploding under our own weight?
It's been my experience that, when going out to eat with a group, once you get more than six people, the conversation tends to fragment. That is, people will stop being a part of the same conversation, and will start breaking off into smaller ones. Practically speaking, you might as well have just gone out with those people sitting near you, because you'll never really talk to people at the other end of the table.

I think K5's the same way. Once we hit a certain number of people, we'll stop engaging in the same conversations, and break out more into our cliques. That's especially true if the story cycle time becomes what this article claims (and I agree).

I'm not saying we should prevent new accounts from being created. I am saying that K5 has a certain optimum "carrying capacity," and our site evangelism should reflect that.

We already have 25,000 seated (4.00 / 2) (#56)
by localroger on Sat Mar 16, 2002 at 09:27:00 PM EST

It's been my experience that, when going out to eat with a group, once you get more than six people, the conversation tends to fragment.

I think we're just a tad beyond that point even now. Having already scaled almost 5 orders of magnitude beyond your observation, it's hard to see why another scaling of one more order would have to change things all that much.

I can haz blog!
[ Parent ]

personalization = horrible (4.18 / 11) (#29)
by dukethug on Sat Mar 16, 2002 at 11:41:31 AM EST

I hate the idea of a personalized homepage/submission queue/whatever on K5. If I'm just looking at the stories about topics that interest me, written by people I agree with, then we've destroyed the best thing about K5- the possibility for exposure to a broad base of ideas from a DIVERSE set of topics and a DIVERSE number of opinions.

If not for K5, I would (gladly) know next to nothing about female genital mutiliation. And yet, this article was probably one of the best written, most informative pieces I have read here.

I trust the K5 community to help expose me to new ideas beyond the traditional media channels. I would never want personalization to take that away.

Re: personalization = horrible (4.25 / 4) (#35)
by DarkZero on Sat Mar 16, 2002 at 02:13:52 PM EST

You actually want to be exposed to ideas that you don't already agree with? And you want to do the SAME THING TO OTHERS?!?!

Man, posts like this make me wish there was a "6" option for ratings.

[ Parent ]
A large community (4.66 / 3) (#31)
by Mysidia on Sat Mar 16, 2002 at 12:45:29 PM EST

If people don't respond to articles, it becomes less likely that people will submit articles, as there is little reward if no-one see the article.

If it becomes less likely that people will submit less articles, then there will be less articles and the cyclical rate will go down. That's fine, what would be bad is for the "good articles" to fall victim to "the mindless crap"

Also, in the face of more submissions the cyclical rate of the front page could be reduced by raising the FP threshold so that articles posted there last long enough.

A larger community means more conversation, it means more articles, and it means we all learn or experience more.

You get more people in conversations but it also means more noise.. while growth is a natural side-effect of success, beyond a certain point, growth is not an end to itself -- it can reach a point where the growth does more harm than good, the "carrying capacity of the system".

The more article submitters and people participating in discussions, the more likely you are to have not only inadvertent redundancy, but a mess -- a mess being a state of having too many things to sift through, for example if a posting gets 800 comments, I may not be able to participate in a discussion I would have, because it is buried or because the download time over my dialup connection is too long, and I have little patience with such things.

In my experience, beyond a point: the more people you have participating in a general discussion the noisier it gets.



-Mysidia the insane @k5
re: self (none / 0) (#32)
by Mysidia on Sat Mar 16, 2002 at 12:58:00 PM EST

rotfl.. what I said about redundancy

here

Guess it can happen with any number of comments when you have stupid people like me around who sometimes miss reading other comments before posting.



-Mysidia the insane @k5
[ Parent ]
A couple of things. (4.25 / 4) (#33)
by mandria on Sat Mar 16, 2002 at 01:34:22 PM EST

Let's take Slashdot as an example. Over there users are not allowed to vote which stories want to see posted. On the other hand K5 lets us vote the stories that we want to see posted. Even if the number of users grow to, let's just make up a big number, 300 000, the only thing that the management here would have to do is to change the number of required votes in order for the article to get posted, as post #24 points out. After all the more the submissions the more the users to vote them up or down.

The biggest problem I think it would be the comments. Post #26 points out that reading the commnets at Slashdot at -1 threshold it's really a mess. For this situation K5 will have to come up with a scheme to hide low threshold posts and if the reader wants to, can choose the threshold that he/she wants to read at.

Now I am not saying that karma should be introduced. Let the users vote down those comments that are really bad and useless without us the users worry about karma. With that we accomplish to have the really low voted comments hiding, and if someone wants to read them still can, or we can browse the highest voted comments and save sometime for ourselves.

Other than that I think that K5 works great.



More users would require some page changes (4.66 / 3) (#44)
by John Milton on Sat Mar 16, 2002 at 04:50:47 PM EST

I seriously doubt kuro5hin has 25,000 active users just as I doubt slashdot actually has 600,000. However if there should come a time when kuro5hin gets at least 10,000 votes on and individual article, some of the html would have to change. For one thing, rusty would have to provide a link to the voting record instead of placing it on the page. Downloading a list with 10,000 names every time I wanted to look at an article would be slightly tedious. Also, there should be a ratings style viewing preference along with nested, flat, and threaded. Once a page has so many comments, I've found that the browser can't properly render all the dropdown boxes. I'm think this would change if there were links for rating comments instead.

While I'm making a wishlist, there is one feature that I think needs to be implemented in the near future. Diary neighborhoods. I'm not fond of the idea of personal homepages, but the diary section has reached the point where divisions really are necessary. The standard argument about cliques doesn't apply to diaries since they already are cliques.


"When we consider that woman are treated as property, it is degrading to women that we should Treat our children as property to be disposed of as we see fit." -Elizabeth Cady Stanton


[ Parent ]
Diaries (4.00 / 1) (#58)
by MrAcheson on Sat Mar 16, 2002 at 09:36:26 PM EST

Yeah I suggested a while back both here and on Scoop that individual users diaries should be able to be hotlisted like stories or individual diary entries. This would allow you to build your own list of diaries that you visit most. Unfortunately I'm not up to coding the feature so nothing ever came of it. Rusty and Hurstdog thought it was cool though. :)

My current workaround is to simply hotlist the an individual diary entry from each person and use that as a short cut.


These opinions do not represent those of the US Army, DoD, or US Government.


[ Parent ]
Things will fix themselves (4.66 / 9) (#34)
by skim123 on Sat Mar 16, 2002 at 02:13:28 PM EST

You seem to worry that there will be too many stories in the queue and shooting through the front page. You say: "Through slashdot migration and word of mouth, suppose we increase in size from 25k to 200k over the next 2 years (a not unreasonable prediction, I believe). This means we move from 6 article in the queue to 48. Who has time to look through all 48? How accessible are 48 articles for a newcomer? Alternatively, look at it the other way - what happens if all 48 articles are voted on by the majority of regular users? The sections would run of control - articles will drop from the frontpage of K5 in hours. People will never get a chance to comment on them."

However, earlier you say: "If our community population increases four times over, we'll start seeing 24 stories in the queue on a regular basis. If the cyclical rate increases, the chance of decent responses to an individual article decreases. If people don't respond to articles, it becomes less likely that people will submit articles, as there is little reward if no-one see the article."

So, according to your own arguments, if the population gets too big and queue submissions get too high, there will be a negative incentive of posting stories, so fewer stories will get posted. Sounds like a problem that will fix itself.

Money is in some respects like fire; it is a very excellent servant but a terrible master.
PT Barnum


Yes, but there's a catch (4.00 / 3) (#86)
by cafeman on Mon Mar 18, 2002 at 01:50:47 AM EST

The inbuilt limiters will mainly discourage new participants, not the old hands. Think of K5 as becoming an exclusive club with pronounced groupthink. Once a particular viewpoint builds up critical mass, alternative points of view will be more strongly discouraged. I believe we want new people, a share and care sort of environment (with plenty of disagreement, but no trolls - joke accounts are ok). A wider community base also means more varied articles. The problem will fix itself, but with a suboptimal outcome.

