Slashdot is powered by your submissions, so send in your scoop

 



Forgot your password?
typodupeerror
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. ×
The Internet

Glasscode Released 152

An interesting article about a web-based discussion system we've mentioned before: nebby writes "A while back Slashdot ran an article which pointed to a k5 submission by myself which discussed ideas about a proposal I had for a new moderation system. Half-empty is an online community with an open submission queue I've been building over the past year which implements this moderation system, among other things, and has been growing steadily with interesting posts and discussion about topics such as government, poetry, stories, and technology. Well, I'm proud to announce that the engine that drives Half-empty, Glasscode, is now available for download. It has advanced features such as distributed skins, category filtering and permissions, and the said global moderation system. Hit the link for how this all came to be, and what I'm hoping for the future."

BBSes, Half-Empty, Glasscode, and my sanity.

Just a bit more than a year ago, me and my good friend Isaac Oates (author of the Eternity BBS software from long ago) sat down and decided to create a website. We were missing the days of BBS yore, where discussion flowed with intellectual posts about all kinds of topics, trolls were sparse, and flames were hearty. We wanted it back. The root of all evil seen in online posting today, Isaac and I decided, was that people were not caring what all of their peers thought of them, and were not in anyway motivated to think through their posts. We also saw alot of the current weblogs out there restricted in what could be posted, and by whom. They were also confusing to the newbie (granted, half-empty is overwhelming right now), and we wanted anyone and everyone with a Internet connection to be able to stop sucking information out and start dumping some back in.

We wanted to create an online community (the kind that Katz has recently been raving about) that would have no limits on discussion and would by its nature make people want to get involved. It would allow the users to get an ongoing rush of content, or eliminate the content down to just being about, say, Birds. It would let the users know what other people thought of them. It would allow for the obvious identification and silencing of blatant trolls. It would be fun to use, and would be addictive.

We started chipping away last January, at the turn of the millenium. Unfortunately, Isaac was sucked away into the depths of UIUC, unable to continue the project. Fortunate or not for myself, this was a project stuck in my head and would not leave me alone until it was finished.. I'm sure most of you can relate. I became addicted to it, adding piece after piece, rethinking the architecture and rating/point system over and over again.. making myself a self-proclaimed psychologist of my users-to-be. "Should points be a reward, or a punishment?" .. "Will they rate stuff down they disagree with?" .. "How much of a focus should be on points, and how much on content?"

I spent most of my second semester freshman year at Cornell (when not doing homework or intoxicating myself) coding this beast. I rebuilt it from the ground up several times, and knew the source code much better than my Chemistry book (and boy, do my grades show this fact..) Summer came and went, and every night after seeing friends I would return home and sit in front of the screen hacking and tweaking away. I saved some cash and got it running on a overclocked Celeron off of e-bay. Half-empty had only one user, one voice, but this would all change soon enough.

I forget the exact day in September, probably the day after we got our cable modem, when I proclaimed to my housemates that we were going to test roadrunner's bandwidth and see what would happen. Knowing that I couldn't afford a real connection, my plan was to open the site, get a gigantic flood of users somehow, and pray that one of them sees what I'm trying to do and decides to help me out. I plugged it in, started it up, set up the DNS, and half-empty.org became live.

Now I needed some users.. I decided the most complicated part of the system (and the most discussable) was the moderation system, which still wasn't perfected. I wanted feedback about both the setup I was doing and the site itself, so I posted a kuro5hin article announcing the site and briefly mentioning the system. I had a steady stream of people checking it out then, the server was stable, and I was happy. A request was made for more information about the specific math involved, so I bit and typed up the in depth explanation linked above. A bunch of "Ideas" (half-empty's content) got posted, and discussion took place with only a few minor bumps.

About 100 people signed up that night. The next day, I was minding my own business when I heard a "Oh shit." from my housemate in the other room.. sure enough, we were about to get semi-Slashdotted (mind you, this was a cable modem) I killed my PC, grabbed the RAM out of it, slapped it into the server, prayed, and surprisingly it survived. I had 500 new users in two hours. The posts were coming in at a pretty crazy rate. (This was the only time that I saw the rate of input that I've envisioned since I started working on the project.)

Within a week or so, roadrunner took notice, and pulled the plug. I thought it was over for a while until I got an e-mail from Tim Wilde of dyndns services.. he had been a member of half-empty during the time it was up, and didn't want to see it fade away. Putting me into their slice of "cool stuff" on their budget (as Tim put it), half-empty would survive. I went into a coding spree for 48 hours, fixing any big bugs I could since the site was going to be dead for a few days. Tim arrived, put the box into the cage and plugged it in, and half-empty was back.

Of course, most of the folks who had been there originally had drifted away because of the downtime. The site has managed to addict a handful of people, however, and we've been trudging on ever since. There have been creative stories and plays discussed, politics, coding, and even a dirty joke or three :) It's become apparent that the moderation system, if nothing else, has caused people to read, preview, edit, and post their thoughts. I'm happy with what it's become, and can only hope that the mentality there will remain the same while the userbase gets larger.

So, today I've reached the end of this road, and probably the beginning of another. I've released the source to Glasscode, and (hopefully) have made it straightforward to setup and install. It's a Java-based servlet application, with many of the features seen in slashcode, with additions such as skinning, appending to posts, selective archiving, user tiers, category permissions and overviews, and plenty more. It provides a component based system for adding new types of content, and there is even an skin development kit to aid in the creation of new skins (which when accepted by the central server will be available to all Glasscode based sites.)

Hopefully this hasn't been too drawn out of a story to culminate in a software release.. I'm hoping that you've been entertained by my struggle against the need to code that most of us have learned to accept and embrace. One thing that many hackers need to learn that computers are just tools, tools which will be ultimately used by people. Linux, Gnome, Glasscode, and all software is there to help people do things or think in ways they couldn't before. With this in mind, Happy Hacking :)"

This discussion has been archived. No new comments can be posted.

Glasscode Released

Comments Filter:
  • I agree. Slashdot has stayed pretty close to its roots when you compare it to some of the other acquisitions going on. I think that Slashdot is missing some of its original personality, but I think that has more to bo with taco being less involved in the day-to-day Slashdot happenings. It used to be Robs site(not saying that noone else was a major part of the site), and him being the typical geek, the rest of us typical geeks identified with him and agreed with most of what he wrote.

    There seems to be a lot less articles posted by taco these days, and I think that most of the complaints on Slashdot (which are for the MOST part, silly complaints) about it changing since the purchase, are from people feeling the effects of taco withdrawals.

    Grow your own dope, plant a man
  • I Visit .5e almost daily.

    Can you say SLASH DOTTED.

    The pages used to take a while to load (over 33.6) now they take about the same amount of time as /. maybe a little slower.

    -- TIM
  • So, would you rather that we lie to you and tell you everything works well? Things break in any business, and you'd be an idiot to base your attitude on the opposite assumption. The difference, folks, lies not in whether one's service seems to break or not, but in how the operators of this service respond to the failure.
  • Perhaps because slashdot has multiple servers with lots of big hardware located at Exodus, and that few other people (except for big names like Yahoo, CNN, etc) can afford big connectivity?
  • I think that I have found a kindred spirit, someone who truly understands Slashdot's moderation / Karma system.

    1. Here is how you get around the Karma max at 50: create a new account! Hell, I have got three! From experience, it took me a year to get my first account maxed out at 50, and then 3 months to max out the second.

