Clay Shirky
( Archive | Home )

Liz Lawley
( Archive | Home )

Ross Mayfield
( Archive | Home )

Sébastien Paquet
( Archive | Home )

David Weinberger
( Archive | Home )

danah boyd
( Archive | Home )

Guest Authors
Recent Comments

pet rescue saga cheats level 42 on My book. Let me show you it.

Affenspiele on My book. Let me show you it.

Affenspiele on My book. Let me Amazon show you it.

Donte on My book. Let me show you it.

telecharger subway surfers on My book. Let me show you it.

Ask Fm Anonymous Finder on My book. Let me show you it.

Site Search
Monthly Archives
RSS 1.0
RSS 2.0
In the Pipeline: Don't miss Derek Lowe's excellent commentary on drug discovery and the pharma industry in general at In the Pipeline


« social network sites: my definition | Main | Friends, Friendsters, and Top 8: Writing community into being on social network sites »

November 20, 2006

Social Facts, Expertise, Citizendium, and Carr

Email This Entry

Posted by Clay Shirky

I want to offer a less telegraphic account of the relationship between expertise, credentials, and authority than I did in Larry Sanger, Citizendium, and the Problem of Expertise, and then say why I think the cost of coordination in the age of social software favors Wikipedia over Citizendium, and over traditionally authoritative efforts such as Britannica.

Make a pot of coffee; this is going to be long, and boring.

Those of us who write about Wikipedia, both pro and con, often mix two different views: descriptive — Wikipedia is/is not succeeding — and judgmental — Wikipedia is/is not good. (For the record, my view is that Wikipedia is a success, and that society is better off with Wikipedia than it would be without it.) What I love about the Citizendium proposal is that, by proposing a fusion of collaborative construction and expert authority, it presses people who dislike or mistrust Wikipedia to say whether they think that the wiki form of communal production can be improved, or is per se bad.

Nicholas Carr, in What will kill Citizendium, came out in the latter camp. Explaining why he thinks Ctizendium is a bad idea, he offers his prescription for the right way to do things: “[…] you keep the crowd out of it and, in essence, create a traditional encyclopedia.” No need for that ‘in essence’ there. The presence of the crowd is what distinguishes wiki production; this is a defense of the current construction of authority, suggesting that the traditional mechanism for creating encyclopedias is the correct one, and alternate forms of construction are not.

This is certainly a coherent point of view, but one that I believe will fail in practical terms, because it is uneconomical. (Carr, in his darker moments, seems to believe something similar, but laments what the economics of peer production mean. This is a “Wikipedia is succeeding/is not good” argument.) In particular, I believe that the costs of nominating and then deferring to experts will make Citizendium underperform its competition, relative to the costs of merely involving experts as ordinary participants, as Wikipedia does.

Expertise, Credentials, and Authority

First, let me say that I am a realist, which is to say that I believe in a reality that is not socially constructed. The materials that make up my apartment, wood and stone and so on, actually exist, and are independent of any observer. A real tree that falls in a real forest displaces real air, even if no one is there to interpret that as sound.

I also believe in social facts, things that are true because everyone agrees they are true. My apartment itself is made of real stuff, but its my-ness is built on agreements: my landlady leases it to me, that lease is predicated on her ownership, that ownership is recognized by the city of New York, and so on. Social facts are no less real than non-social facts — my apartment is actually my apartment, my wife is my wife, my job is my job — they are just real for different reasons.

If everyone stopped agreeing that my job was my job (I quit or was fired, say), I could still walk down to NYU and draw network diagrams on a whiteboard at 1pm on a Tuesday, but no one would come to listen, because my ramblings wouldn’t be part of a class anymore. I wouldn’t be faculty; I’d be an interloper. Same physical facts — same elevator and room and white board and even the same person — but different social facts.

Some facts are social, some are not. I believe that Sanger, Carr and I all agree that expertise is not a social fact. As Carr says ‘An architect does not achieve expertise through some arbitrary social process of “credentialing.” He gains expertise through a program of study and apprenticeship in which he masters an array of facts and techniques drawn from such domains as mathematics, physics, and engineering.’ I agree with that, and amended my earlier sloppiness in distinguishing between having expertise and being an expert, after being properly called on it by Eric Finchley in the comments.