What's your take on it?



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


[ Parent ]
Sites and critical mass (4.00 / 3) (#87)
by cafeman on Mon Mar 18, 2002 at 02:02:31 AM EST

I see two stages (if not more) of community growth online. First stage is reaching sustainable mass. That's where K5 is now. Enough stories hit to keep things interesting, so people keep coming back.

The second stage is critical mass, where the site becomes one of "those" sites on the net. Eg, slashdot (and not many others). Word of mouth and a large community base means that these sites end up attracting content like a black hole. I believe K5 has the potential to move that way due to it's links to slashdot and flows between the two. I also believe this is a good thing - K5 fills an area with no other real comparions (in size and quality). It's this stage of growth I think we should plan for because the dynamics are very different.



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


[ Parent ]
Not the size, the basic setup. (4.00 / 2) (#103)
by Lenins Left Testicle on Mon Mar 18, 2002 at 02:48:08 PM EST

K5 will never become /.

The /. control mechanism (moderation) is totally fucked up. Meta mod adds another kludge on top. The editors then play games with the ratings to satisfy their own particular bent.

If a particular comment on K5 is good/bad, anyone can mod it at any time.

Sort of like an armed, peacekeeping populace.

So what if there is more content? Some of us only have time to read once or twice a week. So we miss some stories. It's not the end of the world.

K5 will grow. It will not become broken at larger user populations, just as it is not broken now.

[ Parent ]

Free market of ideas (4.28 / 7) (#38)
by wji on Sat Mar 16, 2002 at 03:17:01 PM EST

Odd to hear this from a committed anti-capitalist, but maybe competition is the solution.

Set things up like this: instead of voting on stories, you rate them. Every cycle (maybe every 0300 EST, or later every 8 hours or something) the highest rated stories in each section get posted, and (let's say) the two highest rated get FPd. The other stories left in the queue can sit there until they time out (4 cycles, maybe). This way you get a nice, consistent flow of stories.

There are of course some downsides. One is that if there is only one, bad story in a section it gets posted... so really you need a few hundred stories in the queue for this to work, or you can introduce some kind of kludge to prevent low-rated stories from getting posted. Another is that if three or four awesome stories get submitted at the same time, there's a chance one or two will get overlooked.

Oh well, this on its own won't work well, but maybe someone will like the ideas.

In conclusion, the Powerpuff Girls are a reactionary, pseudo-feminist enterprise.

Who's (4.00 / 2) (#39)
by medham on Sat Mar 16, 2002 at 03:26:37 PM EST

The "committed anti-capitalist?"

And, I don't know if this has been raised before, but I think a small percentage of ad revenue should go to the authors and prolific commentors. It is they who keep the site "dynamic," as the kids say nowadays; and they thus deserve some of the income.

I anticipate that some of you may object that contribution would have to be considered qualitatively, not quantitatively, for any such idea to work. You might even further claim that determination of "quality" would be an intractable problem. You may think you're smart because of this, but I have one question for you: Where would America be if our media pioneers had felt this way? Like Holland?

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

Oh, wonderful (5.00 / 2) (#40)
by wji on Sat Mar 16, 2002 at 03:34:47 PM EST

As if there aren't enough K5 addicts already. Now they're going to be quitting their jobs and working full-time as a kuro5hin knowitall.

Speaking as a trusted user who's been on the frontpage (well, it was just an MLP) I think the site can survive fine without paying people to have high-rated comments or FPd stories.

In conclusion, the Powerpuff Girls are a reactionary, pseudo-feminist enterprise.
[ Parent ]

Trusted? Front Page? (4.00 / 2) (#49)
by medham on Sat Mar 16, 2002 at 06:52:24 PM EST

I didn't realize I was writing to aristocracy. As one of the elect, wouldn't you have the most to gain from my proposal?

(PS: I have five accounts, and I have "super trusted user" status on all of them. This basically means that I can edit any comments I wish. Every story on the front page was either written by me or benefitted substantially from my changes. So, don't go getting high-and-mighty.)

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

please find another diversion (none / 0) (#50)
by demi on Sat Mar 16, 2002 at 07:09:08 PM EST

you silly little troll, we are trying to have a discussion here.



[ Parent ]

You have to admit (4.00 / 1) (#51)
by medham on Sat Mar 16, 2002 at 07:11:31 PM EST

That you simply do not know that what I wrote above is wrong. You may think that it is, but you cannot offer any evidence to support that hunch. Therefore, I would suggest that you put away your blue pencil and try to contribute something substantive to the discussion at hand, halfling.

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

I want to talk to another one of the spirits... (none / 0) (#52)
by demi on Sat Mar 16, 2002 at 07:23:04 PM EST

knock once if it is still medham, knock twice if it is not.



[ Parent ]

knock knock (5.00 / 1) (#53)
by Lode Runner on Sat Mar 16, 2002 at 07:23:56 PM EST



[ Parent ]
lost spirit (none / 0) (#54)
by demi on Sat Mar 16, 2002 at 07:29:26 PM EST

Why do you haunt this place? Are you restless? Knock once for yes, knock twice for no.



[ Parent ]

That's what you think (4.66 / 3) (#55)
by wji on Sat Mar 16, 2002 at 07:50:19 PM EST

Actually, "super trusted" status was invented by myself as a subtle trap. It's only given to those posters whom I wish to execute. You see, by analyzing the changes an individual makes to other stories, we can obtain a clear profile of that person's psycho-geographical makeup. By running the changes through our Joint Unified Nooscopic enGine (JUNG) we can identify the exact location, down to a few feet, in which a certain set of thoughts were made. In other words, we can identify the exact location of the seat in front of your computer.

Now, based on this information, the kill is trivial. We simply wait for a period of computer activity and fire our satellite-guided missiles at the co-ordinates arrived at previously.

You might think I'm giving too much away here, but in fact the plan has proceeded past the critical point. Nothing can stop me now.

If you were a rational individual, you might grab some posessions and flee your house at this moment, but our profile indicates you will take this as a joke, kurrode for a while, and then go view some pr0n.

Well, at least you'll die happy.

In conclusion, the Powerpuff Girls are a reactionary, pseudo-feminist enterprise.
[ Parent ]

Dear Sir: (2.40 / 5) (#63)
by RobotSlave on Sun Mar 17, 2002 at 12:21:46 AM EST

Since you have "super-duper extra-special hyper-trusted user" status, you will no longer be needing your mere "trusted user" status.

I have, accordingly, given several of your comments a very low rating.

[ Parent ]

It's only a ... (4.50 / 2) (#100)
by dipipanone on Mon Mar 18, 2002 at 02:15:11 PM EST

PS: I have five accounts, and I have "super trusted user" status on all of them. This basically means that I can edit any comments I wish.

Damn. Now I understand what that guy's trying to say when he keeps posting 'it's only a website'.

--
Suck my .sig
[ Parent ]
Like Holland? (none / 0) (#99)
by dipipanone on Mon Mar 18, 2002 at 02:09:57 PM EST

Where would America be if our media pioneers had felt this way? Like Holland?

What, you mean moderate, rational, pragmatic and lacking a tendency to puritanism and fundamentalism?

You should be so lucky.

--
Suck my .sig
[ Parent ]
Use in combination (4.50 / 2) (#78)
by Kwil on Sun Mar 17, 2002 at 06:10:18 PM EST

I think this is a good idea, just needs to be combined with the current system. If something meets a threshold it gets posted immediately. If it doesn't meet a threshold by the cycle end, it gets evaluated with the top one or two being posted.

If something hits the low threshold, it gets dropped immediately. Otherwise it hangs around in the queue.

If you really want to control the queue, set the cycles to be based on queue size as well, so an automated queue evaluation runs a maximum of once every six hours, but will only run when there are more than X posts in the queue.