    2. Other ways to rack-up Karma: Write something (anything) humourous about Microsoft (carefull, sometimes you get marked down for being a troll). And finding misc. links to similar information is always a good way to get a few points, but it can be time consuming. Here is a great way to get a few extra Karma: Write something that is particularly scathing of Slashdot itself, as long as it is on topic :-)

  • So. Use Karma as a sort of currency?
  • If you wanted something like that, you could probably hack the slashdot code to do it (IE, open story submissions and moderation).

    The Scoop engine [kuro5hin.org] (which powers Kuro5hin [kuro5hin.org]) has story moderation and a comment moderation system that always gives you the points you want.


    Like Tetris? Like drugs? Ever try combining them? [pineight.com]
  • Users are given a certain number of karma points. They can attach those points as positive or negative values, to other accounts. Once they use up those karma points, they're gone -- but they can re-arrange those points if they want

    Think of "attaching karma points to an account" as "certifying that account" and you've just described Advogato [advogato.org]. Too bad Advogato AFAIK doesn't even let you reply to articles without having been certified by a number of other users.


    Like Tetris? Like drugs? Ever try combining them? [pineight.com]
  • It's surviving, just slow.
    ---
    Tim Wilde
    Gimme 42 daemons!
  • It's damn slow, and I really don't see why you'd want to run it. If you wanted something like that, you could probably hack the slashdot code to do it (IE, open story submissions and moderation).
    --
  • by Tin Weasil ( 246885 ) on Thursday January 11, 2001 @09:44AM (#513832) Homepage Journal
    It would be nice to see Slashdot adopt some new form of moderation... one that actually makes sense.

    How many Slashdot readers have been moderated down with a (-1 Off Topic) because the moderator failed to read the (on topic) thread that they had been replying to. This can be real humourous when the original comment in the thread was moderated up.

    Or my favorite: A comment gets scored as a (5 Interesting) and yet NOBODY replies to the thread that the (5 Interesting) comment begins. If it were really interesting, don't you think that it would generate SOME sort of response?

    The moderation system on Slashdot is simply no longer viable for a community of this size.

  • The /. effect tends to make even some of the biggest commercial sites around come to a crawl.
    ---
    Tim Wilde
    Gimme 42 daemons!
  • ...if Glasscode could sue /. for launching a "Denial of Service" attack on their site ;)

    I've been trying to pull up that site for 30 minutes now. *grr*

  • He wouldn't have to walk into a dev group and rewrite their tree. If they weren't interested, he could (gasp) fork. Amazing these freedoms you get with this open source thingy, isn't it?

    The real problem with all these start-from-scratch efforts that are so pervasive nowadays has nothing to do with cooperation -- it's ego, pure and simple. "I made it all by myself!"

  • It should be reasonably accessible right now, we've done some server tweaking and it's being reasonably responsive, considering the load it's under.
    ---
    Tim Wilde
    Gimme 42 daemons!
  • by THB ( 61664 ) on Thursday January 11, 2001 @09:48AM (#513837)
    The problem with any public moderation system is that it tends to promote posts that agree with general consensus, and stifle any other opinion. I have seen this time and time again on slashdot, and it leads to a closed minded discussion based on a single idea of what is right and what is wrong.

    This is especially apperent on topics based around opinions, such as politics. As someone with a many opinions that do not agree with the far left majority of slashdot, I consistently see perfectly valid points and opinions moderated down, or never moderated up, just because they do not agree.

    By only seeing one side of the story it is impossible to get a fair view of the situation, and a now biased personal opinion cannot form. This manifests itself into a closed minded group of people, moving farther and further in one direction, leaving reason and individuality behind.
  • I agree as well, the editors do a pretty good job of cancelling bad stories, even my own.

    But, sometimes, bizarre stuff happens. For instance, if Jon Katz was a normal user, he'd get shouted down or routinely ignored. As it is, his posts are first page news by default.

    When Linux 2.4 was "announced" (Linus posted it, so the developers knew, and waited to see how long it would take the mainstream users and the press to notice), it was done as a story in the Linux subtopic, not the main page. CmdrTaco even commented [slashdot.org] that he had to post it quick, or he'd get a ton of submissions. Well, he probably still got a ton of submissions, since most people assumed it was front page news,and when they didn't see it there, submitted the story. There were 772 comments to that story, which seems like critical mass to make it a front pager. If stories were moderated like comments, this may have gotten pushed to the front page. Of course, the great leader may have been making a selfish choice, to keep the bandwidth pipes open for himself, by limiting the SlashDot effect for a few hours...

    Also, some mod to stories may help with commenting problems. I may have the most insightful comment in the world, but if I don't see the story in the first hour, the moderators won't see it. If you look at comment posts, most highly moderated comments are the first few, giving a definite advantage to quick posters, even those who may not think their idea out clearly. It seems like a recipe for trolls.

    Anyway, some system that mods comments may help, by keeping a story alive. If there is a lot of activity in a story, it could be kept high on the list, and people would be inclined to read it again. I could imagine a system where stories get points based on activity. New stories could have such a high score that they start at the top, and lose points just for existing, so that all new stories start at the top. Old stories where people are still discussing it would rise to the top, based on story mods and new comments. Moderators would get fresh points to use on a story, when it reaches a certain threshhold.

    Yes, it's not fully thought out, but a system where the users have more control over the stories may make for interesting discussions - or at least less knee-jerk posting.
  • 1. make sure Java is really set to use the memory (yes, I read your reply about lilo.conf). Check the -Xms and -Xmx VM options.
    2. Is JIT on?
    3. Use an in-process servlet engine/webserver like Resin (www.caucho.com). Every out-of-process SE that I know of uses a socket to send the request from Apache to the SE.
    4. caucho.com also has a mysql driver that they say is faster.
    5. Optimize the code, simplify, reduce object allocation, etc. Java is not slow if used properly, but that takes work.
  • He wants to run his site, using his software, and receive a bit of recognition for his accomplishment. Well, gee, I can clearly see what your problem is with that. Please. He spent time and effort designing, writing, and planning it, and Half-empty hasn't had a chance to establish itself, yet. But he released the code, anyway. How self-serving is that? If he was really interested in cashing in, he would've either waited until Half-empty became (magically) profitable before releasing any code or just never released the code at all. As it stands, he released Glasscode probably before Half-empty could take too much competition, and to avoid losing Half-empty to some loser who is solely interested in cashing in on someone else's work, he placed a temporary commercial use restriction. Well, shit, we better go and lynch him, now.
  • I honestly have yet to see the Java system that performs as advertised.

    Where are these advertisments, for reference?

  • .5e empty lives in our (DynDNS.org's [dyndns.org]) datacenter. Time to see if our bandwidth is worth all those big bucks. Now if you will excuse me, I have MRTG to stare at. ;)
  • trolls were sparse on most boards, because membership required peer (board) review. twits were kicked+banned, and that was that. and, many boards were payforuse, which probably added to the troll filter.

    as for offtopic conversations, some interesting stuff comes out of that ... i like offtopic, but can't stand the trolling twits.

  • by Anonymous Coward

    Erwin says that Glasscode should be used by Slashdot so that everyone can moderate the submission queue and get the stories they want to read, not the stories that Slashdot decides you want to read.

  • I fine that rather insulting. We survive though nothing but donations, and unlike a lot of dyndns providers we have a policy of being up front about outages. Are uptime over the last year is about 99.3%. We provide support with 24 hours of any query, I would like to see your ISP do that. In addition, I think it is rather rude of you to take are honesty about problems we have had, and insinuate that we are incompetent. In fact, DynDNS is a rather slick little system, and you'll be hard pressed to find other DynDNS service with the kind of uptime that we have. IN fact, we have the reputation of being dependable. We've had people tell us stories about checking every other thing in there set up, cause they could not concieve of us being offline.
  • For the record. If you want to write it in Java, write it in C or C++. Don't give me the cross-platform BS, there are plenty of cross platform large C/C++ projects (all the Quakes; GNOME; Mozilla; the kernel).