However, though Carr’s description is accurate, is it incomplete: an architect does not achieve expertise through credentialing, but an architect does not become an architect through expertise either. An architect is someone with expertise who has also been granted an architect’s credentials. These credentials are ideally granted on proof of the kinds of antecedents that indicate expertise — in the case of architects, relevant study (itself certified with the social fact of a degree) and significant professional work.

Consider the following case: a young designer with an architect’s degree designs a building, and a credentialed architect working at the same firm then affixes her stamp to the drawings. The presence of the stamp means that a contractor can use the drawings to do certain kinds of work; without it the drawings shouldn’t be used for such things. Both the expertise and the credentials are necessary to make a set of drawings usable, but in this fairly common scenario, the expertise and the credentials are held by different people.

This system is designed to produce enough liability for architects that they will supervise the uncredentialed; if they fail to, their own credentials will be taken away. Now consider a disbarred architect (or lawyer or doctor.) There has been no change in their expertise, but a great change in their credentials. Most of the time, we can take the link between authority, credentials, and expertise for granted (its why we have credentials, in fact), but in edge cases, we can see them as separate things.

The clarity to be gotten from all this definition is a bit of a damp squib: Carr and I are in large agreement about the Citizendium proposal. He thinks that conferring authority is the hard challenge for Citizendium; I think that conferring authority is the hard challenge for Citizendium. He thinks that the openness of a wiki is incompatible with Citizendium’s proposed form of conferring authority, as do I. And we both believe this weakness will be fatal.

Where we disagree is in what this means for society.

The Cost of Credentials

Lying on a bed in an emergency room, you think “Oh good, here comes the doctor.” Your relief comes in part because the doctor has the expertise necessary to diagnose and treat you, and in part because the doctor has the authority to do things like schedule you for surgery if you need it. Whatever your anxieties at that moment, they don’t include the possibility that the nurses will ignore the doctor’s diagnosis, or refuse to treat you in the manner the doctor suggests.

You don’t worry that expertise and authority are different kinds of things, in other words, because they line up perfectly from your point of view. You simply ascribe to the visible doctor many things that are actually true of the invisible system the doctor works in. The expertise resides in the doctor, but the authority is granted by the hospital, with credentials helping bridge the gap.

So here’s the thing: it’s incredibly expensive to create and maintain such systems, including especially the cost of creating and policing credentials and authority. We have to make and enforce myriad refined distinctions — not just physician and soldier and chairman but ‘admitting physician’ and ‘second lieutenant’ and ‘acting chairman.’ We don’t let people get married or divorced without the presence of official oversight. Lots of people can drive the bus; only bus drivers may drive the bus. We make it illegal to impersonate an officer. And so on, through innumerable tiny, self-reinforcing choices, all required to keep the links between expertise, credentials and authority functional.

These systems are beneficial for society. However, they are not absolutely beneficial, they are only beneficial when their benefits outweigh their costs. And we live in an era where all kinds of costs — social costs, coordination costs, Coasean costs — are undergoing a revolution.

Cost Changes Everything

Earlier, writing about folksonomies, I said “We need a phrase for the class of comparisons that assumes that the status quo is cost-free.” We still need that; I propose “Cost-free Present” — when people believe in we live in a cost-free present, they also believe that any value they see in the world is absolute, not relative. A related assumption is that any new system that has disadvantages relative to the present one is therefore inferior; if the current system creates no costs, then any proposed change that creates new bad outcomes, whatever the potential new good outcomes, is worse than maintaining the status quo.

Meanwhile, out here in the real world, cost matters. As a result, when the cost structure for creating, say, an encyclopedia changes, our existing assumptions about encyclopedic value have to be re-examined, because current encyclopedic values are relative, not absolute. It is possible for low-cost, low-value systems to be better than high-cost, high-value systems in the view of the society adopting them. If the low-cost system can increase in value over time while remaining low cost, even better.