Anything remaining in the queue after five or six evaluations gets dumped - or alternatively, shifted to the submitter's diary page if there's a current conversation going on (ie, more than one comment posted in the last two cycles.)



[ Parent ]
Some clarifications (4.77 / 9) (#41)
by cafeman on Sat Mar 16, 2002 at 03:40:44 PM EST

I've slept on it, and re-reading my article in the cold light day makes me realise I didn't entirely explain what I meant. The current system works well as long as the population doesn't get too large. We know that, simply because that's the stage we're at. Whether or not the system will work with a greater population is pure speculation.

It all depends on how you define whether the system "works". My feeling is that as the community base grows, the number of submitted articles will increase. However, as others have pointed out, the thresholds will also increase, making it harder for articles to get posted. We end up with a situation where a similar number of articles get posted to the frontpage / sections of K5, but many get rejected. The submission queue becomes very large, as it takes longer for articles to meet a print or dump threshold.

I would argue that this will act as a disincentive for people to submit articles, as it's unlikely that the majority will get posted. We'll end up with high quality postings, but it'll drive the participating population back down to a certain carrying capacity.

If this is where rusty et al see the site going, that's fine. However, think of it like a club with membership. Most of the people on K5 will be the regulars, simply because newbies may give up on the site too quickly. It's disheartening to take the time to write up an article only to see it rejected, especially so if you're new to the site. Remember - the newbies are the ones who bring additional points of view.

My guess is that the current system has a carrying capacity greater than the current numbers. Anything above that, and it has built in limiters to inhibit growth and encourage decay (one submission queue, limited space on the first page, increasing thresholds, etc). Actually, another interesting point there is that an increasing population will see the thresholds increase to a point where it's likely that very few articles will ever make it to the front page. Meanwhile, the sections will revolve hourly. But I digress.

I'd like to see K5 continue to grow. I'd like to see high quality articles in every section, rather than a few high quality articles spread across everything. I'd like to see this for purely selfish reasons - K5 could then become my central site for most news, culture, and technology. To do this, K5 needs more people (to submit articles and to comment on them).

Personalisation of the first page could still include all of the different sections. Make that an option, similar to the way it is now. All submissions could still be displayed in the submission queue, but sort them and separate the ones you've selected as being of interest to you. We all do that heuristically at the moment, why not make it easier? I'd be amazed if rusty hasn't thought long and hard about this already, and whether to keep K5 in a niche or make it bigger (bearing in mind bigger == more eyes == more text ads == more revenue). I'm interested to know what his opinion is. The restaurant / conversation analogy provided by jasonab (#28) is a good one. But, it can be solved by breaking up the restaurant into smaller groups, with one person who yells out over the top of all groups so they can all hear if something important is going on.

Don't get me wrong - things don't need to be changed now. But it's better to plan ahead and figure out what to do now rather than try to respond reactively. I haven't seen a discussion about this so far, and so thought it would be worth having.



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


Would it really be a disincentive? (3.66 / 3) (#73)
by jacoplane on Sun Mar 17, 2002 at 04:54:09 PM EST

I would argue that this will act as a disincentive for people to submit articles, as it's unlikely that the majority will get posted. We'll end up with high quality postings, but it'll drive the participating population back down to a certain carrying capacity.

One could argue that the prestige of getting your story posted on k5 would go up, since your story would be standing amongst the other high quality postings. I believe that as k5 grows the moderation should be better too kind of like the "infinite monkeys" argument that's often used in open-source advocacy. I feel the meter for k5 scalability is the quality of submissions and the ensuing discussions. It will probably be harder to submit a story, but isn't that a price we should be willing to pay if the overall quality goes up?

[ Parent ]

Think it would work slightly different in practice (4.00 / 1) (#76)
by cafeman on Sun Mar 17, 2002 at 06:01:37 PM EST

I think you're right that the prestige would increase. However, I also think MLP stuff would disappear. The quality of writing would probably go up, but I'd also guess there would be a bias for longer articles. This would discourage people from throwing together a two paragraph story with a link.

Some people would argue that this is a good thing. It all depends on what you want from the site. I like a constant stream of varied material, similar to a news site. I also like the thought-provoking commentary. I think as K5 gets bigger, the current system will move towards longer, fewer, better written articles. But as I've said earlier, the active population is limited because of this. More people with a constant number of stories means that the noise to signal ratio for each story will most likely go up. Eventually, you end up with slashdot.

I really like the suggestion made by post #69 and extended by me. I won't repeat it here, read the post. This approach would mean that we encourage longer, well written articles, while also encouraging other areas of interest to develop along with their own content. If all sites are linked via some headline propogation mechanism (not hard if you stick with a single codebase at first), you get all the benefits of both worlds.



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


[ Parent ]
How is loss of MLP a bad thing? (4.00 / 1) (#83)
by Kaki Nix Sain on Sun Mar 17, 2002 at 10:58:48 PM EST

You made some half-assed sounding chain of connections (in your second paragraph) that started with longer, better articles and ended with k5 being like slashdot, but that doesn't make any sense at all. Slashdot is all MLP; if it weren't for the few links that are interesting and the moderately decent comments that float to the surface, /. would be of no value at all. If k5 lost all of the MLP it would be even more dissimilar to /..



[ Parent ]

Sorry, time constrained (4.00 / 2) (#85)
by cafeman on Mon Mar 18, 2002 at 01:36:52 AM EST

Sorry if I wasn't clear. I'm trying to do six other things at the same time. Here's what I was trying to say:

More people coming to the site (and staying) means the total population increases. As the site becomes popular, the proportion of story posters to comments posters will change, with the number of comment posters increasing faster than the number of people posting stories. See slashdot and my reply to Hemos about transaction costs above for the logic. A tighter submission queue means that it's harder to get a story through. The prestige for getting a story through increases, but MLPs fall by the wayside. This is because people start looking for good writing and depth.

Fewer stories making it out of the queue means that the number of comments per story increases (assuming the site continues to grow). That's what I was talking about with slashdot - cultures of trolls sitting around at the -1 equivalent. Huge numbers of comments, most irrelevant, leading to high noise, low signal. That side of slashdot. Not the MLP aspects.

I think loss of MLP at K5 is bad, simply because I'd rather know a little about a link rather than nothing about it. MLP serves a purpose, it's just that it's one small purpose in the gestalt that is K5. Too much MLP is not good. Slashdot offers little value in the story section, some value in the comments. As a split, I'd say 10% / 90%. At the moment, K5 probably goes for around 50% / 50% (bearing in mind that I'm talking about proportional breakup - the total value delivered between in each site may vary). I believe that a tight submission queue combined with community growth will lead Kuro5hin down the 90% / 10% path, almost the opposite of slashdot. The focus will become the articles, not the comments. I'd rather see it stay at 50% / 50%, similar to the current model, but make the base bigger. That's what I was talking about with the distributed scoop suggestion.

It's hard talking about this stuff due to the scope of it. I need more time to fully explain, time I don't have.

I hope that made more sense? Let me know if it didn't.



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


[ Parent ]
We have this already. (4.33 / 3) (#97)
by dram on Mon Mar 18, 2002 at 01:16:11 PM EST

In reality we already have this. There are plenty of scoop sites that are more specialized than K5 is. K5 covers everything under the sun, technology, politics, you name it, K5's got it. However there are other sites, like Ingenuitas.org that just cover one topic (politics in this case).

At the same time these sites do share headlines. Go look under your display preferences and see if you can find the Ingenuitas.org RDF feed. It's there, I submitted it myself. So what you are talking about has already occured. What needs to happen now is a migration from K5 to these other sites that are more specialized.

-dram
[grant.henninger.name]

[ Parent ]

Amazing ... (4.00 / 3) (#106)
by cafeman on Mon Mar 18, 2002 at 09:27:09 PM EST

Damn, rusty codes fast - I no sooner suggested it, and poof! There it was!!

