The interesting thing about Citizendium, Larry Sanger’s proposed fork of Wikipedia designed to add expert review, is how consistent Sanger has been about his beliefs over the last 5 years. I’ve been reviewing the literature from the dawn of Wikipedia, born from the failure of the process-laden and expert-driven Nupedia, and from then to now, Sanger’s published opinions seem based on three beliefs:
1. Experts are a special category of people, who can be readily recognized within their domains of expertise.
2. A process of open creation in which experts are deferred to as of right will be superior to one in which they are given no special treatment.
3. Once experts are identified, that deference will mainly be a product of moral suasion, and the only place authority will need to intrude are edge cases.
All three beliefs are false.
There are a number of structural issues with Citizendium, many related to the question of motivation on the part of the putative editors; these will probably prove quickly fatal. More interesting to me, though, is is the worldview behind Sanger’s attitude towards expertise, and why it is a bad fit for this kind of work. Reading the Citizendium manifesto, two things jump out: his faith in experts as a robust and largely context-free category of people, and his belief that authority can exist largely free of expensive enforcement. Sanger wants to believe that expertise can survive just fine outside institutional frameworks, and that Wikipedia is the anomaly. It can’t, and it isn’t.
Experts Don’t Exist Independent of Institutions
Sanger’s core animating belief seems to be a faith in experts. He took great care to invite experts to the Nupedia Advisory Board, and he has consistently lamented that Wikipedia offers no special prerogatives for expert review, and no special defenses against subsequent editing of material written by experts. Much of his writing, and the core of Citizendium, is based on assumptions about how experts should be involved in a project like this.
The problem Citizendium faces is that experts are social facts — society typically recognizes experts through some process of credentialling, such as the granting of degrees, professional certifications, or institutional engagement. We have a sense of what it means that someone is a doctor, a judge, an architect, or a priest, but these facts are only facts because we agree they are. If I say “I sentence you to 45 days in jail”, nothing happens. If a judge says “I sentence you to 45 days in jail”, in a court of law, dozens of people will make it their business to act on that imperative, from the bailiff to the warden to the prison guards. My words are the same as the judges, but the judge occupies a position of authority that gives his words an effect mine lack, an authority only exists because enough people agree that it does.
Sanger’s view seems to be that expertise is a quality like height — some people are obviously taller than others, and the rest of us have no problem recognizing who the tall people are. But expertise isn’t like that at all; it is in fact highly subject to shifts in context. A lawyer from New York can’t practice in California without passing the bar there. A surgeon from India can’t operate on a patient in the US without further certification. The UN representative from Yugoslavia went away when Yugoslavia did, and so on.
As a result, you cannot have expertise without institutional overhead, and institutional overhead is what stifled Nupedia, and what will stifle Citizendium. Sanger is aware of this challenge, and offers mollifying details:
[…]we will be posting a list of credentials suitable for editorship. (We have not constructed this list yet, but we will post a draft in the next few weeks. A Ph.D. will be neither necessary nor sufficient for editorship.) Contributors may then look at the list and make the judgment themselves whether, essentially, their CVs qualify them as editors. They may then go to the wiki, place a link to their CV on their user page, and declare themselves to be editors. Since this declaration must be made publicly on the wiki, and credentials must be verifiable online via links on user pages, it will be very easy for the community to spot false claims to editorship.
We will also no doubt need a process where people who do not have the credentials are allowed to become editors, and where (in unusual cases) people who have the credentials are removed as editors.
Sanger et al. set the bar for editorship, editors self-certify, then, in order to get around the problems this will create, there will be an additional certification and de-certification process internal to the site. On Citizendium, if you are competent but uncredentialed, you will have to be vetted before you are allowed to ascend to the editor’s chair, and if you are credentialed but incompetent, you’re in until decertification. And, critically, Sanger expects that decertification will only take place in unusual cases.
This is wrong; policing certification will be a common case, and a huge time-sink. If there is a value to being an expert, people will self-certify to get at that value, not matter what their credentials. The editor-in-chief will then have to spend considerable time monitoring that process, and most of that time will be spent fighting about edge cases.