Pick your Innovator’s Dilemma: the Gutenberg bible was considerably less beautiful than scribal copies, the Model T was less well constructed than the Curved Dash Olds, floppy disks were considerably less reliable than hard drives, et cetera. So with Wikipedia and Encyclopedia Britannica: Wikipedia began life as a lost-cost, low-value alternative, but it was accessible, shareable, and improvable. Britannica, by contrast, has always been high-value, but it is both difficult and expensive for readers to get to, and worse, they can’t use what they see — a Britannica reader can’t copy and post an article, can’t email the contents to their friends, can’t even email those friends the link with any confidence that they will be able to see it.

Barriers to both access and re-use are built into the Britannica cost structure, and without those barriers, it will collapse. Nothing about the institution of Britannica has changed in the five years of Wikipedia’s existence, but in the current ecosystem, the 1768 model of creation — you pay us and we make an Encyclopedia — has been transformed from a valuable service to a set of self-perpetuating, use-crippling barriers.

This what’s wrong with Cost-free Present arguments: the principal competitive advantages of Wikipedia over Britannica, such as shareability or rapid refactoring (as of the Planet entry after Pluto’s recent demotion) are things which were simply not possible in 1768. Wikipedia is not a better Britannica than Britannica; it is a better fit for the current environment than Britannica is. The measure of possible virtues of an encyclopedia now include free universal access and unlimited re-use. As a result, maintaining Britannica costs more in a world with Wikipedia than it did in a world without it, in the same way scribal production became more expensive after the invention of movable type than before, without the scribes themselves doing anything different.

If we do what we always did, we’ll get the result we always got

Citizendium seems predicated on several related ideas about cost and value: having expertise and being an expert are roughly the same thing; the costs of certifying experts will be relatively low; building and running software that confers a higher degree of authority to them than on non-expert users will be similarly low; and the appeal to non-experts of participating in such a system will be high. If these things are true, than a hybrid of voluntary participation and expert authority will be more valuable than either extreme.

I am betting that those things aren’t true, because the costs of certifying experts and insuring deference to them — the costs of creating and sustaining the necessary social facts — will sandbag the system, making it too annoying to use.

The first order costs will come from the certification and deference itself. By proposing to recognize external credentialing mechanisms, Citizendium sets itself up to take on the expenses of determining thresholds and overlaps of expertise. A masters student in psychology doing work on human motivation may know more about behavioral economics than a Ph.D. in neo-classical economics. It would be easy to label them both experts, but on what grounds should their disputes be adjudicated?

On Wikipedia, the answer is simple — deference is to contributions, not to contributors, and is always provisional. (As with the Pluto example enough, even things as seemingly uncontentious as planethood turned out to be provisional.) Wikipedia certainly has management costs (all social systems do), but it has the advantage that those costs are internal, and much of the required oversight is enforced by moral suasion. It doesn’t take on the costs of forcing deference to experts because it doesn’t recognize the category of ‘expert’ as primitive in the system. Experts contribute to Wikipedia, but without requiring any special consideration.

Citizendium’s second order costs will come from policing the system as a whole. If the process of certification and enforcement of deference become even slightly annoying to the users, they will quickly become non-users. The same thing will happen if the projection of force needed to manage Citizendium delegitimizes the system in the eyes of the contributors.

The biggest risk with Wikipedia is ongoing: lousy or malicious edits, an occurrence that happens countless times a day. The biggest risk with Citizendium, on the other hand, is mainly up front, in the form of user inaction. The Citizendium project assumes that the desire of ordinary users to work alongside and be guided by experts is high, but everything in the proposal seems to raise the costs of contribution, relative to Wikipedia. If users do not want to participate in a system where the costs of participating are high, Citizendium will simply fail to grow.

Comments (12) + TrackBacks (0) | Category: social software


1. David Gerard on November 21, 2006 9:27 AM writes...

In fairness, there are also all manner of reasons why Wikipedia can't possibly work in theory - if you'd said five years ago "we're going to write an encyclopedia by letting any idiot edit our website and hope the good idiots outnumber the bad ones, and by the way we'll be top-10 in the world by the end of 2006," people would have looked at you funny.

The Citizendium test wiki (not public yet) is getting a lot of work and there's an enthusiastic community forming. I was rather sceptical myself of Citizendium at first, but I'm surprised how active the small community is being. In six months it'll be possible to say what's actually happening in practice.