    But Java lets you send secure, cross-platform code across a network. The most obvious use of this is web applets, but there are others.

    Arguing for Java on the basis of ease-of-coding (basically saying "I like GC") is a cop-out -- you're trading away a lot of speed because you're too lazy to manage your memory

    I disagree. It's not laziness, it's a tradeoff. Rather than spending the time to manage your memory you spend the time on other aspects of your code. You can pretty much always improve software by spending more time on it, and it's not always worth it to spend the time on memory management instead of another area.
  • Is it me, or did every community have a BBS called "The Pirate's Cove"?

    There was one here, in 509...


    - Jeff A. Campbell
  • But this reminds me of the old "do we need this many text editors" debate. While im sure this is one of the things to keep, there are so many programs of this type on freshmeat alone, even more on the net. With literally 100's or 1000's of any given project type out there, is it really worth it to start one from scratch that you know is going to be mediocre when you can just help the community to make a smaller number of really great versions? It's one thing to write a quick one version to learn how the code works, but does the net need a million varieties of production code? Come on, its not a baskin robbins.
  • /* Slashdot link v0.1.2

    * copyright 2001

    * released under the terms of the GNU Public

    * License v2.0 (see LICENSE.TXT)

    */

    if ((link.category != MAJOR_NEWS_ORGANIZATION) || ((number_of_comments < 6) || (number_of_comments > 200))){

    return SLASHDOTTED;

    }

  • I think trolls were sparse back then because access was much harder to come by.

    Trolls were not sparse back then... There have always been trolls, and there will always be trolls. I will probably never fully understand what makes someone troll though.

    --
  • by QuantumG ( 50515 ) <qg@biodome.org> on Thursday January 11, 2001 @09:54AM (#513851) Homepage Journal
    You're kiding right? The only reason Slashdot posted this was to kill em off! Distributed DOS attack stylez.

    Just Kiding. Does appear to be down however ;)

  • by JWhitlock ( 201845 ) <John-Whitlock@NOSpam.ieee.org> on Thursday January 11, 2001 @09:54AM (#513852)
    I fully support this effort and other efforts. This is why:

    Slashdot is where it is because of community. At first, it may have been the editors doing most of the work picking out cool stuff. Nerds of different flavors showed up because it was cool stuff, and started discussion groups. Soon, the editors didn't really need to start looking for cool stuff, the readers started submitting it. It's barely neccessary for me to look at ZDNet, Salon, or the AP wire, since SlashDot seems to always pick up on the best stories in a short amount of time.

    Slashdot isn't entirely community driven, however. The editors do take some editorial liscense, deciding what is post worthy, what is front page news, editing submissions, etc. Occasionally, a story will only get posted because of the massive number of submissions, against the editor's tastes, but this is rare. The posters don't know what's going on - they just see a little "reject" flag next to their submission. This is probably one of the most frustrating parts for someone who is just starting that level of participation - no feedback from the editiors, besides a binary responce and a 6-month old FAQ.

    This particular project doesn't seem to be "SlashCode, but with X!!!" Instead, he is making an effort to allow the community to decide how things are organized. I think it's a laudible experiment.

    Just like voting systems, no post-and-moderate system will make everyone happy. This particular one may not even scale well. But we need the experiementers, and this one seems to have a good start.
  • by Slackrat ( 128095 ) on Thursday January 11, 2001 @09:55AM (#513853)
    I was gonna mod this up, but you didn't get any replies yet.
  • The T1 is apparantly not the problem. Try to view any of the graphics directly off the server, and they load fine. The main page just takes awhile to process your request, once it does it downloads lightning fast. I would'v suggested using a faster php/perl based approach than... java... for such a cpu intensive site.
  • This is known and, in fact, a big part of the point of half-empty; they want to get a large enough, diverse enough base of users that all opinions will be represented almost equally. Now, mind you, a slashdotting isn't the best way in the world to do that, because /. has a lot of biases, but eventually, things would theoretically even out and work pretty well. We've got some diverse views there already, in fact.

    ---
    Tim Wilde
    Gimme 42 daemons!
  • Or my favorite: A comment gets scored as a (5 Interesting) and yet NOBODY replies to the thread that the (5 Interesting) comment begins. If it were really interesting, don't you think that it would generate SOME sort of response?

    I've had this happen, and I laughed out loud when I saw it.

    I think moderation should be weighted based on the number of replies a post generates, with the additive value weighted by how many replies are direct and how many are more deeply nested.
  • Sorry to say this, but I don't like it one bit.
    • Its too damn slow
    • Its bloated
    • Moderation systems blow
    • Ugly UI
    Thats about it =)
  • by nebby ( 11637 ) on Thursday January 11, 2001 @03:08PM (#513858) Homepage
    Well, it's hard to argue such vague points, but it's slow because it's being slashdotted, it's ugly because i'm a programmer not a graphics artist .. some people like it anyway .. (download the skin SDK), the moderation system isn't too important, and as for it being bloated.. well, whatever. :)

  • I'm going to GPL it fairly soon, I just didn't want to see someone open up a clone of half-empty and oust me before things got started (via better advertising.. remember nebby == poor college kid) I'm not going to enforce it on banner ads or anything like that, really. The only situation I'm worried about like I said is a clone of half-empty.. the rest I could care less about.

    I have already stolen the Glasscode, and would hearby like to announce my clone site for optimists called ... Half-Full. Thank you, good night.

    -thomas
  • Here's a long, convoluted explanation of why I think you have a good idea. You get the best Slashdot discussions when the ratio of well-informed people to ignoramuses is above a certain critical value.

    • Case #1: High ratio of clueful to clueless. Misinformation gets corrected quickly. People can learn a lot by reading the discussion. Example: discussions of Linux on Slashdot.
    • Case #2: Moderate clueful/clueless ratio: You get lots of back-and-forth arguments over facts. "Proposition X is true!" "No, X is false!" "True" "True" "No way, you guys, false!" This is kind of a waste of time, but eventually it gets resolved by someone knowledgeable who says something like, "Well, X is a common misconception among my students at Harvard" or "Here's a link to the relevant DMOZ category -- all the sites there agree that X is true."
    • Case #3: Low clueful/clueless: The people who have a clue get tired of correcting all the people who are misinformed. This is how I feel when I visit the Slashdot science section. It's just too much like work -- I teach science for a living. The people who post intelligent stuff get shouted down by a bunch of loudmouths who don't know anything.
    The really horrible thing about case #3 is that it has a perverse positive feedback effect: the more factually incorrect stuff gets posted, the more likely the well informed people are to give up in disgust. If they're like me, they cruise the section for links to interesting articles, but stop participating in the discussions.

    So here's why it could be really good to moderate people instead of comments. When moderators mod comments, they end up mainly just looking at whether it's a troll, and whether it's logically self-consistent and interesting, and stuff like that. But that doesn't tell you anything about the story's relationship to the universe outside of Slashdot -- you know, that thing called reality. It would be great if there was a way to tell whether the person posting was a veritable fountain of misinformation, or was a world-wide authority on the subject. It also might be good to make the rating topic-specific. E.g., the same guy who posts in Science saying "First you send out 100 space probes in every direction..." might actually be very knowledgeable about Linux.

    BTW, I think Epinions does something like this, although in general I think Epinions is so badly maintained that it's not a model that should be emulated.