Seriously, thanks for that - been here since the beginning, and I never knew about that particular feature. Guess I should finally have a look through all the different preferences now.

Migration is an interesting situation. The stuff that's in the queue right now about emergence overlaps quite a bit with what I'm thinking about (a coallition of sites propagating news, the biggest news items propagating the furthest), but I don't want to post any links across for fear of oversaturating everyone with this stuff. I wonder how many other people know about the news feeds - it's not something I've come across before. Maybe a taxonomy for news feeds would help people find what they're interested in, thereby helping migration and propagation.



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


[ Parent ]
Yeah,... (4.00 / 2) (#98)
by Kaki Nix Sain on Mon Mar 18, 2002 at 01:50:26 PM EST

... o.k., cafeman, you made a lot more sense in that last post. Further, I agree that an explosion in the number of comments could be big trouble. However, there are ways to filter the noise out and make it disappear; ways much better than the one we sorta use now.

They all come at some cost though, and I imagine that the reason we don't use any of them already is just b/c they haven't yet become necessary. That is, the cost is too much for what would now be gained. When the noise level gets to be too much, the costs will look better and better and something will be implimented, maybe a few somethings, until a combination that is right for k5 is found.

There probably are already enough articles archived in the meta section about how to change the comment rating and displaying to give rusty and the gang lots of choices for what to impliment when that time comes. Hell, maybe it would be cool if someone went through them all in a new article, catagorized them, and layed out the costs and benefits of each. A meta-meta article on comment systems, if you will. That would show us all just how many contengency plans we already have to fall back on, and it might thus calm your fears.



[ Parent ]

Fixed link (sorry, worked when I clicked on it) (none / 0) (#77)
by cafeman on Sun Mar 17, 2002 at 06:10:14 PM EST

Here.



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


[ Parent ]
Silicon Heaven (5.00 / 1) (#109)
by sgp on Mon Mar 18, 2002 at 10:13:25 PM EST

...of course, I know you're lying. Otherwise, what would be the point to it all? ;-)

There are 10 types of people in the world:
Those who understand binary, and those who don't.

[ Parent ]

Comments, not stories, will explode first (4.26 / 15) (#42)
by chipuni on Sat Mar 16, 2002 at 03:48:37 PM EST

I have to agree with G hoti: that comments, not stories, will explode first.

In my opinion, the winnowing of the best comments depends on the number of disinterested people voting on them. In other words... to pick out the best comments, volunteers need to rate as many comments as possible, not just the ones that they agree or disagree with. As the number of comments explodes, fewer people will be willing to do this... leaving the comment rating to those who are most moved (for or against) a comment. Comments will become rated by whether people agree or disagree with the comment, not whether the comment is of high quality.

Perhaps Scoop should give recognition to people who rate frequently? Perhaps the Scoop code should be amended, in a user's page, to show how many ratings a person has given... or, better yet, how many ratings of each score a person has given? (I do not feel that special privileges should be given for rating. Otherwise, people will write 'bots to give random scores to every comment.)

Your comments?
--
Perfection is not reached when nothing more can be added, but only when nothing more can be taken away.
Wisdom for short attention spans.

Oh, the irony... (2.20 / 5) (#59)
by yicky yacky on Sat Mar 16, 2002 at 09:41:41 PM EST

Over the past year, I've seen the average number of stories in the queue increase from around 2 to 6.
And yet the number of quality articles remains the same as ever...curious and strange don't you think?...

As the dialogue of a cheesy blockbuster once ran:

"You are not part of the solution, you are part of the problem. Quit being part of the fucking problem and put the other guy back on..."

Happy Trails...


yicky yacky
**************
'The actual reasonable Britons are correct, you're being a cock.' - Hide The Hamster.
Some K5 History (4.58 / 12) (#60)
by MrAcheson on Sat Mar 16, 2002 at 10:29:18 PM EST

Well I've been here for about a year and a half. I started here right after K5 can back up after it was shitstormed out of existence. You obviously haven't been here very long (nothing wrong with that mind you) since you haven't used some terminology like the "development arm" instead of Scoop.

Anyway, during this time I have seen the following:

1) Stories taking weeks, not days, weeks to post. It sucked. Rusty adjusted the thresholds and now we're back to days in the queue again. Days is not a problem or a bad thing. Provided a topic isn't talked out by the time it reaches the main page, days is preferable to hours because a wider cross section of K5 gets to vote. If things get bad the thresholds can be altered again.

2) Multi-page submission queues. This does suck. However the reason people were really pissed was that the queue was so long because nothing was moving out of it, not because too much new content was continuously coming in. My personal preference is for material of questionable worth to get killed off more and revised/resubmitted rather than always posted because it has "some merit". Make the posting standards tougher.

3) I have seen lots of these "Scoop should work this way" articles, usually by people who haven't been here long. They seem to go in cycles. Mostly of them are people pissed because of some breach in Scoop etiquette they want fixed. (Wah! A story author modded down my post in his story! He shouldn't be able to do that! Wah! Ricky!!!) You have a point though on this one.

Once upon a time when we were having so much trouble with the queue the last time, it was suggested that the queue use a system more like comment ratings and less like the sh***y slashdot +1/-1 mod system we use now.

Basically the current mod system has two flaws. As the number of users increase the thresholds need to change to accommodate this. No good way to automate this has been found since the last automation attempt failed miserably. Rusty currently controls the thresholds manually IIRC. Secondly if a post isn't good enough to be dropped or posted it can linger in the queue.

The comment rating system however is very very good. Posts will quickly converge to an equilibrium value which changes little with time. (Frankly I wish we used a median instead of a mean to prevent ratings being influences by outliers, but thats my personal Scoop nit.) The trusted user spam removal system is also good. Read the FAQ on how it works.

The solution that was widely discussed is to adapt the comments rating system to the submissions queue. Rate stories 1-5, make the posting decisions when the story rating converges. You can set a minimum number of ratings so it doesn't post/drop too soon, and a max so it doesn't linger.

Since then however K5 has had minor hardware problems, one huge hardware failure that left K5 down for over a month, and now Rusty has had to scramble to keep us funded when OSDN pulled out. He's been busy. Plus the current fix seems to be working fine. If at some point it becomes obvious it isn't it will change.


These opinions do not represent those of the US Army, DoD, or US Government.


Don't be too quick to judge (4.75 / 4) (#62)
by cafeman on Sat Mar 16, 2002 at 11:21:15 PM EST

I was also one of the original people around, prior to the submission DOS that shut K5 down the first time. Given that I was here before K5 was DOS'ed (admitedly as a lurker), between the two of us I'd probably win the "old fogie" award. But that's not important for what we're discussing here.

I used "development arm" rather than scoop simply because with all the newcomers, I'm not sure how many people would understand what I meant when I said scoop. Your points are valid. However, the issues K5 faced then are not necessarily the same that K5 will face in the future. Some problems are teething problems, others related to use and size. Like I've said elsewhere, I don't see any problems with K5 at the moment. I'm not presumptuous enough to assume I know how to fix the minor issues that are kicking around at the moment. But, we've seen K5 get a lot bigger in the last year. I'm thinking ahead, trying to figure out where K5 is going, and what future problems may occur.

It's simple really - figure out a solution now before it becomes a problem. If the problem hits, you can solve it immediately with no associated issues. Classic strategy - figure out where you want to be, then work out a plan to get there.

I think I've said it elsewhere, but I'll repeat it here. I don't have a problem with the mod system. I can see some potential issues as K5 becomes bigger, but I'm not dealing with them here. I'm specifically looking at the growth in the submission queue and the artical turnover on the first page. I don't want to see articles hang around for ever in the queue. I know about the 36 hour rule, but if the queue becomes too big, all articles will hang around for 36 hours because people can't be bothered rating articles that don't look interesting to them. Then, you need to either adjust the thresholds again, or change the time period. Both have problems associated with them. I think rusty has come up with a good solution for the current moderation queue. The problem is, what happens when too many articles hit the submission queue at once? We were up to 9 earlier, we're back down to 7 now. I've seen well into double digits, and K5 keeps getting bigger every day.



