User:WereSpielChequers/forking

(forking) This is an essay. It expresses the opinions and ideas of some Wikimedians but may not have wide support. This is not policy on Meta, but it may be a policy or guideline on other Wikimedia projects. Feel free to update this page as needed, or use the discussion page to propose major changes.

Forking is the division of a community when one part of the community building an open source project decides to take a copy of the project and edit on another site in rivalry to the original, usually with different rules. The right to fork is a feature of any project building content licensed under a license that allows people to copy, use and modify that content.

The right to fork is fundamental to Wikimedia projects. In a situation where one project has just forked and others are considering it I can see the temptation for the Foundation to respond by making forking more difficult. But a more healthy response would be to make forking easy, but respond to each fork by working out what went wrong and trying to learn from the experience. As well of course as not annoying editors so much that they feel the need to fork.

External forks that the community has experienced include the Spanish Wikipedia forking because of the threat of advertising and the English Wikinews forking because most articles were out of date before they were reviewed. Arguably several wikis on Wikia and elsewhere have been created in response to Wikipedia's notability criteria and no original research rules which limited those who wanted to write in great detail about particular Science Fiction settings and other topics. Citizendium could be considered a Wikipedia fork that aimed to be closer to Nupedia and whose failure reaffirmed the Wikipedia model over the Nupedia one. Conservapedia and possibly some of the other POV based pedias are arguably forks of Wikipedia, there hasn't yet been a successful pedia that has replaced Neutral point of view with an ideological point of view, partly because once you explicitly forsake neutrality there is a drift to fundamentalism, and it can be very difficult to differentiate between a more fundamentalist POV and parody.

Internal forks are arguably far more numerous, though not always considered forks. Wiktionary was created in part because people wanted to write dictionary definitions in Wikipedia, hundreds of different language versions of Wikipedia were created partly because people started writing wikipedia articles in those languages. Commons as a shared resource for images, wikinews a news service. The Wikimedia movement within a decade of the start of Wikipedia already contains a thousand wikis. In theory these are not quite forks because their remits should each be unique and material posted to one that belongs on another will in theory eventually get deleted or transwikied. But the boundaries are sometimes vague, English and Italian wikipedias both contain large numbers of images that for copyright reason may not be loaded to commons, as for the role of a breaking news item as opposed to the more widely read encyclopaedia article on the same topic.... Arguably the most famous fork in the history of the project was the creation of a less formal entry level wiki for Nupedia in the hope that maybe the crowd would write a few encyclopaedia articles that could be fed into Nupedia.

The drawback of forks is that in splitting down one divide you split across other interests. I'm a pedant who fixes typos, on the English language wikipedia I've fixed thousands of them, on my visits to other English language projects I've fixed typos on several other projects. But those visits are rare and didn't start until I'd been at Wikipedia for some time, and the tools I have for typo hunting only run on EN wiki. Similarly I'm interested in Easter Island and have half a shelf of books on the topic. But the only places where I've done any Easter island related edits are EN wiki, EN wikisource and maybe commons. In traditional printing and offline knowledge work it has been conventional and may even by logical to have completely different publications for dictionaries, encyclopaedias and newspapers. I'm not convinced that the same rationale applies online, especially when you could have one wiki with shared userspace, template space and file space but also have a wiktionary, Wikipedia, wikinews and wikisource each as a name space within the same wiki. So I'm an advocate for the idea that each language in which we operate should have just one wiki, with one common wiki to handle the meta coordination, outreach strategy and common resources that all languages need to access. It is a truism of w:Knowledge management (WikiWikiWeb) that an organisation should only have one system, to subdivide means inefficiency, and in the case of a crowd based site such as wikimedia the wisdom of crowds is eroded by having an ever more subdivided crowd.

The advantage of forks is that as with divorce sometimes an amicable parting of the ways is better than continued bickering. The idea being that it is better to have two smaller communities each focussed on doing something that may be very similar, instead of one fractious community focussed on an internal debate as to what they should be doing. There may also be a maximum community size above which collaboration becomes difficult and the ratio of "acquaintances" to "strangers" reduces people's mutual tolerance in a similar manner to motorists in London as opposed to a small town.