    The Assayer [theassayer.org] - free-information book reviews

  • by runswithd6s ( 65165 ) on Thursday January 11, 2001 @04:09PM (#513861) Homepage

    Moderation systems are normally designed for the purpose of promoting the quality of a forum. It's interesting then, that the most prominent moderation system uses a method that assigns subjective opinion a quantitative value. "Scoring" a post "up" or "down" is inherently flawed in that you are allowing an individuals subjective opinion to 'grade' the post, rather than a more effectively classifying or categorizing a post.

    Filtering these forums based on this flawed quantitive value will obviously result in quality posts being ignored. Slashdot does offer some kind of categorization of the post, but it still relies upon scoring to order and filter posts.

    Another factor to consider in moderating systems is accountability. Slashdot, and many others, use an anonymous moderation system. kuro5hin [kuro5hin.org] does not follow this poor practice. Everyone can moderate, and everyone is accountable. You can view who has moderated your posts and view the value they selected.

    So let's tie this together. We want a system that reflects a true subjective and qualitative analysis without the impedance of scoring. We want a system that is accountable. "Grading" a post becomes "Classifying" a post, and filtering becomes organizing. For example, let's say the categories to classify a post include the following:

    • Informative
    • Opinion
    • Off Topic
    • Troll/Flame
    • Pro
    • Con

    If subjective categories such as "interesting" or "boring" are available, so be it. They are simply classifications. Opinion is important, but if we base our filtering on grades that do not accurately reflect content, we loose any advantage we gain through moderation.

    So, what would a forum look like with this system? It could be displayed in exactly the same way we see /., but instead of showing a score, we provide a link to show the categories and number of people that moderated the post. For example:

    The Ozone needs an fsck!
    11 Jan 2001, ^chewie [mailto]
    Informative(11 [slashdot.org]), Boring(1 [slashdot.org]), Pro(1 [slashdot.org])
    The Ozone is in serious need of repair! The US Department of...

    ...and so on. (Man, I hate mozilla keybindings...*sigh*). Thus, you have the system I propose as a base. Quantitative measurements are possible, but should only reflect actual physical facts about the post, such as size, number of links, number of moderations, etc.

  • yeah, you like my UID.
    -Davidu
  • by nebby ( 11637 ) on Thursday January 11, 2001 @10:01AM (#513863) Homepage
    I'm going to GPL it fairly soon, I just didn't want to see someone open up a clone of half-empty and oust me before things got started (via better advertising.. remember nebby == poor college kid) I'm not going to enforce it on banner ads or anything like that, really. The only situation I'm worried about like I said is a clone of half-empty.. the rest I could care less about.

  • Whoever modded this Insightful:

    You suck. I took a break from modding unjustly -1ed ACs Underrated and gave the parent a Funny just to piss the guy off (and support his point), and you made it disappear within about five seconds.

    May your overclocked Celeron disfigure your genitals somehow!

    --Cranky Old School Moderator

  • Doesn't "sound" very robust, but cool nevertheless! ;>

    "Don't throw stones in Glassco... Err, houses."

  • Arg! I can't seem to get in from my slow school connection - krellis has advantage of being there.

    Tell me what makes you so afraid
    Of all those people you say you hate

  • by 2nd Post! ( 213333 ) <gundbear&pacbell,net> on Thursday January 11, 2001 @10:07AM (#513867) Homepage
    Well, one thing I am slightly bothered by is that moderators shouldn't be the ones judging on or off topicness; interesting (as a personal interest), overrated (again, personal judgement), insightful (personal), etc.

    I can trust a moderator to make judgements concerning themselves, but not for the judgement of the community.

    Metamoderation is a way to determine if a person can moderate intelligently, hopefully.

    I can't agree to your view that a +5 interesting doesn't get comments. A really powerful, interesting, insightful, whatever, comment, doesn't need to be provocative or controversial. It doesn't need to generate comments, though obviously it would be nice if it did; I always enjoy getting comments!

    Geek dating! [bunnyhop.com]
  • Here you go again troll.

    Bwahahahaahha!
    Most of my posts make it to 5 (check my karma if you don't believe me). Especially considering how rarely I post to /., being marked a troll is beyond funny :)

    Ok maybe just immature

    Honestly, i don't see why an observation about the performance of GlassCode isn't a valid comment when the code itself is released.

    Dude maybe you do know something about coding, and maybe you don't. But apparently you know nothing about hardware. The Celeron (the machine this app is running on) has only 128k of cache. You ever see a server with low cache try to handle high I/O? Ever wonder why server cpu's have up to 8 megs of cache sometimes?

    Plus, this guy is sharing bandwidth with someone else. Also as someone so appropriately pointed out earlier, many much larger sites have failed beneath the /. load.

    I'll be the first to admit that Java likes lots of memory in the heap, and he's only got 128 megs of ram. I'd be interested to see how slashcode compares to this on the same hardware. Plus, the guy's probably using an non-raid IDE drive. Not the most efficient choice for any server. I'm amazed the dang thing didn't just give up the ghost. It got slow, but didn't die.

    So either point out possible improvements or shut the hell up and write your own.

  • Way to go nebby!
    Cornell Student Linux Users Group @ http://cslug.cornell.edu [cornell.edu]
  • Another reason to support it if you're a slashdot fan: Slashdot got where it is by experimenting. It isn't obvious in advance what will work. But the pace of experimentation has slowed since the buyout and vastly increased popularity. Slashdot is no longer likely to break new ground. This is fine - they now have an important role as a well-understood service.
    Smaller projects like half-empty can afford to innovate because they have less to lose. And if they find some new principle to apply to online communities, maybe slashdot will implement the idea after it's been proven.
  • I like the idea of assigning points to people;

    But I might want it on a per user basis, and not as a community wide thing. Or a blend of the two. I don't like raw mob rule.

    So I can assign points to people, akin to handicaps. To use a popular example, Sig11 automatically gets -3, -2, -1, 0, or +1, because I like his posts.

    This gets modified by the moderation system, which is a per discussion ranking, and then is also modified by a global karma; Sig11 tends to have high Karma because a lot of people like him, respond to him, mod him up, or assign bonuses to him; a general -1, 0, or +1.

    This may devolve into a popularity contest, unfortunately.

    But if this is configurable (ignore global ranking, double global ranking, ignore local ranking, etc), it should be okay.

    I for one don't want to listen to '-' no matter that sometimes he's a real treat, and sometimes he's a real pain. I just don't want to deal with it.

    I dunno, maybe it's a bad idea...

    Geek dating! [bunnyhop.com]
  • The moderation towards consensus issue is a real problem, but how about this as a possibility?

    On my site () I wrote up a piece on using micropayments to enable people to set themselves up as "editors" of a data stream, specifically in this case USENET (but it's applicable to any data stream, actually, including the entire net). People with less time can then 'subscribe' to these editors, who get some kind of funding based on how well they help turn data into information by pointing to useful stuff.

    A similar setup could work on /. -- instead of the current rating system, throw it out. Let everyone rate everything -- but also let people decide who they want to use ratings from. that way, instead of being moderated towards some consensus, they get moderated towards a consensus based on who they feel has a clue (or who they've decided isn't clueles,s if you want to go to opt-outs instead of opt-ins).

    If there are people who's opinions you respect, you naturally give them more weight than people who you dislike. you could build a moderation system that allows for that -- technically it's not too difficult, although it's going to be computationally somewhat expensive. But CPU is cheap. Wetware cycles aren't....

    I think the idea of everyone being able to build a customized "editorial staff" to customize /. or other boards is a really great concept, because then each user really can take the entire set of information, and customize it ot their preferences, down to the individual message level, based on the consensus opinion of those they've chosed as their personal experts....

