Toward a New Compendium of Knowledge (shorter version)

September 15, 2006 (and getting slowly out of date!)

Larry Sanger

This shorter version was delivered at the Wizards of OS conference.  It might be more readable than the longer version, but the longer version has more details.


1. Thinkers of the world, start imagining.

There are tens of millions of intellectuals online today.  What is possible with tens of millions of intellectuals working together on educational and reference projects?   The very thought of that makes me literally quiver with excitement.  I am amazed that we, educated people throughout the world, have barely begun to imagine what new reference and educational materials could come into being, if we pool our efforts in the open, collaborative ways demonstrated by open source software hackers.  Even less have we begun to take such possibilities really seriously, or actually get to work on them.


2. Wikipedia.

We may take Wikipedia as an early prototype of the application of open source hacker principles to content rather than code.

Wikipedia, started only five years ago, now has millions of articles in over one hundred languages, and has nearly singlehandedly introduced the world's intellectuals to the possibilities of enormous collaborative efforts.  By giving intellectuals the world over an open platform on which to work together Wikipedia has shown a global audience what enormous, distributed knowledge collaboration can achieve.  The work of the Wikipedians has astounded the world.

I always have been an enormous fan of Wikipedia, and I still am.  But I now want to help launch something better, if that's possible.  Let me explain why I am doing so, and then I will describe the project itself.

When I first started working on Wikipedia, I was also employed as editor-in-chief of Nupedia, a more straight-laced, peer-reviewed free encyclopedia project.  Wikipedia was to be a free-wheeling, fun side project, a dynamic source of content, for the more serious Nupedia project.  Wikipedia was closer in spirit to the old open source, hacker ways, where Nupedia was essentially an academic project.  But together, Nupedia and Wikipedia were going to be an "unstoppable high-quality article-creation juggernaut," or so I said in 2001.

I think there are many misconceptions about Nupedia and why it failed, but I don't have time to go into that here.  But it did wither and finally die due to neglect.  The result was that only half of my own original conception of "the finest encyclopedia in the history of humankind"--the wild and anarchical half--was preserved.

Not surprisingly, with only one-half of the original design, some problems with Wikipedia--problems that continue to this day--emerged.  Here are some of them:

  • The community does not enforce its own rules effectively or consistently.  Consequently, administrators and ordinary participants alike are able essentially to act abusively with impunity, which begets a never-ending cycle of abuse.
  • Widespread anonymity leads to a distinguishable problem, namely, the attractiveness of the project to people who merely want to cause trouble, or who want to undermine the project, or who want to change it into something that it is avowedly not--in other words, the troll problem.
  • Many now complain that the leaders of the community have become insular: it has become increasingly difficult for people who are not already part of the community to get fully on board, regardless of their ability or qualifications.
  • This arguably dysfunctional community is extremely off-putting to some of the most potentially valuable contributors, namely, academics.  Furthermore, there is no special place for academics, so that they can contribute in a way they feel comfortable with.  As a result, it seems likely that the project will never escape its amateurism.  Indeed, one might say that Wikipedia is committed to amateurism.  In an encyclopedia, there's something wrong with that.

Can Wikipedia recover from these problems?  The first step to recovery is admitting that you have a problem in the first place.  Wikipedia's most passionate defenders, if they react at all, will probably do nothing but explain why I am mistaken in each of these criticisms.  I know of course that there are movements afoot to reform Wikipedia in various ways.  But I see little evidence that the community, whatever its discontents, will go so far as to admit the problems I've listed.

The failure to recognize these serious problems is a reflection of the fact that, at bottom, they are political problems.  Like all open communities online, Wikipedia's community is self-selecting, and its policies have determined who stays and who leaves (or is driven away).  For this reason, online communities tend to become rather conservative in their attitudes toward their own systems, and Wikipedia is certainly no different.


3. A new community and a new project: the Citizendium.

[Note, January 2007: while we are still full steam ahead with a fork of Wikipedia's processes, we are having second thoughts about forking Wikipedia's articles.  We've decided as an experiment actually to delete all the unedited Wikipedia articles from our database, to encourage people to start new articles altogether.  We think this might encourage even more activity than we've had--though we're pleased with the amount of activity we've had so far--and help us to create a more distinctive Citizendium culture.]

Often, if you wish to make any very important changes to an open source or open content project that has an entrenched community, the only way to do it is to start a new community.  And that is what I propose.

I propose a fork of Wikipedia to be called the Citizendium, that is, the Citizens' Compendium.  Legally, of course, a fork is clearly permitted.  I think it is also morally permitted--perhaps even morally recommended--if there is a chance of retaining Wikipedia's virtues while eliminating the problems I just mentioned.  It will be a progressive fork, that is, all the articles will be imported at once, but individual articles will keep getting refreshed with Wikipedia's latest changes until they are changed in the Citizendium.

Obviously, you want to know how the Citizendium editorial system will differ from Wikipedia's system.  There will be two main areas of dissimilarity: the involvement of experts, and the creation of a more mature community.  There will be other differences as well.

Before I elaborate, however, let me assure you that I do not think that I necessarily have all of these details right.  Obviously, I am fallible and, like all of us, my understanding is limited.  Clearly, this project needs as much intelligent input as it can get.

