[Foundation-l] We need to make it easy to fork and leave

[Tom Morris]

On Mon, Aug 15, 2011 at 08:26, Nikola Smolenski <smolensk at eunet.rs> wrote:
> On 15/08/11 08:16, David Richfield wrote:
>> It's not just financial collapse.  When Sun was acquired by Oracle and
>> they started messing about with OpenOffice, it was not hard to fork
>> the project - take the codebase and run with it.  It's not that easy
>> for Wikipedia, and we want to make sure that it remains doable, or
>> else the Foundation has too much power over the content community.
>
> I'm fairly confident it would be much easier to fork Wikipedia than
> OpenOffice.
>

Technically, it's much easier to fork code than it is to fork wikis
especially now in an era of distributed version control systems (Git,
hg, bzr) where everyone who checks the code out of a repository has a
full copy of the repository. The only technical infrastructure you
need is some hosting space for the repo and the other common bits you
need for software dev (mailing list, bug tracker etc.)

One thing I've been thinking about from the failure of Citizendium is
how an expert community could set up their own external version of
pending changes: basically a simple database of stable versions, so
any individual or group could set up a server with stable versions of
articles, then you could subscribe to a set of stable version sets -
so, say, the International Astronomical Union mark a bunch of
revisions of astronomy articles as stable, and if you've got the
browser plugin installed with their dataset installed, when you visit
one of those pages, it'd show you the stable version they chose. And
the flipside is that if you are (in my humble opinion) a cold fusion
nut or a homeopathy nut, you could find some crazy person who believes
in those things to come up with his or her own set of crank stable
versions.

And the stable version could be marked as checked by a particular
person from a particular institution with their real name if that is
the practice in that community: perhaps in physics or philosophy or
psychology or some other academic subject, having a real name person
sign off on a particular stable version is fine and dandy, but in,
say, the Pokémon fan community, they don't really have the same
assumptions. (Again, one of the failures of Citizendium: you don't
need a guy with a Ph.D to approve the articles on Pokémon in the way
you might want a credentialed expert to sign off on, say, an article
on cancer treatment.)

The essential thing is to separate out the things that people want:
some people want "distributed Wikipedia", but why? Well, one good
reason seems to be so you can have stable versions with expert
oversight (like Citizendium) - well you can get most of the desiderata
that led to Citizendium by having a third-party distributed approval
layer and browser plugins etc. A little bit of hacking provides a lot
of opportunity for different communities to take Wikipedia and run
with it in the ways they want to. This kind of proposal would provide
a lot of what Citizendium was shooting for but without the
coordination problem of trying to get disparate communities of people
to work together in a way the CZ community kind of failed to do.
Consider for instance the ethnic studies/women's studies people who
didn't find Citizendium a welcoming environment.[1] Under this kind of
proposal, if there is a community of people involved in ethnic studies
who want to participate in Citizendium-style expert approval, they can
set up some very lightweight software and organise their approvals in
whatever way fits best with their academic community norms.

Essentially, in software terms, this would be like a 'packager',
someone who takes Wikipedia's output on a certain topic and marks
specific revisions or whatever as good or bad. They'd still be welcome
(and indeed encouraged) to participate in editing on Wikipedia in the
traditional way, and ideally the community wouldn't take participation
in such an enterprise against them as an editor (just as they
currently don't or shouldn't take participating in Wikinfo or
Citizendium or even Conservapedia against someone), and any comments
that come up in the 'packaging' process could be taken as feedback in
the normal way just as if packager at Debian finds a bug with a piece
of software, he or she can point that out the upstream maintainer.

Feedback?

[1] see http://cryptome.info/citizendium.htm and
http://rationalwiki.org/wiki/Citizendium

-- 
Tom Morris
<http://tommorris.org/>