[foaf-dev] announce: F2F MediaWiki extension - pre-alpha OpenID trust group RDFa exporter

==Background==

  * public facing, open Wikis are being spammed heavily. Information
in the wiki is either damaged by spam/malware attacks, or locked away
in read-only mode, making community contributions needlessly
difficult. Many such wikis exist and have overlapping communities of
contributors and potential contributors. Spam techniques are
increasingly subtle, and machine filtering is proving inadequate. It
is difficult for admins of a wiki to know who they can trust with
write access, if they want to maximise contributions to their content
and at the same time, take care to protect those contributions from
vandalism. If a wiki is left open to the public, it will be
vandalised. If it is made editable only by a small group, it will
often fail to reach its full potential. This is unfortunate.

 * This situation causes a trend towards centralisation - both in
terms of culture/content (people simply use Wikipedia and other large
sites), but also in terms of identity technology ('comment with your
twitter / facebook id', etc.) which can exclude users who choose not
to use the market-leader sites. Both aspects threaten the diversity
and pluralism that make the Web interesting.

 * Many users increasingly have access to openids (and via eg.
openid4.me, from webid too).

 * Recent MediaWiki releases makes it easier for wiki pages to be
automatically generated with machine-readable markup RDFa).

==Overview==

 * This is an experiment in sharing trust-related metadata about
people's login-capable Web identifiers - homepages, blogs etc that are
wired up to OpenID.
 * Wikis and blogs can both produce and consume this kind of data.
Every time you accept a comment into your blog or wiki, and that
comment was made by OpenID, you are accumulating evidence that people
can use when deciding whether to trust that OpenID elsewhere.
 * F2F exporters are simple addons for blogs and wikis that (with some
tweaking/customisation) can publish descriptions of the groups of
openids known to some software installation

==Status==

Previously...

* from 2009 http://danbri.org/words/2009/10/25/504 'Syndicating trust?
Mediawiki, Wordpress and OpenID' initial writeup
* http://danbri.org/words/2009/12/29/523 'WordPress trust syndication
revisited: F2F plugin'
* Mediawiki notes http://wiki.foaf-project.org/w/DanBri/MediawikiOpenid

There is already a basic exporter for Wordpress. Today I got same
working for MediaWiki (thanks to mhausenblas, presbrey in irc for
help!)

This is pretty rough still, but I hope it gives others some basic
pieces to build upon.

Sample installation: http://wiki.foaf-project.org/w/Special:F2F (gives
an RDFa page describing some trusted openids)

Source code: check out F2F/ from
http://svn.foaf-project.org/foaftown/2010/mediawiki/extensions/F2F/
subversion into your MediaWiki extensions/ dir.

See http://svn.foaf-project.org/foaftown/2010/mediawiki/extensions/F2F/F2F_body.php
for the SQL, which needs tweaking/customising. Please read the
code before installing this in a live site, it is very much
experimental work in progress. I won't list all limitations here; you
can see from the code link that there is not a lot to this yet, beyond
the raw idea. In particular it doesn't distinguish between the groups
some account/openid is in, nor provide any information for consuming
data.

So anyway - the idea is that if we can get a few such installations
running, they can syndicate useful/trusted openid lists back into the
wider community.

For example, the FOAF wiki already takes a feed of openids from my
blog, so that commenters accepted as non-spammers in the blog are put
into a group in the mediawiki. Since the FOAF wiki edit list is now
locked down to known/trusted openids only, I am interested in finding
ways of broadening that list without too much manual effort. I am sure
other Wiki admins are in a similar situation.

Research topic: how to avoid loops in the trust graph :) If
wiki.foaf-project.org and (say) a Dublin Core wiki both produce and
consume these lists, how to distinguish in the data which groups are
populated based on external evidence, and which from first-hand
knowledge?

thanks for any thoughts,

Dan