Slashdot

From Metagovernment - Government of, by, and for all the people

Slashdot pioneered a mass moderation system whereby every comment posted has a starting score which can be incremented or decremented by semi-randomly chosen moderators. Logged in users commentaries start at 1 (although this can vary from 0 to 2 based on their overall contribution to discussions) and anonymous users start at 0.

When a moderator is given access (i.e. has been chosen), they are given 5 points of influence to play with. Each comment they moderate deducts a point. When they run out of points, they are done serving until next time it is their turn.

Moderation takes place by selecting an adjective from a drop down list that appears next to comments. Descriptive words like 'Flamebait' or 'Informative'. Bad words will reduce the comments score by a single point, good words increase a comments score by a single point. All comments are scored on an absolute scale from -1 to 5.

As such, posts not only are scored, but characterized ("20% insightful, 80% interesting"). The descriptors available are normal, offtopic, flamebait, troll, redundant, insightful, interesting, informative, funny, overrated, and underrated.

Moderation points added to a comment are also added to a user's karma score. Having high karma gives one bonus point to posts made by that author. (Being a registered poster adds one more, so that the highest normally achieved starting score is 2.) Conversely, users with low karma have penalties imposed on them.

Those who can moderate are selected by their karma score (and other criteria). Slashdot editors, can moderate limitlessly. Non-editors moderation points expire after 3 days if they are left unused. Then they go back into the pool and might someday be given access again.

Slashdot users can set a personal threshold where no comments with a lesser score are displayed. A person browsing the comments at a threshold of 1 will not see comments with a score of −1 or 0 but will see all others. They can also configure the value of each descriptor. Moderators can not participate in the same discussion as both a moderator and a poster, this is to prevent abuses but it is one of the more controversial aspects of the system.

Contents 1 Advantages of Slashdot 2 Criticisms of Slashdot 3 See also 4 External links 5 Our Discussion

Advantages of Slashdot

The primary use of Slashdot scoring is to provide filtered presentation of posts. Instead of having to manually sort through each post, readers can use the scores and tags applied to posts to pre-filter the posts for them. As the number of posts to a Slashdot article grow, the Slash system automatically applies some filtering: posts with low scores are still shown by title, but their content is suppressed.

The advantage of filtered presentation is that it allows users to use the community scoring system to help them evaluate which posts to read, and perhaps also how much credibility they should lend to that post.

Criticisms of Slashdot

There are numerous public criticisms of Slashdot.

A primary flaw with the Slashdot post presentation system is in the way posts are sorted. Replies to the first post can take up so much of the conversation, that later posts end up being seen and read far less. Even the second post on an article can be largely ignored if there are hundreds of replies to the first post.

See also

• Good examples

External links

• Slashdot website
• Wikipedia article
• Slashdot FAQ - explains the scoring system

Our Discussion

>>A primary flaw with the Slashdot post presentation system is in the way posts are sorted. Replies to the first post can take up so much of the conversation, that later posts end up being seen and read far less. Even the second post on an article can be largely ignored if there are hundreds of replies to the first post.

Maybe a solution to this might be to hold all the votes for 24 hours before applying them. This would make sure that the comments will have a more fair trial. Also the comments on the same level could be sorted randomly every time you look at them. Still the comments that are made later will have less time to be voted. But this might also be cleard by making sure that every new comments appear for at least x hours from the time it was written. --Pietro 06:44, 8 July 2008 (UTC)

The bad thing about this, is that people I think have come to expect chronological ordering. I hope this is mainly due to the nature of the systems they are using, and that they will use metascore differently, once they see how it works (however that may be). I've put some thought to the ordering of posts a while ago, I hope I can remember the ideas. The goal should obviously be, to find the most interesting/relevant posts to a certain topic. One idea was to intermittently put posts with the least amount of ratings at the top, and encourage users to rate them. This ensures, that all posts get a minimum amount of visibility and ratings. Another idea, was to aggregate the scores in threads, so that not only the most interesting posts would be displayed first, but the entire discussion. The tagging system will allow messages to be marked as referring to other messages and/or as replies. Maybe this diagram will make things clear.