    Seems to me that's what computers are about -- we're letting the computer do the hard work of sifting all this, by teaching them what to look for -- and doing that by using clues other people are willing to give us. (then add in some form of micropayment system, even if it's a shareware-type thing, and the editors, if they're good enough, can turn it into a revenue stream. The better they are at leading audiences to what they want, the easier it is to pay the rent off of the data mining. If guys can make a living building castles in Ultima for a living, why can't we find ways to make a living in other parts of cyberspace where it might make a real difference?)

  • The moderation towards consensus issue is a real problem, but how about this as a possibility?

    On my site () I wrote up a piece on using micropayments to enable people to set themselves up as "editors" of a data stream, specifically in this case USENET (but it's applicable to any data stream, actually, including the entire net). People with less time can then 'subscribe' to these editors, who get some kind of funding based on how well they help turn data into information by pointing to useful stuff.

    A similar setup could work on /. -- instead of the current rating system, throw it out. Let everyone rate everything -- but also let people decide who they want to use ratings from. that way, instead of being moderated towards some consensus, they get moderated towards a consensus based on who they feel has a clue (or who they've decided isn't clueles,s if you want to go to opt-outs instead of opt-ins).

    If there are people who's opinions you respect, you naturally give them more weight than people who you dislike. you could build a moderation system that allows for that -- technically it's not too difficult, although it's going to be computationally somewhat expensive. But CPU is cheap. Wetware cycles aren't....

    I think the idea of everyone being able to build a customized "editorial staff" to customize /. or other boards is a really great concept, because then each user really can take the entire set of information, and customize it ot their preferences, down to the individual message level, based on the consensus opinion of those they've chosed as their personal experts....

    Seems to me that's what computers are about -- we're letting the computer do the hard work of sifting all this, by teaching them what to look for -- and doing that by using clues other people are willing to give us. (then add in some form of micropayment system, even if it's a shareware-type thing, and the editors, if they're good enough, can turn it into a revenue stream. The better they are at leading audiences to what they want, the easier it is to pay the rent off of the data mining. If guys can make a living building castles in Ultima for a living, why can't we find ways to make a living in other parts of cyberspace where it might make a real difference?)

  • This sounds like an excellent idea. Perhaps you could have a "training period" where you rank every post to initially set your preferences according to the rest of the community. The resources consumed just for the moderation system might be a little prohibitive though.
  • I think this is an intriguing idea, but I agree that choosing the N dimensionality would be difficult. Perhaps it would be best to think it polar?

    Everyone would start at the origin. When the first user in your system mods a comment, it would establish the first relation. If they agreed with each other, they would both move to 1 Each user would come to have a polar coordinate assigned to them as they moded other users up and other users down. Say you really agree with someone's post? You would add their vector * 10 to yours. Say you kind of agreed? Their vector * 5. If you disagreed, negative values would be used.

    I'm not sure whether you moderating their comment should change their vector as well. I'm inclined to say no. Just because you agree with them doesn't mean they would agree with you...

    The only complexity I see to this problem is when things get started. Say you have four users (A,B,C, & D). A mods B and C mods D, establishing two relationships. However, we can't compare these two relationships until somebody from the one group mods somebody from the other. It would be tricky but doable.

    So finally, you just set your threshold by setting an angle and maybe a magnitude and seeing who is close by.

    I'm not exactly sure where trolls would fall into this... and maybe it would be cool if you used 3D vectors instead. You could print a globe of your user set!

    The moral here, BTW, is to define a system that doesn't use landmarks, but, rather, one that allows you to look at the map when you are done a see the landmarks that establish themselves.

    Me thinks I might go find my 3D Calc book and write a moderation system of my own.

  • Well, I had never intended for this to be something I would release. It started out as me making a website from scratch, not necessarily making a forum system from scratch. It evolved into something that I thought others could use so I started writing it in a manner that would allow people to use it for any site, not just half-empty.

    Besides, I couldn't fork a Perl/PHP project like slash/scoop and use Java, now could I? :)
  • "Just another case where the admin of an obsure site on the Internet tries to get his website known by posting a story to slashdot (free advertising). That really bugs me..."

    Their submission was relative, technically related, and most definitely "News for Nerds." That's what /. is all about, besides, these stories need some sort of site to provide more information with and who better to benefit from the /. onslaught than projects like Glasscode.

    It's not as if /. posted a story from Betty Crocker because of a new cookie (cookie = edible food | cookie = internet term) product they're putting out.

  • A comment gets scored as a (5 Interesting) and yet NOBODY replies to the thread that the (5 Interesting) comment begins. If it were really interesting, don't you think that it would generate SOME sort of response?

    Unless everyone who read it was just in awe of the post that they're wiping the drool from their mouth instead of replying to it ;)

  • He tried to re-span k5 about 4 weeks ago but people got the drift and moderated his story submission down to the point where I don't believe it was posted. Well gotta hand it to him for persistence, however shameless it is.
  • This is a simple facet of democracy which we're all aware doesn't work. Mainly because we're a bunch of hypocrits who say one thing yet go and do another.
  • Then how to promote diversity?

    Allegedly, the electoral college system in the US voting scheme is supposed to allow for that;

    Gaining the majority vote is not enough to win, you need to get the majority vote in several geographic areas, and thus force yourself to appeal to several demographics, and not just general mob rule.

    Can something like that occur here?

    Something like that happens, in the sense that supposedly anyone can moderate. If an even distribution of moderators exist, then the statistical model should represent the views of slashdot. Supposedly.

    But then there are other problems; even I sometimes don't read at -1 or 0, because I want to avoid the trolls. I try to avoid modding up +2 or +3, unless they are *really* good, and concentrate on modding the +1s and the responses to +2 or +3, that provide good counterarguments.

    Is that appropriate? I dunno, is there a moderator's training page to provide good behavior? That's the best I can do, for now.

    Geek dating! [bunnyhop.com]
  • umm.. actually Krellis ISN'T there, he's a few hundred miles away. Nebby is the only one close to where the machines actually are....
  • crashing said competitor's site and all. :-)
  • See he only cares about making it big like CmdrTaco did. Under the facade of 'community building' he just cares about making cash.
  • The number of people who actually bother to read DynDNS.org's status announcements is so pitifully low it's not gonna make a difference.
    ---
    Tim Wilde
    Gimme 42 daemons!
  • About 130 people have clicked on that link. I seriously doubt that it's making much of a difference.
  • I wish I was. I'd rather be there than here. *sigh*
    ---
    Tim Wilde
    Gimme 42 daemons!
  • Hee hee! What irony that your post has been moderated up to 4 so far!

    My experience with Slashdot moderation has been grrrreat! No, I don't think I've ever been moderated down unfairly. I got hooked on the whole game of being a karma whore, and once I understood a little about how the site's culture worked, I found it was really easy to rack up karma. F'rinstance, I realized that a really good way to get some karma was to look up links with more detailed information on some of the science articles being discussed, which were often short fluff pieces from Wired, etc. Oops! You know what? It may have been karma-whoring, but it was also kind of, er, informative...

    My only criticism of the moderation system is that now that I've maxed out at 50 karma, it's kind of boring. Makes me want to goat-sex myself back down to zero and start over.


    The Assayer [theassayer.org] - free-information book reviews

  • by Ross C. Brackett ( 5878 ) on Thursday January 11, 2001 @10:43AM (#513889) Homepage
    Actually, there once was a time when /. was pretty devoid of trolls. Taco and Hemos used to post in open forums. People posted patches in response to Ask Slashdot questions. Technical arguments by industry experts were common.

    Then the trolls arrived. At first they posted genuinely - and since everything they said was mindless drivel, they decided that /. was lame and that trolling was okay.