Potential Forks

There are a number of potential forks that may unfold in the near future

1. Inclusionism/Deletionism On EN wiki there has been a huge amount of debate and quite a bit of rule stretching between those who would like to move the project towards stricter standards of notability and sourcing, and those who would support policies of only reverting edits as "unsourced " if they are contentious.
2. Civility. Currently on the English Wikipedia there is a deep divide within the community as to whether being a valued contributor of good content should excuse you from general rules on civility. Some go as far as to consider that certain stellar contributors should be allowed to make personal attacks on lesser editors.
3. Porn and censorship. The Foundation is contemplating an image filter which depending on the way it is implemented may divide the community in a similar manner to the fork of the Spanish Wikipedia, with the Germans and perhaps some other projects leaving the movement.
4. Children's pedia. A perennial proposal and one of the counter proposals in the censorship debate is that we could and perhaps should create a children's version of Wikipedia.

Existing forks that may change

Forks by product

As the community ages and dwindles so the wisdom of subdividing the crowd into ever smaller crowds becomes more open to challenge. Equally as the number of language with Wikimedia projects grows so the argument for language level wikis is strengthened, if only because a language with fewer than 150 active wikimedians is probably more viable as one community. Vandalism and personal attacks require the same admin skills and tools whether they occur in the wiktionary, Wikisource or wikinews of the Masai language wiki.

Forks by language

Translation software has steadily been improving and becoming more commonplace. In recent years search engines have started to offer a free translation screen, and many readers of the English language wikipedia are apparently using it already. The current quality of automated translation is fairly poor, however we have the ingredients to make the next stepchange in that quality. Translation Software commonly works by producing the most likely translation of a particular piece of text, it doesn't say this sentence could mean x, y or z depending on these two ambiguous words. But with the ability to suggest the most likely meaning of a text comes the ability to produce lists of anomalies and ambiguities. If Wikimedia was to collaborate with one of the translation outfits it would be possible to identify the hundreds of thousands of anomalies that make the translation of a project such as the English wikipedia problematic. Currently some such anomalies would be fixable, and probably quite rapidly if combined with AWB. Others would require us to change the structure of the pedia so that when someone used a word such as bonnet wikipedia then recorded whether this was bonnet as in the chili, the hat or the part of a car. By Wikipedia standards this would be a major project, and it might take a couple of years to implement. But once it was done then even those who only speak languages in which we have quite small Wikipedias would be able to access several rival versions of Wikipedia. The step beyond that would be to enable anyone to edit in their own language, with the edited version stored in the core language of that version of wikipedia.

If we move in this direction then different language versions of Wikipedia will become competitors as translation software improves. Strange alliances may emerge as advocates of the survival of various minor languages see their future in getting a high quality translation route to more common languages. Already we have significant differences between different language versions of Wikipedia, some are much stricter than others on requiring reliable sources to be cited by editors; Some are stricter on vandalism with German and several other languages having introduced Flagged revisions so that every edit will be checked at least for obvious vandalism. Others like the English Wikipedia have chosen to be more tolerant of vandalism in order to maintain the promise that anyone can edit and have their edit immediately go live.

The next eleven years could see a whole family of very different Wikipedias emerge, ostensibly and technically with the difference being in the core language in which they store pages. But as the language become merely a display choice, the key differences may be over such things as inclusionism, quality of referencing, community culture, tolerance of controversial content and perhaps even a scientific POV wikipedia emerging to compete with the NPOV ones.

If the Wikimedia Foundation remains above the fray and acts as an enabler then this could be allowed to develop in a relatively painless way. Servers and bandwidth are getting cheaper each year and while a translateable core language with ambiguous words tagged to their correct wiktionary meaning would require more resource than current wikis, the difference is probably minor compared to the price performance improvements that computer technology will deliver in the next decade.