First, here are some details of the editorial system I propose: 

  • We will have a new role in the system: that of editor.  Others will be called authors.  Generally, authors will defer to editors when editors are speaking about their areas of specialization.  When authors get into a dispute, they may work out a compromise, or they may consult an editor.  Editors' decisions will be logged in a new, standard part of each article's discussion page.
  • For the most part, editors will work in the system just as everyone else does, "shoulder-to-shoulder" with ordinary authors.  Editors will not be able to direct work in a top-down fashion, or to "squat" on articles and prevent others from making any changes.  Editors who are not comfortable with this arrangement should not participate.  Those who attempt to make articles their personal bailiwicks, shooing off everyone else, will be ejected from the project, no matter what their qualifications.
  • Editors will be self-selecting, in a certain way.  There will be no editor selection process.  Rather, editors will be invited to come to the website and simply declare themselves to be editors, if they meet certain benchmark requirements--the same straight-up credentials that the offline world relies on.  Editors will be required to state their credentials on their user pages, for everyone to examine.
  • Editors will have the right to place articles in an "approved" category.  Other (qualified) editors may remove articles from the "approved" category; disputes are to be settled by discipline-oriented editorial workgroups.

Second, some differences in the community:

  • There will be no logged-out editing and no anonymous editing.  Anyone may participate, but all must be logged in under their own real names (we will use the honor principle to begin with), and with a working e-mail address.  Where Wikipedia shares the culture of anonymity found in the broader Internet, the Citizendium will have a culture of real-world, personal responsibility.
  • The community will launch with a charter that articulates the project's goals, fundamental policies, governance, and the rights of participants and of the public.  As with most charters, the Citizendium community charter will be relatively brief and vague, difficult to change, and regarded as "binding law" of the project.
  • All contributors will, as a condition of their participation, be expected to support the community charter.  If they cannot do this, they are not welcome to participate and may be ejected from the project.
  • The charter and rules will be enforced by "constables."  In time, an effective and fair "legal" system will be established.

Third, over the years, Wikipedia has suffered some feature creep.  The Citizendium will be deliberately simplified and kept simple.  In addition, copyright and libel abuses will be handled quite differently.  And finally, the Citizendium won't officially call itself an encyclopedia.  We might call it an experimental workspace.

So much for the differences, and as I hope you can see, they are significant.  But it is important to note that the system will still work very much the way Wikipedia does, in many respects. 

Areas of similarity include:

  • It's more or less, kind of like, an encyclopedia.  It's a wiki that aspires to be as good as a real encyclopedia.
  • It's open to virtually everyone.  Virtually anyone can come to the website and, within a few minutes, be working on an article.  It is not Expertpedia.
  • It will use the same license, the GFDL.
  • There will be no advertisements.  There may be unobtrusive non-profit sponsorship statements.
  • It will come under the control of a non-profit foundation.
  • The neutrality policy will be virtually the same, and the rule against original research will be virtually the same.
  • The MediaWiki software will be the same (though some settings will be different) and there are no plans to fork it.


4. The way forward.

[Note, Jan. 2007: this section is now old news entirely and may be safely skipped.]

Finally, I want to discuss the way forward--in other words, how to get the project started.

Whatever else we do, we should have the servers and the wiki software set up, and available at least to some beta testers, as soon as possible.

While waiting for the wiki to be set up, what can the ordinary rank-and-file future authors or editors do?  Three things: first, join the project forums or a project mailing list; second, contribute your thoughts to the discussion; and, third, wait for the announcement that the wiki is ready to edit.  I want to start the project with a bang.  I would like to strain the limits of whatever server configuration we put in place.

So much for the short term.  As to the medium term, over the coming weeks, I hope the Citizendium project will be contacted by individuals as well as universities, foundations, and companies.  My personal belief, which I do not hold very strongly, is that the latent interest in and support for this project will allow us to collect a truly stellar group of advisors and partner institutions; if so, this experiment might succeed brilliantly.  As to what we need, precisely, please see the project FAQ, which is now online, for some ideas.

Once the wiki has launched, we should begin to form discipline-specific editorial workgroups.  Also once the wiki has launched, I think we should immediately hold a series of monthly face-to-face meetings in about a half-dozen major cities.  This would lead up to a big meeting about the project charter.

I am making this announcement in Germany in part because of my personal commitment to making this project fully international as quickly as is feasible and responsible to do.  On the other hand, I do not want to presume to speak for the interested German-speaking community online.  Knowing as you now do that I personally am getting behind a fork of the English language Wikipedia, I put the question to you: should we also make a similar fork of the German language Wikipedia?  That's something to talk about in the "beer-to-beer networking" that will be going on this evening, maybe.

While this organizational work is going on, I wish to put together the finest possible set of advisors for this project.  Their role as advisors will give them a special platform from which to advise the project and speak on its behalf.  Moreover, it is this group of people who will meet at the constitutional convention to ratify the project's charter.

So, let us create a vibrant yet responsible new community.  Together we will, I think, have an amazing journey.

For more information on the Citizendium project, see the website.