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


[ Parent ]
Literacy requirements? (3.42 / 7) (#64)
by medham on Sun Mar 17, 2002 at 01:16:28 PM EST

The community could vote on an essay test that would have to be judged before a new account could be formed. Literacy committee duty could be assigned on a revolving basis.

I myself have noticed that all of the > 20500 UIDs have really been dragging the site down lately, and even many of the old-timers could learn some new tricks.

The real 'medham' has userid 6831.

thank you for the revelation [nt] (none / 0) (#71)
by infinitera on Sun Mar 17, 2002 at 03:35:35 PM EST



[ Parent ]
Thanks 20737 (5.00 / 1) (#118)
by Mitheral on Wed Mar 20, 2002 at 04:26:52 PM EST

That just cracked me up. I'm reminded of something my grandfather used to say: Illegal aliens have always been a problem in the USA -- Ask any Indian

[ Parent ]
I had a similar idea (none / 0) (#126)
by r00t on Mon Mar 25, 2002 at 01:38:42 PM EST

I had a similar idea and posted it here. I see two problems with a literacy test and a panel of judges...

a)All member selection will be based on what the judges think.
b)It takes too long.

I think an IQ type test would be a much better idea because it is not susceptible to human judgement and could be marked immediately. Not everyone has superior literary skills and that is not what this site is primarily about. If you read the mission statement for k5. This site is for people who like to think, ideas and logical thought are the most important characteristics to judge for membership exceptance.

Where humans exist, greed and corruption are possible. Why communism fails and why the US is so f*cked up. Lets learn from that and not have a system dependent on peoples decisions, feelings or thoughts. I would rather not have my membership exceptance based on what some bozo judge thinks.


-It's not so much what you have to learn if you accept weird theories, it's what you have to unlearn. - Isaac Asimov
[ Parent ]

The link should be.. (none / 0) (#127)
by r00t on Mon Mar 25, 2002 at 01:45:14 PM EST

here

-It's not so much what you have to learn if you accept weird theories, it's what you have to unlearn. - Isaac Asimov
[ Parent ]

/. k5 serve different communities, diff population (3.42 / 7) (#66)
by cicero on Sun Mar 17, 2002 at 01:34:19 PM EST

Kuro5hin currently has population of around 25,000 (plus or minus 5,000). This is a far cry from the supposed 600,000 on slashdot.org.
just one quick note which I think invalidates the need for this type of disaster planning, at least of a little while.

k5 is not the type of news site that slashdot is: slashdot is abc news, k5 is call in radio. slashdot spoon feeds people what it thinks is newsworthy, k5 is an interactive discussion site.
yes, the k5 user population will grow. yes, this should be planned for as soon as possible. no, that does not mean that the sky is falling

anyway, here's to the first 25k.


--
I am sorry Cisco, for Microsoft has found a new RPC flaw - tonight your e0 shall be stretched wide like goatse.
Adaptive user submission filtering (4.00 / 3) (#67)
by pla on Sun Mar 17, 2002 at 01:48:27 PM EST

Okay, what I see as the big problem - As more people use K5, the number of submissions will grow too large for each user to realistically read and vote on (I would say it has almost reached that point... I usually have perhaps half an hour at a time to review *all* the news I check up on, and voting on up to a dozen new submissions per day ranks low in my list of priorities).

So, my suggestion:

For each user, K5 would "learn" the type of article they find interesting (perhaps judging by what they read, with a lot of weight on what they consider worth posting a comment on.

Although a user should have the *option* of voting on every new submission, have the "Moderate Submissions" page default to only displaying topics the user will likely find interesting enough to read. If each person only has to vote on half-a-dozen articles per week, I would personally consider that an acceptible level.

Some benefit may also come from having the "Moderate Submissions" page appear as the first a user sees after logging in. If I need to rush and just look at news, I might never see something in the queue that I would really like to vote on. If I see that first after logging in, I can quickly skip it and go back to the front page, but I will at least catch the titles of new submissions.

Additionally, to address the problem of articles leaving the front page too soon, I see a simple solution... As much as people tend to slam /. around here, the way it separates front page stories by day seems to work fairly well. Keep the last ten to twenty articles on the front page, and have easily accessible links to the last week's front pages, which will only cover articles from the particular day the user selects.

Alternatively, perhaps decrease the number of lines of an article that appear on the front page, going from a full summary for the most recent five, to three lines for another ten, to only the titles for the entire last week's articles.


Alternativly (none / 0) (#74)
by Weezul on Sun Mar 17, 2002 at 05:03:06 PM EST

You could allow people to vote on articles by section, but not count the +1 FP votes of people who votes from the section, i.e. if tehy only read the section there is no reason to let them vote it to the front page.


"Fascism should more appropriately be called Corporatism because it is a merger of state and corporate power." - Benito Mussolini
[ Parent ]
Submissions on the side of the page. (4.00 / 1) (#88)
by nefertari on Mon Mar 18, 2002 at 04:16:51 AM EST

Some benefit may also come from having the "Moderate Submissions" page appear as the first a user sees after logging in. If I need to rush and just look at news, I might never see something in the queue that I would really like to vote on. If I see that first after logging in, I can quickly skip it and go back to the front page, but I will at least catch the titles of new submissions.

You could have this effect also by putting the titles of the new submissions in one of those boxes. Then you would see them too.

[ Parent ]

newspapers already have solved this problem (3.80 / 5) (#68)
by turmeric on Sun Mar 17, 2002 at 02:10:25 PM EST

look at the new york times or the dallas morning news. 100+ pages of crap spewed out from thousands of reporters from all over the world. They also include stories from AP, Knight Ridder, and other newswire services (basically this is mindless link propagaation but with the actual text of the article pasted that they are linking to). And yet somehow they manage to organize it, put certain things on the front page, etc. So as far as the 'www.kuro5hin.org' front page presentation goes, no problem, just rely on similar principles to the newspapers, separating stories into sections based on area of interest, like sports, business, whatever.

You say people 'wont have time to look at all 48 stories in the queue'. that is fine. They will just look at stories that they think might be interesting based on whatever criteria they want. some will look mostly at the headline before clicking and reading and voting. some will look at the author. etc. they just won't read through every single article every day, and they won't vote on every single article. Nobody could possibly read through every single article that goes into the New York Times or the Wall Street Journal or Barron's every day either. They rely on their sub-editors and other co workers to look through it all.

But how do newspapers choose which sections will exist? I dont know, but I am betting it comes from the top down. K5 perhaps might want to democratize the 'section' arrangement, as this is perhaps one of the last bastions of dictatorship left in the K5/scoop structure. Of course that may take 80 hours to implement.. and there are other things to do.

Modularisation vs. going elsewhere (4.62 / 8) (#69)
by TheophileEscargot on Sun Mar 17, 2002 at 02:38:17 PM EST

I'm a bit baffled as to the point of modularisation. There are plenty of different community sites, quite a few of which use Scoop or Slash. There's also quite a bit of overlap between them.

So, I don't really see the benefit of splitting K5 into modules. If the community is going to split anyway, you might as well split it into different sites entirely. A distributed solution is more robust, for a start!

I don't really get this "I'm leaving site X for site Y" attitude. Posting on K5 doesn't mean you can't still post to slashdot. I get the feeling people still aren't thinking in Internet terms: other sites are only a link away, so why should the new K5 "modules" exist on a single site? The K5 front page will still be available to anyone who's interested.