Permalink to Comment

2. David Gerard on November 21, 2006 2:56 PM writes...

(BTW: I'm a volunteer press contact for Wikimedia, I've edited on Wikipedia for four years now and I'm also on the Citizendium mailing list and will be participating in the project at least a little. Because more open content is good for everyone and should be encouraged.)

Further note: This is not to say that discussing theoretical problems is not important - it's useful to stay aware of possible pitfalls. But a useful rule of thumb at Wikipedia is that it's not a problem unless and until it's a problem.

One thing that theoretical objectors to Wikipedia miss is that just about every possible obvious objection to Wikipedia - point-of-view pushers, idiot kids, the well-meaning clueless, cranks, trolls, marketers, vandlalbots/spambots, arrogant anti-experts, arrogant experts, the most persistent revert-warrior winning, people are a problem, etc., etc., etc. - actually *has happened* and *does happen*, and we deal with them as they come up. Wikipedia is very far from perfect, and no-one is more keenly aware of that than those who work on it, but it appears to be *good enough* to achieve "JESUS CHRIST IT'S A LION GET IN THE CAR" web ratings (number 13 on Alexa today) for a site run by a nonprofit with two technical employees.

I see no inherent reason the Citizendium community can't achieve the same, given time and the harsh real world. We'll see in six months and twelve and so on. There's got to be more than one way to do this.

Permalink to Comment

3. Clay Shirky on November 21, 2006 3:30 PM writes...

David, absolutely. It should go without saying that every blog should appear under a banner reading "My argument may be completely wrong." Indeed, many smart people, including one of Wikipedia's co-founders, think just that.

However, as I've been using wikis since the 90s, and teaching Wikipedia in my classes since 2001, I can say that I have never looked at anyone funny for advocating communal modes of production (though five years ago I might have been skeptical about any Top 10 claims, as you note.) In other words, I don't think that there is a kind of theoretical equivalence between Wikipedia and Citizendium, simply because both are experiments. I think, and have always thought, that wikis are a better way of managing production than systems structured around serious workflow.

In making that argument, I'm trying to do two things at once: first, use what I think I know about social costs to make correct predictions about the design of new systems, and second, if I'm wrong, to be wrong publicly, decisively, and informatively.

If, contra my predictions, Citizendium creates a system of governance that raises costs of participation slightly but value highly, it will succeed, and my argument will be discredited. If my argument is wrong, then discrediting it would be a good outcome (albeit fairly embarrassing for me.)

The only bad outcome would be saying nothing now, and then weighing in after the fact to explain why whatever happened was inevitable.

Permalink to Comment

4. Duncan Lock on November 24, 2006 5:55 AM writes...

I think this simply boils down to work vs. play.
If you're expecting people to write you an encyclopaedia for free in their leisure time, then it needs to be easy and 'fun'. If you make it like work, then people won't do it unless you pay them - I get enough 'work/bureaucracy' at work, thanks.
This is largely why the original GnuPedia failed - contributing to it was way too much like hard work, so virtually nobody did.

Permalink to Comment

5. Zbigniew Lukasiak on November 24, 2006 8:47 AM writes...

I think this should be relevant here:

It is possible that unstructured peer production has some upper limit of quality. For some uses you can live with low quality encyclopedia - but for reference you need something reliable - and this is not a marginal usage of encyclopedias. So I would argument that the value of a structured process of Citizendium can have high value.

Permalink to Comment

6. Zbigniew Lukasiak on November 24, 2006 8:51 AM writes...

I think this should be relevant here:

It is possible that unstructured peer production has some upper limit of quality. For some uses you can live with low quality encyclopedia - but for reference you need something reliable - and this is not a marginal usage of encyclopedias. So I would argument that the value of a structured process of Citizendium can have high value.

Permalink to Comment

7. Rimantas on November 25, 2006 10:58 AM writes...

David, Citizendium faces different question than that five years ago "will this work?". Wikipedia shows that "this works", so question for Citizendium is "will those millions of Wikipedia users who think WP is 'good enough' find any reason to switch?".
I doubt, that content on Citizendium will be THAT much better, and I also doubt that Citizendium will offer any other benefits to justify the cost of switching.

