Most recent edit on 2005-04-03 09:58:24 by PaulResnick
Additions:
One of the problems with making things invisible is that no one else has a chance to look at the item and decide that it should be more visible. (In a paper on "The Market For Evaluations", we included a little analysis of how this phenomenon is similar to and different from the notion of "herding" in information economics.)
An interesting solution might be to maintain a score for each comment. The score would determine the probability that the comment would be displayed to the next user. If reinforced by a user approval, the probability would go up. If not reinforced, it would go down. But it would never get to a probability of 0, so there would always be some chance of rescuing comments.
--PaulResnick
Edited on 2005-03-30 06:40:28 by 209-204-144-35.dsl.static.sonic.net
Additions:
I think that requiring positive user action to retain comments is a powerful and interesting idea. Could you explain the rationale behind your design of the rule? For example, why is 10 visits the initial lifetime, and why does a vote prolong the lifetime only by up to 5 visits? Intuitively I would choose the extension to be longer than the original lifetime, since having a comment approved by two independent parties is much more significant than having it simply written by one party. Can a comment's lifetime be extended indefinitely? Should a comment become permanent after enough approvals? If there are a lot of comments on one page, should fewer approvals be required since clicking on all the good comments takes more work?
Here's a random suggestion made up completely on the spot. Suppose each new comment is created with one "life unit" and has a "burn rate" of 0.1. Each visit decreases the comment's life by its burn rate. Each approval increases the life by a constant increment I and multiplies the burn rate by a reduction factor R. For example, I = 0.2 and R = 0.7. This method has the property that with each successive approval the amount of extension increases. Earlier approvals have a stronger effect than later approvals (since presumably an earlier approval suggests a higher probability of approval, which would suggest a more worthy comment).
The public annotation system i built in 1997 (see http://zesty.ca/crit/∞) allowed anybody to attach comments to specific words or phrases on any web page, and it suffered from many of the problems you mentioned. Pages of popular websites received lots of irrelevant comments and useful comments were buried among them. Your idea addresses a significant problem and I'd be interested to see how well it works in practice.
-- KaPingYee
Edited on 2005-01-31 16:08:44 by DerekHansen
Additions:
[Please insert comments in this section followed by your WikiName]
Deletions:
Oldest known version of this page was edited on 2005-01-31 15:58:43 by DerekHansen []
Page view:
Scott A. Golder, Judith Donath, (2005)
Content as Conversation Space: Promoting Positive Contributions Through Vetted Annotations