Distributed wikis

Distributed wikisBrianna Laugher

Freedom in the Cloud miniconfLCA2011

[[SPEAKER:IS]]Wikipedian c. 2005-2010

[[TALK:IS]]Strictly vaporware

Philosophising

[[TALK:ISNOT]]A demo

Concerned with technical specs

Distribute/decentralise what?

Part Software Wiki

Interface Already existed with centralised VCS.

Barely exists, although possible. Vast majority of access via web UI.

Repository/storage

=DVCS meh

Access point Possible but not done, projects use official releases.

“marketplace of ideas” model

Community Kinda, like Linux? Not really

● Multiple versions of articles● Opposite of “One True Version”● Some mechanism allows the best to “rise to the top” (like PageRank?)

● Isn't that like the internet before Wikipedia? …● Similar to Knol? UrbanDictionary? StackOverflow?● Problems:

● rewards older contributions● evaluating is boring ● no canonical/reliable version● does not force/reward collaboration

“Marketplace of ideas” model

No more “One True Version”?

“A new-generation Wikipedia based on Git-style technologies could allow there to be not just one

Ocelot article per language, but an infinite number of them, each of which could be easily mixed and

merged into your own preferred version.”

– Anil Dash, “Forking is a Feature”http://dashes.com/anil/2010/09/forking-is-a-feature.html

● Wiki = VCS + prose text project + web UI.● Copyleft license => “right to fork” => “keeps the bastards honest”.● (Software) releases : (wiki) approved versions?● English Wikipedia is 10. Can it survive to 20?

● Too big to fail?● Too big to fork?

Some ideas

Wiki = web front-end for VCSfor prose text content

What's missing from wiki VCS?

● Diffs need to be per-word, not per-line ●Code contributions generally expected to be self-contained, generally in larger chunks than w/prose ●Code needs to be machine readable, (optionally?) human readable. Onus is on contributor to check machine readability

=> higher technical barrier to contributing is widely accepted

●Drive-by vandalism virtually non-existent● Prose projects rarely do “releases”

VCS for code vs prose

Merging for code vs prose

Code for unrelated technical functions should be able to be merged

Can we make the same promise for prose?

Can Wikipedia survive another 10?

Sense of dissatisfaction in the community

Unlike software, a certain critical mass is needed to stave off vandalism

Low barrier to entry(incl. anonymity)

+high visibility

+many pages

=>vandalism

Wikipedia the monopoly

● One destination – convenient and simple for users

● Great SEO (=> project growth)

● Potential for serendipity in editor activities

● Consistency (at least superficially)

● Practically, impossible to fork● hardware/bandwidth● community

● Widespread bureaucratese, instruction creep

● Impersonal

MediaWiki has a write API!

# Init site objectimport mwclientsite = mwclient.Site('commons.wikimedia.org')site.login(username, password) # Optional

# Edit pagepage = site.Pages['Commons:Sandbox']text = page.edit()print 'Text in sandbox:', text.encode('utf-8')page.save(text + u'\nExtra data', summary = 'Test

edit')

“Pending changes” aka “FlaggedRevs”

“Pending changes” separates

“what change I want to make”(a commit)

from“what I want users to receive”(tag as approved=”release”)

It's almost like having a release branch

mixed in with trunk....

Parsing mark-up :(

Templates :( :(

Free license + API – what's the hold-up?

“WikiProjects” FTW

● Self-organised groups of editors dedicated to a particular topic (e.g. Australia) or, less commonly, focus (e.g. standardising dates)● Very informal, light-weight● Narrower focus => better opportunity for community

Fork the UI?

(a la Twitter API)

“Why would anyone contribute to a feeder wiki?”

Promise of Wikipedia visibility+

Domain-specific and relevant interface+

Community

What is community?People

Intent/aims

Social norms- for interacting

- for contributing (eg. style guide)

License

Meta-planning for all of the above

In summary...

If forking Wikipedia is too hard, what can we do to make it practical again?

CreditsScreenshots and logos are © their respective ownersWikipedia 10: David Peters, CC-BY-SAhttp://commons.wikimedia.org/wiki/File:Wikipedia_10mark_rev_k.svg

Vandalism post-its: cdaltonrowe, CC-BYhttp://www.flickr.com/photos/30485180@N06/3490537301/

Coloured post-its: Michael Goodine, CC-BYhttp://www.flickr.com/photos/watchsmart/3227691975/

Branches: Piotrus, CC-BY-SAhttp://commons.wikimedia.org/wiki/File:Shaped_tree_branches_Tenerife.JPG

Wikimania schedule editing: Kat Walsh, CC-BY-SAhttp://commons.wikimedia.org/wiki/File:Wikimania2007_everythings_a_wiki.jpg

WikiProject Council logo: Neurolysis, CC-BY-SAhttp://commons.wikimedia.org/wiki/File:WikiProject_Council.svg

flagged revs maybe: Neurolysis/Kotra, CC-BY-SAhttp://commons.wikimedia.org/wiki/File:FlagRevsMaybe.png

pending changes clock logo: Adam Miller/Anomie/Dodoïste, CC-BY-SAhttp://commons.wikimedia.org/wiki/File:Pending_changes_clock.svg

brianna@modernthings.org

identi.ca/pfctdayelise

brianna.laugher.id.au

This work is © Brianna Laugher and licensed under the Creative Commons Attribution ShareAlike

license, except where otherwise noted.

Thanks!