KarstenSelf wrote on KuroshinSubmissionQueue that the "ultimate goal of K5 and Scoop is: maximize content quality". : I should probably change that to "the goal of K5 and Scoop moderation systems (both comments and stories) is to maximize content quality". The mechanism for doing so is to promote high-quality content, demote low-quality content, and to provide means for improving the quality of submissions through an editorial review process. : -- KarstenSelf 22 Apr 2001 I think this is false--the real "ultimate goal" seems to be to provide a satisfying community experience. Many (but not all) KuroShin visitors believe that the best method to achieve that experience is to maximize content quality. Other visitors have different goals like maximizing the feeling of community participation, or ensuring that everyone's opinions are given a fair hearing. Perhaps one should ask what it means to maximize content quality? Consider if every K5 story had the quality of a professional/refereed journal article--why would K5 be any better or different than an online magazine? Suppose there were 10 new stories posted every day, but only half of those were of the highest quality--would this be a good or bad thing? Alternately, suppose there was 1 story posted every day, but it was of the highest quality possible--would this be better than more stories of varying quality? Who should decide these questions? Much of the original rationale for K5 was to provide a space for higher-quality discussions than SlashDot. Should the goals of the founders be the foundation of K5, or the popular will of the current community? --CliffordAdams (who reads K5 for the comments, not the stories) ---- One of my impressions about K5 is that the "community quality" is considered far more important than "individual quality". Suggestions to improve individual quality like killfiles or rating groups are often countered with concerns about the community quality. (For example, if killfiles were allowed there might be less peer pressure against trolling and other undesirable behavior. If a rating threshold like SlashDot was implemented then fewer people would rate. Etc.) I think the "community quality" approach will inhibit submissions of unusual or uncomfortable content even if it is of very high quality. I used to be bothered by this until I accepted the community-centered nature of K5. (See MultipleViews for my "community be damned" essay.) Now I think that sites like K5 and SlashDot benefit from their strong sense of community, even though it often conflicts with my individual sense of quality. --CliffordAdams P.S. I hadn't read the [http://www.kuro5hin.org/?op=displaystory&sid=2001/4/22/12246/1974 Ratings Killfile] story when I wrote the text above, but it illustrates some of the community/individual differences. --CliffordAdams ---- First, the opening premise of this node is based on sloppy writing on my part. I should have described the moderation subsystems rather than the site as a whole. Note that another design aim is to provide a scalable CollaborativeFiltering engine, from both technical and social standpoints. I've seen sites that scale to a few dozen users (eg: ZDNetTalkback, which falls apart utterly at about that level, also LinuxTodayComments). Threaded discussion boards, without any other filtering, eg, the OldIWEForums, handled a userbase of up to about 10,000, with a hundred or so active participants. SlashDot's current membership is on the order of 100,000, with probably 10% or so actives. KuroShin has about 15,000 members. I don't know WikiStats, but suspect that a good Wiki could scale to several tens of thousands of participants, with sufficient size. Does anyone know PerlMonks stats? The EverythingDevelopmentCompany page gives 100K hits/day as a basis for their minimum license fee (yes, it is free software). ScoopEngine itself seems to max out at about 6-8 sustained TPS, or about a half-million hits daily, on a single-server configuration. In reality, we're talking an hourly peak of about 28,800 hits. K5's load profile suggest that actual peak daily load would be about one third to one quarter theoretical, or about 150k - 180k daily. Actual hits per hour: max: 7627, average: 2679. Per day: max, 96,056, average 64,758 (March, 2001 stats). There are other aspects to KuroShin, including the idea that the whole site is rather more community/user-driven than SlashDot, which is where I believe RustyFoster started from. The moderation pieces are largely my idea, though there's been input from others. As for what people want from K5 as a whole -- that's probably a mixed bag. I'm somewhat disappointed by the experience currently, though I suspect improving the KuroshinSubmissionQueue will help considerably. I'd like a spot where I can toss out ideas for discussion and have informed feedback, as well as provide input on ideas others bring to the table. The topic and quality of discussion flagged somewhat. I'm a tweaker, I'd like to see what can be done to improve it. Regarding the [http://www.kuro5hin.org/?op=displaystory&sid=2001/4/22/12246/1974 Ratings Killfile] stuff. There's actually some merit to people identifying users they do or do not trust. The question is then to decide what to do about them. As I indicated in my response to that article, there's a narrow range in which it really matters -- about 1% of users are "trusted", and only a very small handful, far less than 1%, are untrusted. One the one hand, people are very good pattern-recognition algorithms. On the other, it's fairly easy to pollute the system with bad data (watching the watchman is an issue in any meta-filtering system). My inclination is to lean toward administrative search tools, though this means building them ;-) I'm not averse to user-initiated flags. But they've got to be built into the system carefully. Summary: content quality (high, low, or otherwise -- Cf: GeekAZoid) is one part of the ComunityExperience. -- KarstenSelf 22 Apr 2001 ---- I have the distinct impression, not just from here but from other comments made elsewhere, that one of the major goals of kuro5hin is to have an immensely large readership. A lot of online space proprietors have this emphasis on traffic. I'm not entirely sure why. When I throw parties, I don't let anyone come in any more. The invitation list is restricted because I know what happens when too many people come: the cops usually come shortly after, and the firetruck if you're really unlucky. (Garden hoses are my friends.) MeatballWiki has been growing quickly. This I don't mind. Though, it has already exceeded my ability to keep up with it. I just don't find any bragging rights in the HitCounts. I definitely don't if the quality of writing drops. Indeed, my bragging rights comes from ExternalReferrals. But this is the QualityOverQuantity vs. QuantityOverQuality argument. You choose what you want to emphasize. I just think that the web is full of traffic grabbing bally-hoo. I'd prefer to create something of meritous value instead. -- SunirShah ---- One word: Google. In the beginning was Usenet. And there was Kibo. And it wasn't good, but it spawned a mythology. Then was Archie. And Archie begat Veronica (how he did this while dating her, I don't understand). And the days and nights became weeks and months, which passed, and we had Gopher. And lo, BernersLee created the Web. And it Was good. And the marca smiled on the ether. But lo!, there was a confusion in the world, and the people knew not how to find things. And Yang and Filo begat Yahoo, which established an order on the Web. And it Was good. And Digital begat AltaVista. And the people could find Many things. And fast. But only on the 14th screen. And it was Technichally Impressive. But it was Not good. And Google appeared in the darkness. And while in its simplicity it was a word and a box, and there was no order, yet there was order in it. And the people Felt Lucky. And they were. And lo, the Enlightened One came upon us to tell us Why Google Worked. And it was not for its size, for Google was not Fearsome, AltaVista was. And it was not for its categorization and Web-enabled Librarians, for this was the Trick of Yahoo. No. The Secret of Google was to Let the Web Order Itself. For there was Too Much data, and the people were lost, though, yeah, they had Yahoo, and AltaVista, and HotBot, and Excite, and Inktomi, Lycos, and AskJeeves. For there was no Sense to the Data. And, yeah, what Google did was Use the data to Order Itself. So here's the scoop. It's not that I necessarily want to scale K5 to tremendous sizes. But I would like to create something, with Scoop, that can be overlayed onto a large problem space. Such as, say, Usenet. So that's where I'm headed. The basic moderation concept works. Stories and submissions -- that's something peculiarly K5ish, I suppose. But it's a part of the problem. Right now K5 has the circulation, effectively, of a medium-sized paper. I'd like to see some quality come into that. So, while traffic doesn't matter, of itself, dealing with traffic does. I don't want to keep running away from where all the people are. -- KarstenSelf 23 Apr 2001 CategoryKuroshin