[Wikipedia-l] RfA

Tue Feb 3 21:35:22 UTC 2004

Summary: Current RfA ineffective. Proposed
automated software evaluation of Wikipedia
contributors. only users with high score will be
able to get lised in RfA. Please comment &
criticise.

The current RfA (Requests for adminship) process
cannot serve Wikipedia anymore. Wikipedia is now
a very popular big site and I am afraid the RfA
process we are using can only support small
communities.

Currently 1 user who had less than 100 edits is
nominated. Another user with about 200 edits
requests adminship.

In the past, a user who acted in the
"politician's way" requested to become an admin
numerous times: make me admin and I will do this
and that... Of course the request was ridiculous
and nobody was taking him seriously.

Also, on January 12, 2004, a user was nominated
for adminship. A developer promoted the user to
administrator just 21 minutes after the
nomination, even before the response of the
nominated user (a nominated user has to respond
in order to accept or reject the adminship).
Although everyone, including me, supported the
nomination (after he was made an admin!), and the
admin is now good and valuable to the project, I
think the quick developer's action was
unnecessary. I say this, not in order to rise an
issue with the developer (who is useful and
valuable to the project) but only to show that
the RfA process has "holes".

The current RfA ineffectivity has been
demonstrated many times.

One of the main problems of the current RfA
system is that everyone can nominate any person,
many times not for Wikipedia's good but only for
personal reasons etc. Uneducated
nominations/requests are also common and waste
our time.

To put it in one sentence: It is very easy for
someone to nominate a user for adminship, or even
request it.

I think RfA should be more difficult, so that
whenever someone is listed on it, it will be more
or less sure that he/she is already wanted or
trusted by the community up to some degree.

Requests could be go away. Personally, I prefer
nominations. Or, the policy and the system could
be formulated in such a way that it would be more
difficult for someone to request adminship, than
to be nominated for it.

A semi-automated software system for user
evaluation may be needed, IMO. Just like most
auction sites (eBay etc): Other users will
evaluate a user's edits.

The system will work like this: In Page history,
or in the version differences page, we could have
three option boxes, one textbox and one button,
all under the caption "Evaluate this user's
edit". The option boxes will read: 1. Positive,
2. Neutral, 3. Negative. It will be required for
the evaluator to write a summary in the textbox
and justify his/her evaluation. When the user
presses the button, the system will record the
evaluation in the user's evaluation log. These
logs will be public (accessible via the user's
page), and updated/maintained by the software.

Because abuse is possible, we can have evaluation
moderators. When a new evaluation is submitted,
it will not be written in the user's log until a
moderator aproves it. The log will refer to the
particular edit and article an evaluation was
about, and it will keep some statistics, such as
how many different users made a positive or
negative evaluation. Evaluations which were not
approved by the moderators will be kept in a
separate log but not counted in the "official"
user's score.

The user's score will be calculated by software
from data gathered from the evaluation log. I
suggest the score's algorithm to pay more
"attention" on recent evaluations. For example,
evaluations one year old can be multiplied by
0.5, evaluations 6 months old by 0.75, while the
evaluations from the past 2-3 weeks will be
multiplied by 1.5. Also, evaluations from
evaluators with a high score will count more in
the final user's score (for example, multiplied
by 1.33). The number of user's edits will be
taken into account, too: The higher edits, the
higher the score. So the score is not simply the
number of positive evaluations, but it is based
on more complex analysis and considerations.

***the algorithm which calculates the score
should be published in wikipedia, written in
simple pseudocode so that non-developers will be
able to understand it***

I suggest that the moderators should be appointed
directly by Jimbo.

There will be a page called "Great contributors"
maintained by software. In this page only users
with a high score will be listed.

In RfA, a user may be nominated for adminship
*only* and *only* iff: 
a) he or she is listed in the "Great
contributors" page, AND 
b) the user has made at least 500 edits. 
Otherwise, the nomination will be automatically
rejected.

I do not think users with less than 500 edits
should be even considered in RfA.

I ask for comments and criticism on these ideas.
Also, please, suggest your own ideas and propose
the changes you would like to see in RfA. I hope
we can built a better and more efficient RfA
system.

--Optim

__________________________________
Do you Yahoo!?
Yahoo! SiteBuilder - Free web site building tool. Try it!
http://webhosting.yahoo.com/ps/sb/