    So trolls will tell you that they're doing a good job saving /. from lameness. I wish they could know that /. wasn't dumb until they arrived. But I can understand - I didn't have much perspective when I was 14 either.
  • Heh heh. I live in Denver, which is in many ways a hellhole and an utterly rotten city, but there is one thing I am very proud of: we had no big party for the '99/'00 rollover, but had a massive blowout for the turn of the millenium. A European fireworks artist, music, parties, the whole bit. It was really cool, and for once we did things right.
  • by Crag ( 18776 )
    This is sounding much more like the perfect moderation system then anything I've heard yet. I believe this would make an excellent model for moderation of other forms of data as well. This model would allow parents to define "pr0n" for their own family.

    The problem with the systems we have in place now (on and offline) are exactly the one-dimensionality you describe. There is certainly a "goodness" we can all agree on at the extremes (love good, murder bad), but outside of those extremes our own personalities and freewill make simple value judgements a detriment, be they restrictive or merely descriptive.

    I'm going to save a copy of your post for later re-hashing and possible implementation. I'll try to give what credit I can.
  • Ok i'll bite.

    We have databases for a reason. They allow you to manipulate, store, and access data faster than text files.

    Lack of useful features? Uhm. Ok.

    Try the "Text-only" skin.. it's not perfect but it looks decent in Lynx.
  • by bcrowell ( 177657 ) on Thursday January 11, 2001 @10:13AM (#513899) Homepage
    Slashcode is great for doing what Slashdot wants to do: handle bjillions of simultaneous users per topic, never have to censor, allow anonymous posts, and be very selective about the new topics that will be started.

    Slashcode may not be the best model in many other cases. For instance, I run a slash-ish site [theassayer.org] for book reviews, focusing on reviews of free books. The selectivity-about-topics part of Slashdot is obviously completely inappropriate for this kind of site, since the equivalent of a Slashdot article is a book, and there's no reason to exclude books. Also, a particular book is likely to be discussed only sporadically, not in a Slashdot-style feeding frenzy, so I didn't need Slashdot's mechanisms for getting rid of first-post trolls, but I did have to implement a system for people to ask to receive e-mail notifications when discussion is posted about a book they're interested in.

    A lot of things are a matter of taste and culture, and one size does not fit all. A lot of Slashdotters are paranoid types who have filled the margins of their copy of Cryptonomicon with conspiracy theories. So it makes sense that Slashdot allows anonymous posts. However, for most discussion sites, the single simplest thing that can be done to get rid of trolls is simply to disallow anonymous posting -- make people at least put their nick on their posts, if not their real name and e-mail. For book reviewing, it's particularly important to have some idea of who the reviewer is and what his qualifications are.

    BTW, this last issue -- does the person posting know their posterior from a cavity in the earth? -- is, in my opinion, the place where Slashdot is the most deficient. It's fine when you're reading discussion on a computer topic, since most Slashdotters are computer nerds, and mistakes get pointed out really quickly. But it's a big problem in the science section. A lot of the people posting there got their ideas about science from Star Wars. You get ridiculous stuff like people saying that asteroid mining can be accomplished by "dropping" asteroids into the Earth's atmosphere, where air drag will slow them down and let them crash to the surface. So this is an example of how one design doesn't necessarily work for everything that even one discussion site tries to do.


    The Assayer [theassayer.org] - free-information book reviews

  • First you e-mail me a while ago asking for my source code so you can make some commercial website. I said its not ready and it will be released eventually for non-commercial use. Then you made some threats or something ridiculous. Then you spread lies about me on k5 .. and now this. Would you please grow up? You're embarassing me and yourself.

    Ironically enough, your selfish behavior was a large factor in helping me decide to release it initally as non-commercial, to protect myself from people like you.

    As for my "community building" being a facade.. look at the team of people (including myself) that bring you half-empty. We're not the most selfish of types. You can't make money from a site like half-empty, anyway. I took a lot of crap on k5 about the fact the site had banner ads. What nobody realized is that banner ads does not equal revenue for myself. It goes to dyndns.org's bandwidth bill.

    I told you to call me (and gave you my phone number) if you had any guts to talk to me in person about what your problem was. You never did.
  • An observation about K5 v. /. moderation:
    On K5, the latest front page story has 60 comments. Some statistics:

    50 are topical, 10 are editorial.

    of the 50 topical, 35 have been rated.
    of the 35 rated, only 8 were rated by just one user. the other 27 were rated by multiples users.

    On /., only 2 comments out of the 85 comments made on this story have made it above a 3.

    K5 lets any user rate any comment at any time. As you can see, it means that a significant portion of posts *do* get rated. Yeah, K5's way is probably more intensive on the server, but it seems that more comments getting ratings can only mean good things for readers.

    --Robert
  • by istartedi ( 132515 ) on Thursday January 11, 2001 @10:45AM (#513905) Journal

    ...have a filter to prevent excessive whitespace in a post? Come on, Cowboy, time to update the lameness filter. How hard is it to check for an excessive number of consecutive <br>s?

  • Yeah, but if every slashdot user could moderate, we'd never be able to load a slashdot page.

    I think perhaps the slashdot source people need to maybe redo or rethink moderation. I can count on one hand (2 times) the number of times I have been able to moderate. Maybe a good compromise would be to have moderation be available when you have more than 10 points, and it's been more than 2-3 days since you last moderated. I think that would be a better balance than the current haphazard moderation slash has now.....
    --

  • I used to be an avid BBS user way back when. Part of the problem is that there is no real sense of community in a lot of places. There are way too many people and sometimes this creates cacophony. BBSes used to be maybe 1 or 2 dialup lines and people used to mostly find out about them thru word of mouth or in the case of "31337 boards", thru invites or reputation or what haveyou. There are still a few telnettable BBSes with real intellectual and indepth discussions out there if you know the right people and know where to find em.
  • What about moderation of people instead of articles? use the "karma" concept.

    Users are given a certain number of karma points. They can attach those points as positive or negative values, to other accounts. Once they use up those karma points, they're gone -- but they can re-arrange those points if they want.

    Say everyone on /. gets 100 points to award (or anti-award, with negative points) to other users. Over time, the posters that are a consensus positive get positive karma values, the trolls go consensus negative. Users can then filter messages in a thread by limiting their view to what the user's net karma value is.

    There are limitations and ways to abuse this system, but they can be limited. For instance, one way to limit the impact of trolls is to base the number of karma points a user has to give out on the karma value of that user. So the more negative your karma value, the less 'damage' you can cause by spreading around karma.

    If users knew their messages would have a global effect on their being read by others, it'd be a nice incentive to be careful about their postings....

  • by babbage ( 61057 ) <cdevers@cis.usou ... u minus math_god> on Thursday January 11, 2001 @10:59AM (#513913) Homepage Journal
    But then there are other problems; even I sometimes don't read at -1 or 0, because I want to avoid the trolls. I try to avoid modding up +2 or +3, unless they are *really* good, and concentrate on modding the +1s and the responses to +2 or +3, that provide good counterarguments.

    Would it help if we took a new look at the browsing system, rather than moderation? What if we were able to add a rule to the display options such that you don't see any posts moderated as "-1 Troll"?

    Better still, what if you like to read Slashdot for a good laugh, don't mind if the topic strays a bit, and really don't like those posts that make you learn something. It would be nice if you could add virtual adjustments for each level of moderation: +1 funny, +0 offtopic, and -2 informative.This way, you can filter out the things you don't want to see, while bringing back in some of the ones you weren't interested in.

    The next step from there would be to open everything way up: you've gone part of the way from crude "good / bad" descriptions to a more meaningful content based description. You can augment that by adding in more adjectives for moderation. These adjectives don't have to be so much in terms of good vs bad, but rather as adjectives are meant to be: descriptions.

