[WikiEN-l] Tech proposal for 3-revert rule

[Jimmy Wales]

I have an idea for an abstract generalization of the ban/throttle/
page protection concepts, unifying them in a single table with great
flexibility.  Details at the bottom, but first some quick motivation.

We have a proposal with some support, but also some strong opposition,
for a rule that says that sysops can 24-hour ban people who break the
(widely supported) 3 revert guideline.

I think it's safe to say we have strong consensus that the 3 revert
guideline is a good guideline.  What we don't have yet is consensus
about what should be done about it.

My problem with it, and I think that this would be echoed by many who
have reservations, is that a 24 hour ban is sort of a big hammer to
use in cases like this.

I have some questions about the technical feasibility of some
smaller-hammer ideas.

1.  per-article bans -- this could be very useful in a number of
cases, particularly now that we have an arbitration committee to
divide up the work of studying such things.

2.  per-article, per-person throttles -- this would be the technical
application of the 3 revert rule, applied to individual people and
individual articles.  Basically, a sysop could put Wik (for example)
into throttle-mode for a given article where he's committed a 3 revert
violation, and after that Wik can only edit that article 3 times per
day.  This might push him to be more co-operative.

Or consider the DNA case.  Pretty much all parties to the dispute,
even those who were not guilty of anything bad could be placed,
without stigma of "punishment", into throttle mode for that article.

3. per-article throttles -- for particular articles, instead of
"protecting" it so that no one can edit, we could "throttle" it so
that each person may edit it only 1 time per day.  The idea is to slow
down an edit war to allow people to calm down, and to keep everyone
else from being bothered by an ongoing fiasco.

---

All of these seem fairly technically do-able to me.  I'm especially
keen on the latter two.  (And suspect that the first one could be
implemented as a form of the latter two, with edit_count set to zero.)

Here's a table to show what I mean:

Jimbo Wales      -   Oregon    -     3 # Jimbo throttled on this article
Jimbo Wales      -   Cats      -     0 # Jimbo per-article banned
*                -   DNA       -     1 # everyone throttled here
*                -   Judaism   -     0 # same as protected
Wik              -   *         -     3 # Wik throttled everywhere
Plautus          -   *         -     0 # Plautus banned

Then, to find out if a user has permission to edit a given article, we
just look them and the article up in the table, and see what the
number is.  If it's not there, then all is well, if it is there, then
we have to make an additional check in the history table to see what's
happened in the past 24 hours.

Obviously, it's easy for me to say that this is easy.