[WikiEN-l] "the experiment" - did it work?

Wed Jun 21 17:01:37 UTC 2006

On 21/06/06, Erik Moeller <eloquence at gmail.com> wrote:
> In addition, a pseudonym is more anonymous than an IP address.

It's also less anonymous though; wherever a pseudonym is used, we know
it's the *same* people, where an anonymous IP address is used it could
be anyone behind a NAT'd gateway.

> Before,
> it was possible to tell that user X came from school network Y. Now
> you have to do a CheckUser to get the user's IP. The Seigenthaler
> vandal was identified because he was an unregistered user.

Right. IP addresses are very useful.

> GerardM also has some great ideas how we can build better
> authentication into our software, so that, for instance, we know that
> certain IP addresses are untrusted, and instead of blocking them
> entirely, we allow users who are authenticated _within_ a school or
> university to use that authentication in Wikipedia.

Yes, it doesn't make sense right now that a logged in user on a
anonymous IP address can't edit the wikipedia. Recently there was a
weird internet routing problem and I found that there was no direct
route to the wikipedia, but I could log in fine via a proxy server,
but couldn't edit. Given that the wikipedia knew it was me, that
behaviour is *broken*.

> Absolutely. My long-term vision of a replacement for both protection
> and semi-protection is "quality protection", where the version you see
> is the last reviewed one, but the article remains fully editable.
> Following this strategy, we can make Wikipedia ever more openly
> editable, continuing the path we have already taken.

I definitely agree, but I think the nupedia experiment tells us that
formal reviews are bad for the wikipedia though.

The problem is they never time out; you could be waiting for a formal
review forever- there's nothing forcing it through. There's no
pressure on the reviewers. I mean, if the wikipedia tells us anything
it's that if most edits to something improve an article, then the
article will improve over time; so the first version doesn't have to
be particularly good, because it will converge to excellence.

So we need to put pressure on the reviewers- how's this as an idea for
the most lightweight scheme for reviewing?:

http://en.wikipedia.org/wiki/Wikipedia:Timed_article_change_stabilisation_mechanism

(summary: the idea is that newbie edits don't go live for a day or so,
to allow more experienced editors to check it over.)

But beyond that, we need to come up with simple mechanisms to divide
trust between well established users; for example how can new users
get disadvantaged when creating a new article, but in a way that still
allows them to create articles?

If a new user (with say, less than 100 edits) creates an article,
should it go live immediately or should there be a cooling off period
of say, a week, to give the established editors a chance to check it
over? Just adding a delay discourages lots of bad guys I think; no
immediate gratification.

> Erik

-- 
-Ian Woollard

"Victory can be perceived but not created."