    Political. Poetic. Surreal. Sublime. Yet Another Microsoft Bashing. Yet Another Linux Stroking.

    You get the idea.

    The final step would be to make moderation a much more common occurance, because you would have to gather much more input to determine which posts have which attributes. I'm not sure where the line there should be -- on one hand, it's tempting to say all users can always moderate once they get the 20 karma points or whatever it is. That would probably be a black hole for system abuse though. Maybe a second tier? Those above, say, 30-50 points {?} could moderate all the time, while those between 20 and that upper tier cutoff would get to periodically like they do now. I'm not sure, it would have to be fleshed out if the idea got anywhere near that far.

    In any event, I've been disappointed in the moderation system for months now, and some sort of change would be very welcome.

    I dunno, is there a moderator's training page to provide good behavior?

    Does this [slashdot.org] count? As they say, read the FAQ... :)



  • by Fervent ( 178271 ) on Thursday January 11, 2001 @09:14AM (#513915)
    For all the flak Slashdot has been getting recently (selling out to VA, not having a sufficient moderating system for the posting of stories, duplicate story posts) there is one thing its always done right: it isn't afraid to tout what could be a potential competitior's wares and site.

    I give kudos to Slashdot for being one of the only commercial sites on the web that does this.

  • Trolls were sparse? I doubt it, you didn't know that you were being trolled.

    Truly. I'll wager that every single person here who was into the BBS scene, at some time called some local lamer warez board, chatted to the sysop, and pretended to be a cool phreakin' dude from the other side of the planet. We sure did.. lock your modem to some crazy speed like 4800 baud, blame it on the noise from all the different boxes you were using to phreak, talk in broken english.. and watch this high school kid running a BBS off his twin-floppy no-hard-drive A500 bullshit you about his 500meg of 0-day warez.. hilarious trolling fun!!

  • You can never go back to "the good old days". Trolls were sparse? I doubt it, you didn't know that you were being trolled.

    I remember the BBS days quite well, I operated a BBS here in the Chicago area (The Pirates Cove). We were a popular BBS, but at its peak there were only 300 active members, and all were computer enthusiasts. No one was given access without being reviewed first.

    With the web, anyone and everyone can visit a web site and post. You will never be able to eliminate trolling.

    Personally, I think that the trolls are an important part of the experience. Don't like it, browse at +1.
  • The thing is getting toasted right now b/c of the /. effect, its hard to say if its because of the servlets or because of the amount of traffic (mind you, it's running on a celeron with 128MB RAM :))
  • Hey there.. I'm the "another web guy" you're talking about, and I've tried my best to make it be functional and readable at the same time.

    There's 6 other skins. Take your pick. If they all suck to you, then sorry I did my best. If you're really that annoyed by it, you can try downloading the skin SDK and making your own skin and sending it to me :P
  • Actually, the server is in Ithaca, NY, and I'm in NJ.. still home for winter break.

    The server is slowly murmuring to its lonely self .. "Why?? Why??"

    :)
  • You can never go back to "the good old days". Trolls were sparse?

    Well, you can put them in a box instead. Hopefully, this site will suck all of the Katz inspired intelectuals, poetry, politics and such out of Slashdot. Fork!

  • As a person who has far-left opinions (read: progressive) I would say that the majority on /. is anything but far-left. But this does re-enforce the idea that this poster is making: how do you moderate opinion? Since confilicting views are necessary to have a productive dialogue, this is something that should be addressed.

    My experience on /. is that a certain slant will start to appear in a discussion, be it liberal or conservaitve, then this slant is made more pronounced by the moderation system. Opposing views never get moderated up enough. The impact is two fold: those with opposing views stop posting; and moderators stop using their moderating points on the conflicting posts, since they have litte chance of moderating a post to visiblity.

    I've never visited half-empty before, but I'm looking forward to see how they solve the problems. Hopefully, some of the ideas can be used on /.

    eric

  • by aheitner ( 3273 ) on Thursday January 11, 2001 @11:23AM (#513939)
    Bwahahahaahha!

    Most of my posts make it to 5 (check my karma if you don't believe me). Especially considering how rarely I post to /., being marked a troll is beyond funny :)

    Honestly, i don't see why an observation about the performance of GlassCode isn't a valid comment when the code itself is released.

    Yes, I'm an anti-Java person. I think i'm a pretty well justified one, in general. I honestly have yet to see the Java system that performs as advertised. Now maybe GlassCode really is just under the weather and is usually better, but you'll forgive me. My last two anecdotal Java experiences were:

    - Using the SameTime (IBM's message client, used a lot internally there) Java client on Linux. The beast took 90 megs of space (RSS). By contrast, the SashXB SameTime client (which my group wrote), took only 30 megs, and ran much more responsively (this was on GHz pIII's, running IBM's absolute-latest JITC-based JRE). And the SashXB version was written in JavaScript on top of Mozilla :)

    - Using JAlice, the new version of the Alice virtual reality tool (built at CMU). They have terrible problems with pausing and catching in the middle of rendering (they had to totally give up on the laughable Java3D, and write JNI around DX8, which kinda misses the whole point, doesn't it?). It would render smoothly for a bit, and then just freeze for a significant fraction of a second, a couple times a second. They're still fighting the problems.

    For the record. If you want to write it in Java, write it in C or C++. Don't give me the cross-platform BS, there are plenty of cross platform large C/C++ projects (all the Quakes; GNOME; Mozilla; the kernel). Carmack himself said (paraphrase) "I like the discipline being cross-platform brings to code". Arguing for Java on the basis of ease-of-coding (basically saying "I like GC") is a cop-out -- you're trading away a lot of speed because you're too lazy to manage your memory (I agree, it's tough. Everyone who writes code should be forced to pass CMU's OS class, 15-412, or restricted to writing HTML/JS/VB and nothing more complicated...). But even if Java was compiled just like C, it would be a lot slower, due to some fairly subtle language level tradoffs (basically the heavy reliance of Java on dynamic type and function resolution, and the fact that almost every primitive operation in the language requires doing such resolution).

    But if you can write it in a language specialized to the task, do so. PERL can be gobbledygook hard to read. It can be quite elegant if coded well. But either way, I guarantee you the PERL will allow a much more concise representation of the problem. Compared to Java accessing SQL ... even if the SQL parts run as fast, the Java will lag heavily in the REGEXPs, and will be no where near as clean as PERL at representing them in code.

    I agree, it was unfair of me to rag on GlassCode under /.'ed loads. But it would be very interesting to do a performance comparison under ideal conditions.

  • by fm6 ( 162816 ) on Thursday January 11, 2001 @10:31AM (#513940) Homepage Journal
    First of all JSP pages will always load slow the first time because they're being compiled.
    I have to say that I've always wondered at the logic of embedding server-side applications in content pages. I find that backwards -- applications should generate content, not vice-versa. Makes for an app that's hard to test and maintain. How many times have you encountered broken ASP pages?
    Secondly servlets, being bytecode, are *plenty* fast. I'm pretty happy servlets were used over .
    Let's not get into "mine is bigger than yours" nonsense. Suffice to say that Java often gets a bum rap. This has less to do with the interpreted-native dichotomy (most commercial-grade Java VMs aren't simple interpreters anyway) than with the nasty rep Java earned early on. And this had mainly to do with primitive VMs, poor browser integration, and bloated apps written by inept Java novices.

    In any complicated system, there are any number of factors that can affect performance. People have a nasty tendency to focus on one "obvious" bottleneck (program speed in web apps, cpu speed in PC apps, buffer size in comm apps), completely neglecting all the other, usually more important, factors.