Sanger himself experienced this in his fight with Cunctator at the dawn of Wikipedia; Cunc questioned Sanger’s authority, leading Sanger to defend it with increasing vigor. As Sanger said at the time “…in order to preserve my time and sanity, I have to act like an autocrat. In a way, I am being trained to act like an autocrat.” Sanger’s authority at Wikipedia required his demonstrating it, yet this very demonstration made his job harder, and ultimately untenable. This the common case; as any parent can tell you, exercise of presumptive authority creates the conditions under which it is tested. As a result, Citizendium will re-create the core failure of Nupedia, namely putting at the center of the effort a process whose maintenance takes more energy than can be mustered by a volunteer project.
“We’re a Warm And Fuzzy Hierarchy”: The Costs of Enforcement
In addition to his misplaced faith in the rugged condition of expertise, Sanger also underestimates the costs of setting up and then enforcing a process that divides experts from the rest of us. Curiously, this underestimation seems to be borne of a belief that most of the world shares his views on the appropriate deference to expertise:
Can you really expect headstrong Wikipedia types to work under the guidance of expert types in this way?
Probably not. But then, the Citizendium will not be Wikipedia. We do expect people who have proper respect for expertise, for knowledge hard gained, to love the opportunity to work alongside editors. Imagine yourself as a college student who had the opportunity to work alongside, and under the loose and gentle direction of, your professors. This isn’t going to be a top-down, command-and-control system. It is merely a sensible community: one where the people who have made it their life’s work to study certain areas are given a certain appropriate authority—without thereby converting the community into a traditional top-down academic editorial scheme.
Well, can you expect the experts to want to work “shoulder-to-shoulder” with nonexperts?
Yes, because some already do on Wikipedia. Furthermore, they will have an incentive to work in this project, because when it comes to content—i.e., what the experts really care about—they will be in charge.
These passages evince a wounded sense of purpose: Experts are real, and it is only sensible and proper that they be given an appropriate amount of authority. The totality of the normative view on display here is made more striking because Sanger never reveals the source of these judgments. “Sensible” according to whom? How much authority is “appropriate”? How much control is implied by being “in charge”, and what happens when that control is abused?
These responses are also mutually contradictory. Citizendium, the manifesto claims, will not be a traditional top-down academic scheme, but experts will be in charge of the content. The only way experts can be in charge without top-down imposition is if every participant internalizes respect for authority to the point that it is never challenged in the first place. One need allude only lightly to the history of social software since at least Communitree to note that this condition is vanishingly rare.
Citizendium is based less on a system of supportable governance than on the belief that such governance will not be necessary, except in rare cases. Real experts will self-certify; rank-and-file participants will be delighted to work alongside them; when disputes arise, the expert view will prevail; and all of this will proceed under a process that is lightweight and harmonious. All of this will come to naught when the citizens rankle at the reflexive deference to editors; in reaction, they will debauch self-certification (leading to irc-style chanop wars), contest expert preogatives, rasing the cost of review to unsupportable levels (Wikitorial, round II,) take to distributed protest (q.v. Hank the Angry Drunken Dwarf), or simply opt-out (Nupedia in a nutshell.)
The “U”-Curve of Organization and the Mechanisms of Deference
Sanger is an incrementalist, and assumes that the current institutional framework for credentialling experts and giving them authority can largely be preserved in a process that is open and communally supported. The problem with incrementalism is that the very costs of being an institution, with the significant overhead of process, creates a U curve — it’s good to be a functioning hierarchy, and its good to be a functioning community with a core group, but most of the hybrids are less fit than either of the end points.
The philosophical issue here is one of deference. Citizendium is intended to improve on Wikipedia by adding a mechanism for deference, but Wikipedia already has a mechanism for deference — survival of edits. I recently re-wrote the conceptual recipe for a Menger Sponge, and my edits have survived, so far. The community has deferred not to me, but to my contribution, and that deference is both negative (not edited so far) and provisional (can always be edited.)
Deference, on Citizendium will be for people, not contributions, and will rely on external credentials, a priori certification, and institutional enforcement. Deference, on Wikipedia, is for contributions, not people, and relies on behavior on Wikipedia itself, post hoc examination, and peer-review. Sanger believes that Wikipedia goes too far in its disrespect of experts; what killed Nupedia and will kill Citizendium is that they won’t go far enough.