Permalink to Comment

8. Bruce LaDuke on November 26, 2006 10:07 PM writes...

Both Wikipedia and Citizendium are noble concepts and headed in the right direction. I say this because knowledge is one, not many. As knowledge multiplies in our present world, so does the complexity, confusion, and chaos caused by silo thinking.

For example, the term creativity should have one definition with maybe a few variants, but today it has hundreds of definitions. If we could define it as one, then social knowledge would be streamlined and we would greatly simplify our social knowledge base.

In the old days we could make it more big deal...but today, as knowledge rapidly increases, people are drowning in complexity that does not need to exist.

That all said, I don't believe anyone is going to ever get over this 'expertise' problem.

The root problem here is that we don't really understand what logic is. If we did, we'd use logic as our knowledge proof, not expertise.

Expertise has as many levels and scopes as the term creativity in my example above. It means different things to different people and groups in different settings.

Even if expertise did have a single definition, it is not a test of knowledge validity...logic is the only true test of knowledge validity. For this reason, the entire concept of expertise is invalid and illogical...and it can't be used to govern knowledge working effectively.

What is missing is the scientific method for rational thought and a better understanding of how knowledge is formed. If we can better understand the evolution and validation of one thought, a microcosm, we can then manage the whole.

Permalink to Comment

9. Paula Knudsen on November 28, 2006 3:04 PM writes...

Hi! I work for a producing company and we want people to send us a video with their stories. It is for a documentary about bikini.

To know more visit our blog or e-mail me!

Permalink to Comment

10. Anthony on November 28, 2006 3:48 PM writes...

To expand on Rimantas' thought, I think that if Citizendium succeeds in creating higher-quality articles than Wikipedia, that there will be a number of Wikipedia editors who will monitor Citizendium for interesting and useful changes, and backport those edits to Wikipedia, thus raising the overall quality of Wikipedia to the point where the casual user won't have much reason to prefer Citizendium to Wikipedia.

Permalink to Comment

11. Peter Frishauf on December 23, 2006 3:48 PM writes...

An overlooked reason for Wikipedia's success is that it strips authors of the destructive force of ego. Because there are no bylines, and academics can not claim credit for their contributions with tenure committees(!!), Wikipedia is largely free of the detritus that fills the "literature" created to serve the "publish or perish" world of most scholars. Reading Wikipedia it is clear that for the majority of contributors, the primary motivation is a love of information, and the willingness to share it.

Like Clay and Larry, I respect expertise and credentials. As the founder of, a site created in 1995 to freely distribute scientifically-sound, peer-reviewed content, I am continually surprised by the availability of high-quality, continually refactored information on Wikipedia.

My naive hope is that Jimmy Wales and Larry might put their differences (egos?) aside and agree that both have a point. In an ideal world, the complex expert-reviewed system Larry would like to build (with all the complexity and expense Clay identifies) could live in an "Expert Reviewed" tab on Wikipedia. Wikipedia could continue its successful trajectory while a working reputation system could be put in place with an expert-reviewed version of the same content. Thus users would have a choice: the Wikipedia content would have immediacy with the risk of some inaccurate information; the "expert reviewed" content might take longer to create, but presumably would be free of some of the abuses that find their way into Wikipedia today.

Wikipedia has a high degree of trust for comprehensive information. Ideally, it will be improved, not replaced. Creating a Citizendiuum site with parallel information isn't a bad thing, but is it necessary when all we really want is a better Wikipedia? Technically, this is possible. Egos permitting, it could happen.

Permalink to Comment

12. Anonymous on May 21, 2007 11:36 AM writes...


Permalink to Comment

TrackBack URL:


Remember Me?


Email this entry to:

Your email address:

Message (optional):

It's Live! New JCMC on Social Network Sites
Race/ethnicity and parent education differences in usage of Facebook and MySpace
User-generated neologism: "Indigenous content"
history of social network sites (a work-in-progress)
New Freedom Destroys Old Culture: A response to Nick Carr
responding to critiques of my essay on class
Tagmashes from LibraryThing
Spolsky on Blog Comments: Scale matters