[Wikipedia-l] Forget mentoring systems, we need a web of trust

Thu Jul 1 19:17:34 UTC 2004

Partially cross posted to WikiEN-l since this is more of a global issue:

--- Geoff Burling <llywrch at agora.rdrop.com> wrote on WikiEN-l:
>...
> However, I just noticed today that I've managed to make something like
> 70 edits from a single IP address, & no one's seemed to notice. Not that
> I care, but one has to wonder how many more anonymous contributors are
> quietly working away on a regular basis, adding positively to Wikipedia
> without anyone noticing.

Lots (especially small copyedits and adding interlanguage links). But most of
our vandalism and other newbie behavior also comes from IP addresses, meaning
people on RC patrol have to check *each* one (lots of duplicated effort). A
unified log-in system will cut down on the proportion of good anon edits vs bad
simply by displaying user name's of people who are adding interlanguage links. 

But what we really need is a way to share the workload by excluding or graying
out and reducing the font size of edits in a special RC once they have been
viewed more than x times (3 perhaps) by a certain group of users (a preference
choice between admins or non-newbie logged-in users would be nice, for
example). To be maximally useful, this would have to be applied to the newbie
RC that Brion gave a link to and the anon RC as well. 

In the long term though, we need a web of trust system where each logged-in
user can select the people they trust and even the people whose trust opinions
they trust - thus trusting by proxy. See
http://mail.wikipedia.org/pipermail/wikipedia-l/2004-February/014300.html 

Somebody created a page in the Wikipedia namespace on en.wiki about this, but I
can't find it. :(

But baby steps first since I understand the coding and the computing power
needs of such a web of trust would be significant. An enhanced newbie/anon RC,
as described above, should work real well in the meantime. As it is, I rarely
look at RC anymore since there are way too many edits (20 a minute is not
uncommon). I simply get bored reviewing edits since the great majority of them
(by volume) are good. But since the volume of edits is so great, even a small
percentage of bad edits will result in a lot of damage in an absolute sense. So
we need to work on ways to share the workload and reduce duplicated effort -
otherwise RC burnout will become more common. 

-- mav

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 