. First post 1
|- Post with least ratings 2
|- Post with highest ratings 3
||- Reply in subthread with least ratings 4
||- Post with highest rating in subthead
|- Post with second highest ratings

1 This post is always displayed, since it is the start of the discussion, all messages following, are in one way or another a reply to this post
2 May have message encouraging reader to rate it.
3 While this post may not have the overall highest rating in the thread, its aggregated ratings give it a higher priority
4 This demonstrates how the algorithm is applied to each sub thread recursively.
--Manuel Barkhau 9:44, 15 July 2008 (UTC)

Hmm, I disagree that some comments (even 1) should always be presented because they started off the conversation. I fear that this would give an impossible to balance advantage to whoever started a conversation. Even to the risk of having people say something silly not to give this position to someone else. But maybe we could let the user clarify when a comment refers to other comments. Like between blog posts you have links and trackback. If then a comment refers to something else, the previous comment would always appear before the second one. Although if the rating is very poor it might be shrink with a javascript. Of course the side effect of this is that messages will tend not to refer too much on other comments to have a higher chance to be above. But this is not necessarily a flaw, as it forces, pardon invites, users to make sure that their comments are self contained. For the rest I keep on thinking that comments on the same level should be randomly sorted. But then open or shrank depending to their rating.--Pietro 22:39, 15 July 2008 (UTC)

How about some form of modified graphical presentation? The analogous format I am think of is the tag cloud, where "importance" is graphically depicted by text-size. In addition to that semi-graphic representation, posts could also be grouped together in a two dimensional plane (or possibly, but probably confusingly in a three dimensional space). Posts could be grouped together by tags that they have in common. In the absence of any commonality by which to group them, they would by default group themselves by date, clockwise around the center (the resolution). -- 21:01, 16 July 2008 (UTC)

I would more think of a mindmap of comments. Where answers to a particular comment appear as subbranch of it. And then the size of the branch represent the support the comment had on the community. Also the links between comments could be represented as interbranch links. All could be done with freemind, and then shown with the flash application. No problem with that.

I'm not a big fan of flash, but if somebody is willing to make a component that represents the structure in a more intuitive way, I won't stop them. Just don't expect it to be the default display mode.

--Manuel Barkhau 11:20, 17 July 2008 (UTC)

I've also been thinking of using colour to depict how much people are in agreement with a post. Say blue for a post which most people have scored positively, red for negative and purple for controversial posts (similar number of positive and negative ratings). Of course that's just the concept, I guess I'll try to make the mechanism generic enough, that any criteria could manifest in a specific colour.

--Manuel Barkhau 11:14, 17 July 2008 (UTC)

Regarding the idea of using tags. What is REALLY important is that we ALWAYS use a broad folksonomy and not a narrow folksonomy. That is a situation where each user can put his own tags on every object that we are tagging, and is made it easier for him to put as many tags as humanly possible (provided he thinks they all matters), and also to reput ALL the tags that are already present that he agrees on. Although this might seem to be an overshooting, and to give a lot of redundant information this is precisely the way to go, because at that point we start to collect not just a set of relevant tags, but a multiset of it. That is a set where each tag carries a certain weight. The weight of how many people agree that that tag is relevant. Once we have this tag cloud we can then calculate the distance in tag space between comments (or resolutions or whatever) and given one find the others that are relevant to it.

I see all this very relevant for resolutions. I am not so convinced for comments. First of all the whole thing really works only if many people tag less resources. It does not work that easily if everybody just tag few comments (because they read them once and don't need to see them again), and it works terribly if everybody tag their own comment. This we really need to avoid. People will forget to tag, do it in the wrong way, with a different mindfram, and eventually it will end up like the keywords in the html page. nice idea, never worked out. But if we all tag the resources, if doing so is easy, if we WANT to do so, because by doing so we can find them again, and if we need to find them again (yes those are many if-s), then tag works wonders. For more info on the topic please also check my blog entry: On Tag Clouds, Metric, Tag Sets and Power Laws.

--Pietro 09:09, 17 July 2008 (UTC)

The way I'm programming it, allows you to tag any resource (post, resolution, revision, user, group, tag [yes you can even tag a tag, a metatag if you will]). Every "tagging" that is the application of a tag to a resource, can be given a score. So if you were to tag a message as insightful, others could rate that tagging, indicating how appropriate that tag is to the message. Another idea that came to mind recently, was to tag specific portions of a post or resolution. This would allow a fine grained expression of peoples opinion, but I think I'll leave it out for now.

--Manuel Barkhau 11:35, 17 July 2008 (UTC)