[Wikipedia-l] Semi-protection

Wed Nov 15 04:31:20 UTC 2006

On 11/13/06, Gregory Maxwell <gmaxwell at gmail.com> wrote:
> The ability to edit competently, accurately, and in conformity with
> our standards (NPOV, BLP, TDMTLAOMG, etc) is a totally orthogonal
> matter with a persons interest in editing with good intentions.  Some
> of the most harmful problem material which I've seen inserted into
> Wikipedia was inserted with good intentions.

Certainly. So in addition to "Assuming Good Faith", our system
currently "Assumes Positive Contributions". This, in many cases, is a
fair assumption -- but in many other cases it is not. But whether a
contribution is likely to be positive or not is not only dependent on
the user in question, but also on the article. Pop culture articles
attract different editors than, say, those on mathematics; highly
politicized topics get different contributions than very sterile ones.

Semi-protection of selected pages is one way to deal with this problem
while respecting the distinction between different types of articles
-- we identify those articles where problems occur frequently, and
disable newbie editing for them altogether. The proposal made here, to
generally vet newbie contributions _instead_ of semi-protecting
articles, eliminates the distinction between the _types_ of articles
which we are talking about, which is a highly significant one. As a
consequence, it would massively alter the experience of all new users.
Such a drastic step is not warranted before trying more gradual steps
of quality assurance.

Accordingly, my proposal is to instead build QA and edit approval
functionality _on top_ of the existing protection model. Instead of
vetting all newbie edits, we should vet those to selected articles --
and in turn, we can make the number of those articles greater than the
current number of semi-protected pages. On the other hand, pages
which, on average, receive contributions that are in good faith and
improve the articles would continue to operate as present, but
_indicate_ if a quality-reviewed prior revision is available. See:

http://meta.wikimedia.org/wiki/User:Eloquence/WikiQA

So, if someone makes an edit to [[Euclidean geometry]], it's going to
be immediately visible just like it is today, whether the user is a
newbie or not--but the page makes it much clearer to the reader that
the article they look at has not been reviewed yet, and whether a
reviewed version is available. If they make an edit to [[George W.
Bush]], it's going to have to be approved by a regular user first.

As a next step, we could also try to identify users from IP ranges
which are particularly likely to make negative contributions, and put
them into the "to-be-reviewed" bin by default. But we should only do
so if it is likely to significantly benefit our readers without
significantly impairing our editors.

> Without seeing some data on the current distribution of semiprotection
> lifetimes, I couldn't comment on the usefulness of automatic
> de(semi-)protection.  I looked at protection lifetimes some time ago
> (back when folks like Tony Sidaway were around and diligent with
> removing old protections) and we looked like we were doing pretty
> well.

Do a cross-language comparison -- last time I did so, there were vast
difference across languages. Particularly, those languages where
(semi-)protected pages are not visibly tagged seem to do poorer at
removing the protection. en.wp seems to do reasonably well in general,
though protections certainly often stay longer than they have to.
-- 
Peace & Love,
Erik

Member, Wikimedia Foundation Board of Trustees

DISCLAIMER: Unless otherwise stated, all views or opinions expressed
in this message are solely my own and do not represent an official
position of the Wikimedia Foundation or its Board of Trustees.