Incidentally, have I hyped Radio Free Tomorrow, the new Scoop site for SF and Fantasy discussion yet? And Hatori42: distributed literature for the people is back up again, too... ;-)
----
Support the nascent Mad Open Science movement... when we talk about "hundreds of eyeballs," we really mean it. Lagged2Death

Excellent idea! (4.50 / 2) (#70)
by cafeman on Sun Mar 17, 2002 at 02:46:37 PM EST

Create an alliance of scoop sites! Set up an XML news notification system between the lot, and you could browse seamlessly between them. I like the idea ... if you did some personalisation, when you hit the first page of K5, you would not only see the current headlines, you would also see headlines from other related scoop sites (like the current articles box). K5 could become a propogation mechanism in addition to encouraging discussion. Build it into the codebase, and then every additional scoop site that's set up automatically has the functionality to participate. It would take some work (some coding, but more in coordinating), but would be very cool.



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


[ Parent ]
Semantics... (4.00 / 2) (#80)
by DarkZero on Sun Mar 17, 2002 at 06:34:41 PM EST

I don't really get this "I'm leaving site X for site Y" attitude.

Well, the "leaving" part is really just semantics. We only have so much of our time to seriously devote to an online "community" (the user base of a site), so when someone says that they're "leaving site X for site Y", it means that they're devoting the primary amount of their online time to site Y instead of site X. If someone said that they were leaving Slashdot for Kuro5hin, for instance, it would just mean they'll be spending more time browsing and participating in Kuro5hin than in Slashdot.

The real problem is that our LANGUAGE is still grounded in meatspace, not our thinking.



[ Parent ]
K5 is already avoiding being crushed by conetnt. (4.00 / 2) (#75)
by Mysidia on Sun Mar 17, 2002 at 05:52:32 PM EST

[o] Moderate Submissions (0/0)

Wow.. haven't seen that one in a while... empty queue [G]

At this rate K5 might be starved of content rather than crushed by it.. but yeah.. the cycle speed varies.

Diaries on the other hand (which are also part of K5's neat content) already cycle far too quickly, perhaps there should be "ratings of diary entries", so the diary "sections page" is reserved for high-rated entries (while any appear in the "newest entries box" like now.. and let diary entries go to the K5 FP if they rate high enough and then get a good percentage of the K5 population to vote for it

I think K5's moderation queue setup works nicely as it is, with variable thresholds based on population and all, and the 36-hr-limit thing improved matters IMO.



-Mysidia the insane @k5
That'd be right .... (none / 0) (#79)
by cafeman on Sun Mar 17, 2002 at 06:19:14 PM EST

Of course, this would happen as my article hits the section area. Murphey's Law.

But seriously though, I'm not worried about K5 at the moment. I think we're just starting to see indications of a problem that's going to become bigger and bigger. I'm looking at the 6 month to 2 year timeframe. It may not happen, but that's why you plan - so you can avoid it if it does.

On a different note, I'm trying to direct people to look at post #69 and my comment to it. I think this may be a viable, relatively low cost solution. It also helps rusty prevent himself from being a bottleneck to growth and / or quality (not that I'm implying he is, it's just that large systems are harder to coordinate and maintain - look at /. I think rusty is doing brilliantly, and has risen to every challenge he's faced). That way, everyone would get what they want - we would still get better written articles, signal to noise wouldn't become a problem, and we'd get huge variety in submissions. It's almost a naturally load-balancing solution. Check it out ...



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


[ Parent ]
Missing The Larger Picture (4.81 / 11) (#81)
by hemos on Sun Mar 17, 2002 at 08:48:40 PM EST

First, a quick correction: on a daily basis Slashdot sees roughly 1/3 of a million people, and roughly 2.3 million people per month.

But the larger problem is the assumption that you are making vis a vis user *involvement*, and that's a much harder problem. What do I mean? Well, basically, once you pass a certain point (from talking to other people, it's somewhere around 50,000 or so) you can continue to get a growth in the number of people, but the time and energy that the people will put into the website diminishs, if you take an average.

So, yeah, you still get quality people - but in contrast to the first launch of the site, where everyone is very involved, you get a increasing percentage of people who lurk, or are merely observing. You can see some of Rob and I's comments about this in the recent IRC forum we did - there's still a massive amount of users on Slashdot, but to be frankly, only *50%* read anything other then the top page.

Comparing K5 to Slashdot though, is one of my pet peeves - and I believe (Back me up here, Rusty! :) ) Rusty agrees. We are two very different sites that exist for very different reasons. My hope is that because of that K5 will avoid some of the problems of overload - but there's still issues with growth to be considered.

The other main issue, at least in my eyes, is the problem of malicious behavior, vis a vis network maintenance. That's where the real cost will start to kick in.

Wow ... (4.25 / 4) (#82)
by cafeman on Sun Mar 17, 2002 at 09:46:23 PM EST

Didn't expect to get a response from you, Hemos ...

One question before I reply - are the 2.3 million people unique people, or unique views? I ask because it changes the parameters of how far you scale before the number of submissions becomes so great that you need to start focusing on filtering posts.

Your involvment point is interesting, but to use an economic turn of phrase, it all depends on the transaction cost of involvement. Writing and submitting an article is costly, so few people will do it. Writing a post has average transaction costs, so more people will do it (possibly your 50,000). Voting on articles is cheap (so you might see 200,000 people get involved). Depending on how you define "involvement", you can get vastly different results. I'd imagine more people that the 50,000 would vote, but the number submitting articles would scale much slower. This would actually solve the submission queue problem, but not the frequency with which the first page would cycle. This is highly simplistic, but you could assume that the number of members (who vote, because it's easy) scales geometrically, while the number of posters scales linearly (as it's harder). The number of article submitters might scale logarithmically. I'd hate to see articles come and go on an hourly basis, especially if it's an article I would have been interested in.

I recognise your frustration at the comparison, but I still think the comparison is valid under certain conditions. Namely, that both have / are experiencing growth, and that both allow users to post comments. The comment rating system is different enough to make meaningful comparisons very difficult. The best you can do is say "this is good because of x, but this one is good because of y".

I compared the two sites because of the activity. Where else do you go for an empirical examination of what happens to a site that allows community postings that goes through extraordinary growth? Slashdot has come up with it's own mechanisms for coping. They're not necessarily the best, but they might have been the best at the time. By seeing what's happened over there, we can try to avoid similar potholes over here. Standing on the shoulders of geniuses, and all that.

The comments system here is interesting as far as growth is concerned, but I didn't want to get into it here because we'd lose focus. I personally think that if K5 grows, the comments system will need to be extended quite significantly to separate the wheat from the chaff. But, that's not what I wanted to discuss here. I only wanted to discuss the story queue.

I think we're in agreement with each other, I'm just trying to explain why I used the comparison. The ethos, philosophies, and operational aspects are very different - no argument there. I guess it's a question about whether K5 should get a bigger population but remain pretty much the same (with reference back to the carrying capacity of the existing system), or come up with models that extend K5 to let it grow with the increased member base.



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


[ Parent ]
displaying my ignorance (3.00 / 1) (#89)
by tenpo on Mon Mar 18, 2002 at 06:03:28 AM EST

Ok, i admit it's been a while since uni. Could someone give me a basic formula and an example of 'scaling geometrically'? Thanks.

[ Parent ]
Easy answer (4.00 / 2) (#102)
by cafeman on Mon Mar 18, 2002 at 02:33:45 PM EST

A search on google for "geometric growth" turns up a lot, but here's a visual example of the difference between linear growth and geometric growth (assuming K5 started on an arbitrary point on the X axis where the two curves intersect). Here's an example.

Hope that helps.


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


[ Parent ]
cheers, math-buddy (n/t) (none / 0) (#105)
by tenpo on Mon Mar 18, 2002 at 08:56:58 PM EST



[ Parent ]
Numbers (4.00 / 4) (#93)
by hemos on Mon Mar 18, 2002 at 10:41:07 AM EST

2.3 unique people. The number of views is 'round about 55 million page views.