    If you are writing a complicated system like Glasscode in Perl, well, God help you.
    Now who's trolling? Perl does seem to attract more than its share of sloppy programmers (slackers like languages that do your string management for you ;-) ) but it's quite possible to write tight, structured, object-oriented code in Perl. I personally will never come to terms with all the weird syntax and non-linear idioms -- but that's just the way my brain is wired.
    Writing it in C is just plain stupid, and is the icing on your troll post.
    Now you're getting personal. My company does Delphi [borland.com] and C++ Builder [borland.com], both of which are very good at generating native-code web apps. These products have a loyal following in the NT world (and will soon move into the Linux world [borland.com]), but I think this has more to do with being a good RAD tool, generating clean and easy to maintain apps, and having solid software component support. Native code performance probably helps, but I doubt if it's a crucial factor.

    __________________

  • Hi Troll,

    You think perhaps the /. effect may have anything do with it?

    First of all JSP pages will always load slow the first time because they're being compiled. Secondly servlets, being bytecode, are *plenty* fast. I'm pretty happy servlets were used over <insert your hackish CGI preference>. If you are writing a complicated system like Glasscode in Perl, well, God help you. Writing it in C is just plain stupid, and is the icing on your troll post.
  • by Tridus ( 79566 ) on Thursday January 11, 2001 @09:26AM (#513952) Homepage
    Back when he started this project, I bet he could not have just walked into say slashcode's development group and rewritten the entire mod system without somebody stopping him. You have a lot more freedom when you write your own project to do whatever you want with it.
  • I think your most informed comment is "GlassCode in C/C++ wouldn't necessarily help, if its algorithms are dumb."

    Perl has it's place, so does Java. If Perl is written with out reguard to how it works underneath, it is slow; same with Java. I've seen many very slow systems that allocate and deallocate tremendous amount of memory for every transaction.

    Java is sort of like VB, it is very easy to write bad code; it is difficult to write good code. In perl it is difficult to write/maintain anything, hence most things have been somewhat thought out. Linus uses this as an arguement to not allow debuggers into the kernel. It would enable lesser programmers to participate creating ugly code (apparently me included).

    Since I haven't connected yet, I'm assuming that there isn't a cache (like squid) sitting in front of the server, which does seem dumb.

    Joe
  • Nothing in life is perfect. Even moderation, every site is different, glasscode would in no way work on slashdot, because there are too many bad apples out there that would like to mess it up. The community determines the moderation, if any, on every site.
    --------------------------------------
    I'm a karma whore, mod me up damn you!
  • Actually, I do get it, and I disagree with "grading" a post. It is an over-simplistic approach that we have seen doesn't work on /., half-empty, and other similar systems. Why? Quality posts get ignored or graded down because people don't "like" them. People grade up or down based on what they "like." Differences of opinion should not be the criteria for "grading" a post. Remove the ability to "grade" a post "up/down", and you remove the bias. Remove the bias, and the pro/con arguments can be graded based on content.

    Posts that get attention will be moderated. Posts that don't get attention will be unclassified. The way you filter the posts is determined on what you would like to read. Would you like to read the 'pro' arguments, the 'con' arguments. Comments that were 'pro' and 'interesting', or 'con' and 'troll'. +1 or +5 tells you nothing more than the idea that a number of people "liked" the post. Now your filters are subject to the general concensus of the population that uses the forum. If that population consists of a bunch of neo-nazis, their grading will reflect their personal biases. If they're a bunch of Windows-huggin users, the BSD-huggin users will likely get a consistently lower grade on their posts.

    No. "Simple" grading schemes may be fast, but they're not effective.They are not useful. They are subject to the bias of the population of the forum. Frankly, I'm tired of them.

  • Can anyone suggest a way to speed it up, software wise? I'm using IBM's VM 1.3, which I'm pretty sure is the fastest available (correct me here if I'm wrong). I'm using JServ, which of about 5 servlet engines I tested with it, ran the fastest.

    I do agree that the bottleneck seems to be the servlets (not Apache).. weither it's the MySQL or the Java I can't tell.

    Any suggestions?

    Then again, since it's a Celeron 450 w/128MB RAM dynamically generating content, I suppose there's only so much you can expect when faced with the slashdot effect :)
  • by blamario ( 227479 ) on Thursday January 11, 2001 @12:07PM (#513963)
    The solution is to split the sheep herd into several sub-herds whose members happily agree, and let the black sheeps which want to hear differing opinions roam among them.

    Imagine an N-dimensional space, where every user and every posting is represented as a point in that space. Whenever you rate a posting with "+", your point and that posting's position move a bit closer to each other. When you rate a posting with "-" these two positions move away from each other. The poster's point also moves in the same direction as his postings.

    Now, when you post a new comment, at first it will appear right where you are. After several "moderations" from other users it will be moved to where it's liked better, and you will be dragged along after it.

    After many iterations, this kind of "Slashdot Space" should evolve into several clusters of think-alike users and comments they like. So if you rate negatively all KDE-vs-Gnome pissing contests, you'll soon get far away from them, and from the users that post them.

    Your reading treshold would be maximum distance between your own position and position of acceptable posts. Or you can think of it as "eye-sight". Another useful thing could be to assign each comment a "size" quality, or visibility. Comments rated positively from many different users "grow" and become visible from afar. This is probably necessary to prevent a split-up of the forum into several group of users who never hear of each other.

    Now the difficult part is the choice of N (dimensionality) and the initial system state. You could set some meta-positions or lighthouses, landmarks, whatever, that represent the unmovable positions people can use to orient themselves. For example, there could be a M$-bashing landmark, Pro-M$ landmark on the opposite side of the universe, goat.cx landmark, Pro-Napster landmark, etc.

    Cool thing is, if you keep a database of all +/- ratings, you can always generate a new version of the space if a need arises, like if a new dimension (er, landmark) pops up. And even better, you can "dumb down" the space to any two dimensions and show it as a graphic on demand. Now that would give a whole new meaning to whereami command :).

  • I sped it up alot by adding mem=128M to lilo.conf.

    And you made fun of me because it was impossibly slow.. it was running Java on 64 megs of RAM

    Someone shoot me :) Let the flames against my lack of knowledge begin.
  • by yesthatguy ( 69509 ) on Thursday January 11, 2001 @12:32PM (#513967) Homepage
    We started chipping away last January, at the turn of the millenium.

    Wow, you coded all this in ten days? We need more programmers like you!
    ---------------
  • I really am not a license snob and in fact in many ways think that the *BSD license is better than the GPL but also think the GPL has it's strong points. But the license he is putting this out under is really bad and it is *not* OSS or free software. Why do people put in commercial use clauses. This really restricts uses of the software and is not really a clear thing. Think about it if I start a site based off of this and start out doing nothing with ads or any kind of income this would be within this license. Now if in a few months it starts to be big and I decide to get a bit of income going so I can improve the site so I start selling a couple of ads. Now do I need another license. Damnit Jim free software is all about being able to do anything I want with it . If I want to make money I should be able to. Just use the freaking GPL or the BSD license and get over it. This is the kind of not_really_free_but_kind_of_free_software hurts us all.
  • as to why Kuro5hin [kuro5hin.org] has been linked to Slashdot's front page, yet AGAIN?

    I found out about it from a slightly more difficult to find sid on slashdot than the TOP article on the FRONT page..I am of the belief that it is best to let people hunt a little bit for such a site, rather than repeatedly subjecting it to the whole panoply of lusers that could possibly be accessing the front page of Slashdot.

When it is not necessary to make a decision, it is necessary not to make a decision.

Working...