-----Original Message-----
From: Lars Aronsson [mailto:lars@aronsson.se]
Sent: Monday, July 11, 2005 11:18 PM
To: wikipedia-l(a)Wikimedia.org
Subject: Re: [Wikipedia-l] collaborative editing system for controversial content
Bennett Haselton wrote:
(c) Indicate that you disagree with the reply if
it's a
rebuttal. If this happens, then a "jury" of other users on the
site will be selected at random to vote on who is right,
One reason that voting is so often avoided in online communities
is that it is unclear who is eligible to vote. While
one-person-one-vote is a beautiful principle, it is not
self-evident who can be counted as "a person". How do you
convince your users that the voters aren't just sock puppets, or
that they are of a mature enough age to make an informed decision?
As to the voter-qualification issue, the theory is that voting on one fact or point of
logic at a time, highly specific, will make it easier for voters to grasp what they're
voting on, in addition to keeping them honest.
The trickier issue is if someone creates hordes of phantom accounts to manipulate the
voting, and that's a real problem that will have to be dealt with eventually.
Adherence to Wiki-community principles means allowing users to participate anonymously,
but that in turn means that one user can create many accounts. Note that since jurors in
"jury votes" are randomly selected among the entire user base, that means if the
number of accounts in the system is large, you'd have to have a very large number of
accounts under your control to ensure getting, say, three spots on a seven-member jury...
but someone might put in that kind of effort.
For right now, my goal is to see if the system can produce useful results *with*
user-verification (that is, I'll just keep track of new users and will red-flag any
large group of "users" with free hotmail accounts or accounts all in the same
domain). If it can't produce useful results *with* user-verification, then it surely
won't be able to produce useful results without it (since that opens up the
possibility of jury abuse). So the first stage is just to get enough users to reach
critical mass where the threads stay active.
Even in the limited discussions that have taken place so far, no votes have been necessary
yet; one unexpected result is that when you're disputing highly specific facts and
logical inferences one at a time, that keeps the users *themselves* more honest, so that
disputes don't have to go to a vote. I hope you'll try it out if you think it
sounds interesting.
-Bennett