I think that comparing Slashdot and K5 is silliness, because they are very different experiences for the readers/users, and as such, will attract different people, albeit with a significant common audience.

[ Parent ]

Good lord (2.55 / 9) (#91)
by rutsy on Mon Mar 18, 2002 at 09:15:29 AM EST

"But to be frankly" that comment contains some of the most illiterate "English" I think I've ever run across. The main issue, at least in my eyes, is the problem of people making absolutely no sense, vis a vis due to poor communication skills, or adding literary Sturm und Drang to make the banalities they're offering seem significant.

Oh, and that latest #forum session was a disgrace on many levels. The only good part was where you promoted alterslash (a knock is as good as a boost :) ) and knee-jerked at it for a few minutes in a deep pool of ignorance.

Wouldn't it be great if everybody who owned a weblog was as great as me? -rutsy
[ Parent ]

Yeah, Not My Best (4.33 / 6) (#94)
by hemos on Mon Mar 18, 2002 at 10:48:04 AM EST

One handed typing while holding baby, and then baby kicked hand, so we skipped right over the 'Preview' step and right into 'Post'. Ah, if only to take back typing already done.

[ Parent ]
Sorry Hemos... (3.00 / 2) (#96)
by ramses0 on Mon Mar 18, 2002 at 12:53:00 PM EST

YHBT by "rutsy". ;^) I didn't think there was any problems with your comment, it gives a lot of insight into the conversation.

--Robert
[ rate all comments , for great ju
[
Parent ]

most illiterate "English" ... ever (3.75 / 4) (#95)
by codemonkey_uk on Mon Mar 18, 2002 at 12:49:37 PM EST

Rusty, are you even reading your weblog these days? I know it's important to maintain the illusion of a Kuro5hin vs Slashdot war, but in this case Hemos' post is a shinning light of clarity in a sea of incoherence.

And no, Hemos, thats not a complement.
---
Thad
"The most savage controversies are those about matters as to which there is no good evidence either way." - Bertrand Russell
[ Parent ]

Look at the username.. (4.00 / 6) (#101)
by Lenins Left Testicle on Mon Mar 18, 2002 at 02:25:58 PM EST

This sad person has used

ruTSy

Not

ruSTy

Mom must be late with the carpool again.

[ Parent ]

Mon dieu (none / 0) (#104)
by bob6 on Mon Mar 18, 2002 at 03:28:08 PM EST

It writes vis--vis, with a grave accent on the a. Usually I don't correct French words but this time I have to make a point.
You made a (very tiny) mistake on the French word either because you are too lazy to type à or you don't speak French. Well let's be indulgent on the English written by non-English speaking persons (like myself).

Cheers.
[ Parent ]
Amen (4.66 / 3) (#111)
by rusty on Tue Mar 19, 2002 at 12:20:36 AM EST

Comparing K5 to Slashdot though, is one of my pet peeves - and I believe (Back me up here, Rusty! :) ) Rusty agrees.

Halleleujah brother. Can I get an amen?!

Yes. K5 != Slashdot. Important fact of life that we all seriously need to get our heads around before I have to read another goddamn article about Slashdot. :-)

____
Not the real rusty
[ Parent ]

AARRGH! (5.00 / 1) (#112)
by cafeman on Tue Mar 19, 2002 at 12:39:04 AM EST

I think I've said it around 20 times so far ... I'm not comparing the two in that way!!!! The sites are different. They have different content. They have different philosophies. They have different rating systems. They attract different users. They place different value weightings on the comments and stories. They have difference focuses.

What I'm comparing is the growth tjhat slashdot went through, and the growth that Kuro5hin is going through. I'm using the growth of slashdot to project possible growth directions of Kuro5hin. I'm looking at the relationship between userbase and comments at slashdot, and trying to map that across to a future state of Kuro5hin, taking into account the differences between the two sites.

Sorry if I sound annoyed, but I'm getting sick of getting dismissed with "The two sites are different. End of story. Everything will work." I keep agreeing that they're different!!! Yes, everything may work out. But are they going to work out the way we want them to? Does it lead to an optimal outcome? Or does it lead to a site where we sit around together and congratulate each other on how great we are while we furiously chase away every newcomer? That's what I've been trying to discuss.

Besides, if you want to try and analyse what happens when a site with high community contribution increases userbase dramatically, what other site can you look to for empirical evidence? This isn't an article about slashdot. It's an article about what growth may do to Kuro5hin. Forget slashdot!!! I wish I had never used it as an example!!!! I wish there was another content rich, high userbase site I could have referred to!!!!!

gasp ... pant pant

Sorry to react that way ...



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


[ Parent ]
Mistargeting, Captain (3.00 / 1) (#113)
by rusty on Tue Mar 19, 2002 at 12:37:55 PM EST

Actually, I wasn't aiming that comment at you at all. I thought that, for a story that had to go and mention Slashdot, you did a good job making it clear what your point was. I was really talking about all the "let's sit around and analyze Slashdot with a sense of smug superiority" articles. This is not one of them.

FWIW, this is one of the better Meta articles I've seen. I even voted for it.

____
Not the real rusty
[ Parent ]

blush (3.50 / 2) (#114)
by cafeman on Tue Mar 19, 2002 at 02:20:44 PM EST

Um ... yeah, sorry about that then ...

I'll just ... be quiet now ;)



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


[ Parent ]
Pre view/vote suggestions (4.14 / 7) (#84)
by kvh009 on Mon Mar 18, 2002 at 01:21:47 AM EST

Things that should happen *** before *** the users view/vote the story.
  1. Please, please, put the story though a spelling check.
  2. Have someone who knows something about the English language look over the story.
  3. Suggest a better title and/or section.

I get really tired of not having #1. If people spend so much time writing a story and submitting it, you'd hope they could at least spell check the story.
How many of the comments on a newly posted story are about #1 and #2...a lot.


...searching for a life less ordinary...

Simple Solution: Stricter Standards! (4.00 / 4) (#92)
by mcherm on Mon Mar 18, 2002 at 09:30:57 AM EST

There's a simple solution to this problem: Stricter Standards. 24 stories a day on the front page is too many. If we are getting 24 SUBMISSIONS a day, though, that's not a problem, because several of those won't make it. If we start getting 100 submissions a day, then we'll just have to raise our standards, so that only a reasonable number of stories get in. Yes, this will make it harder to get a story accepted on K5, but the result will be BETTER WRITTERN stories.

I already so this to some extent: if the queue has been particularly empty for a while I'll relax my standards, while if there's been a blizzard of new stories then I'll be more demanding for a while. If other readers do the same, then the system will be self-moderating.

-- Michael Chermside

For all those still reading this story (5.00 / 4) (#107)
by cafeman on Mon Mar 18, 2002 at 09:37:22 PM EST

dram just pointed something out to me that I wasn't aware K5 had the capability of. At the huge risk of being flamed for not knowing about it, but for the benefit of everyone else who also doesn't know about it, you can customise the first page to display news feeds from other sites (including slashdot etc). To get to it, go to your display preferences. Interestingly enough though, no-one else was mentioned it, which makes me wonder how many people know about it.

This is exactly what I was talking about. I don't have the time at the moment to go fully into it, but good usage of this would effectively help K5 scale indefinitely. With enough propogation between participating sites, participation (through voting) would be high enough to manage the submission queue. Signal to noise ratio within articles would be good because people will tend to hang around and post on the sites they're most interested in. Same goes for story submissions - if you want to get a particular range of responses, you'll submit the article to a particular site. All that needs to happen is for each site to clearly define their particular niche while integrating into the coalliton.

The one thing that would help significantly would be a taxonomy for site categorisation. At the moment, I'd have to check each site out individually to find out who it's target audience is.

Been here for ages, and I never knew about this. Well, I'm ready for your worst. Got the flame suit on and everything ...



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


I agree with your assesment. (5.00 / 3) (#110)
by dram on Tue Mar 19, 2002 at 12:01:11 AM EST

There are a few problems with only using the RDF feeds.

Like you said, people may not know what each site is about. I would be that most people here do not know that Ingenuitas.org is a political discussion site. We would need some sort of index of sites to find our particular niche. This index could also be known as Google.

There is quite a bit of overlap between sites. I don't just mean in types of stories, I mean in the stories themselves. There are many stories that are on /. that will show up in the K5 queue as an MLP.

People don't like going from site to site, having different logins and such. People want just one site that fulfills all of their needs. This is the hardest obstacle to overcome. If there were multiple sites and a user wanted to participate in political and OSS discussions, the chances are, he would have to use two different sites. And people just don't want to do that.

Just sharing headlines is not the same as sharing users. While having smaller, more specialized sites is good, it is only for the people that are only looking for that specialization, and not a myriad of discussion topics. If you take me, for example, I started Ingenuitas.org because all I really care about on K5 are the political discussions. I am not into technology really. So I started a site that had just what I wanted.

On the whole, I think that breaking up K5 is a bad idea. I think that a more distributed userbase would not be advantageous to K5. I think that if it does happen there would be little user overlap. One of the features that Rusty really likes, I believe, is that people bring all of their different interests here and exchange different, competing views that are born from a variety of experiences and interests. The interaction between the political science and the computer science people, or any other person with a different interest, would be lost if the K5 userbase migrated to more specialized sites.

-dram
[grant.henninger.name]

[ Parent ]

Keep dreaming (none / 0) (#108)
by acheon on Mon Mar 18, 2002 at 09:49:03 PM EST

Let's state the ruthless truth :
=>Slashdot crumbled under the weight of idiots =>*Anything* that becomes popular crumbles under the weight of idiots =>We don't want Kuro5hin to become popular ; it is already too much for my taste

Then let's state as a consequence the life cycle of anything that becomes popular :
=>A small kernel of nice people start something that makes sense =>Then they open it to the world and improve it =>As it attracts attention, it draws more and more outsiders, which flows become more and more difficult to control and filter =>Like a party when everyone is invited, it becomes filled with a lot of people who don't belong there and spoil everything, like a disease =>A new kernel of nice people quits because it sucks too much and they start something else

And you want Kuro5hin to become more popular and improve ? Keep dreaming. There is only *one* way to keep such an initiative alive while preventing the above cycle to complete : start to tell the truth people don't want to hear. So nice people will stay and losers will quit. You don't even have to filter anyone. Nice, eh ?

There is only one problem : this is incompatible with democracy and free speech where every idiot can defend his "right" to think and do everything he wants. Like it or not, this is the undeniable truth.

--Martin

How to limit membership.... (none / 0) (#125)
by r00t on Mon Mar 25, 2002 at 01:25:11 PM EST

Slashdot crumbled under the weight of idiots...

During membership signup, a user must write some sort of "IQ type" test to join. Create a database of 500 or so questions, and have a script randomly select 20 or so. If you are intelligent you get in. What would be the cutoff score?


-It's not so much what you have to learn if you accept weird theories, it's what you have to unlearn. - Isaac Asimov
[ Parent ]

Use this algorithm for submissions (none / 0) (#115)
by lvogel on Tue Mar 19, 2002 at 04:12:18 PM EST

Written in pseudocode so that it can be implemented in any language you choose;

int new_submission_count;

function SubReceived(article as string)
{

new_submission_count++;
if mod(new_submission_count/2) = 0 then
move(article, '/dev/null')

else
post(article);
}

This should reduce our sloppy slashdot seconds count drastically.


-- ----------------------
"When you're on the internet, nobody knows you're a dog!"

-a dog
Another K5 load reduction idea (5.00 / 1) (#116)
by lvogel on Tue Mar 19, 2002 at 04:13:26 PM EST

Remove the MLP category. Link the icon to Slashdot. Save a buck or two.
-- ----------------------
"When you're on the internet, nobody knows you're a dog!"

-a dog
Non discussion oriented stories (none / 0) (#117)
by etherdeath on Wed Mar 20, 2002 at 11:39:29 AM EST

How about the ability to submit stories that you don't foresee a lot of discussion on? Maybe have them appear on the right hand side like the news feeds appear.

We get a lot of story submissions about things that almost everyone wants to know about - for instance the recent ice shelf breakup. It's important, but there might not be much in the way of discussion on it.

Maybe I've mistaken what K5 is for... I thought it was a good thing when the front page stories have a lot of discussion. The other question is even if a story contains important information, like the ice shelf story, do we need to have it on K5. Don't people use other news sources? I'm sure almost everyone here does... but then the question is, do we want to make this a news source, tailored in that special K5 way?

If the answer is yes, then I think we could use an area of headlines. Nothing would prevent anyone from posting discussions attached to the story, but it doesn't need to take up 10-35% of the front page.

I spent about 15 minutes looking to see if this idea has come up before.. and I'm sure it must have, I just can't find it, so my appologies if this has been covered before.

alarm notification and searching before posting. (none / 0) (#123)
by johwsun on Thu Mar 21, 2002 at 08:31:14 AM EST

Two things:
First, we need an alarm notification, in order to be alarmed if a submitted story fullfils our personal alarm notification criteria.

Second, we need a pre-post stage, where we will be notified that similiar articles have already been posted, thus it is recomended to go there and comment the previous articles rather than post a new article.

Many algorithms exist that can help somene find postings similiar to others and postings that fullfil some selected criteria. Just look for example at speach recognision or image recognision algortihms.

The question is:
Can someone among the k5 community understand speach or image recognision algorithm and is he/she willing to implement them for us?

Exponential comment ratings (none / 0) (#128)
by revscat on Fri Mar 29, 2002 at 03:35:10 PM EST

Something that may help to balance out the ratings under the load of many users would be the following:


  1. Broaden the possible ratings for comments from 1-5 to 1-10 (or 1-7, or whatever)
  2. Use simple averages (the current system) for ratings of 1-5
  3. Allow trusted users to rate comments higher than 5
  4. BUT (and this is the important part): Ratings that are greater than 5 weigh less in the average that those that are less than or equal to 5. For example, it would take 2 users rating a comment a 6 for one 6 value to factor into the average, 3 for 7, 4 for 8, etc.

So if we had the following ratings for a comment:

Rating     Number of Votes     Weight
   3               10                     1
   4               8                       1
   5               7                       1
   6               3                      .5
   7               2                    .33
   8               8                    .25

Considering only the > 5 ratings, we would have the following:


  • It requires 2 "6" votes for a 6 to be factored into the average. There were 3 "6" votes registered above, and it takes two of them to count, 3/2 = 1, so we factor 1 "6" into the average.
  • "7" votes have a weight of .33. 2/3 = 0, so there are no "7" votes to count.
  • "8" votes have a weight of .25. 8/4 = 2, so there are 2 "8" votes to count.

    My two bits.




    - Rev.
    Libertarianism is like communism: both look great on paper.
  • Friendship web (none / 0) (#129)
    by twh270 on Sat Apr 27, 2002 at 12:18:52 PM EST

    Warning: I'm brand new here on the K5 community. I have no idea what I'm talking about.

    Outside the Internet, people have a small network of friends and acquaintances. A relationship between two people has a certain amount of value to each of them. When John's good friend Sue comments on how good LotR was, John is encouraged to see it -- he values Sue, and her comments and thoughts.

    Perhaps K5 could adopt a similar model, where everyone maintains a list of friends. The articles you see are those posted by your friends, or that your friends recommend.

    The result is a network of friendships. When Sue posts a recommendation of LotR, her friend John sees that, and he is encouraged to both see the movie and pass on her recommendation to his own friends when he likes it. Or, if he doesn't like it he won't pass on the recommendation. So an article's readership will depend on how many people consider it worthy of recommendation.




    How will K5 avoid being crushed by content? | 129 comments (114 topical, 15 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!