
Jimmy Wales on Wikipedia 03/09/2009
What this covers
This 2009 EconTalk conversation pairs Russ Roberts with Jimmy Wales to trace how Wikipedia was built and why it holds together. Wales recounts the failure of Nupedia, his earlier project, whose application process and expert review he calls intimidating and joyless for volunteers. He then describes how an open Wikipedia outproduced it within two weeks. Much of the talk is about Hayek's 1945 essay, which Wales read as an undergraduate. He treats it as a practical design test: he checks whether a proposal centralizes something that could stay decentralized, since centralization creates a single point of failure. Roberts adds the Oxford English Dictionary as an early case of crowdsourcing.
The middle of the conversation turns to how the project governs itself and how reliable it is. Wales explains that policy is descriptive, with pages rewritten to match actual practice, and that the arbitration committee works like a court that cites its own precedent. He discusses neutrality) through an example of Mormon editors who wrote from their own perspective and accepted simple attribution as a fix. He also makes a counterintuitive claim: obscure topics, such as niche anime, suffer more bias than controversial ones, because only fans are watching them. On reliability, Roberts compares Britannica's treatment of Milton Friedman with Wikipedia's. They discuss the study that found about three errors per Britannica article. Wales also says nobody should cite either encyclopedia in university work.
The closing stretch is more speculative. Wales gives some figures: 280 million monthly visitors, a staff of 25, and 2.7 million English entries with obvious gaps. He expects English growth to slow. He disputes the idea that a slice of television time could be converted into many more Wikipedias, because contributors are self-selected. Roberts and Wales then argue that this voluntary coordination illuminates how markets work. Wales also describes the wiki movement as a revival of folk culture, distinct from both commercial and subsidized culture.
Jimmy Wales argues that Wikipedia works because it follows Hayek's insight that knowledge is dispersed. Pushing editorial decisions out to the people who hold the information, under evolved and descriptive norms, beats top-down expert curation, as the failure of the expert-gated Nupedia showed. Wales and Russ Roberts further argue that voluntary bottom-up coordination of this kind says something about markets and about culture.
- Nupedia's top-down expert review was intimidating and produced little, while open Wikipedia did more in two weeks than Nupedia did in two years
- Hayek's 'Use of Knowledge in Society' frames centralization as an information-flow bottleneck and single point of failure, which Wales uses as a design heuristic
- Wikipedia's governance (policy, ArbCom precedent, neutrality) evolved descriptively, and quality depends on opposing viewpoints scrutinizing articles
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
The Oxford English Dictionary, as recounted in 'The Professor and the Madman', is probably one of the first known examples of what we now call crowdsourcing. Thousands of individuals unknown to each other contributed dispersed Hayekian knowledge to a large project.
“one of the first examples that we know about of where what we now call crowdsourcing where thousands of unknown individuals unknown to each other contributed the dispersed Hayekian knowledge”
The Mormon editors were not intentionally biased or subverting Wikipedia; they wrote what they knew and it did not occur to them to phrase it otherwise. When someone reworded the entries to 'according to the Book of Mormon, so-and-so was a prophet', the original editors were not offended and agreed it was correct. This shows that much non-neutrality is a perspective blind spot fixable by simple attribution.
“they weren't intending to be biased they weren't trying to subvert Wikipedia they just wrote what they knew”
When evaluating proposals for Wikipedia, one of the first things Wales checks is whether the proposal tries to centralize something that could be left decentralized. Centralization creates a bottleneck and a single point of failure, which is bad design.
“is this an effort to centralize something that can be left decentralized because the real problem with centralizing is that you now have a bottleneck a single point of failure”
Wikipedia policy is descriptive rather than prescriptive. When someone notices that a policy page does not document what is actually being done, they change the page to match practice.
“rather than policy being prescriptive written policy it's actually descriptive somebody will go and change the policy page when they notice that what is documented there is not what's actually being done”
When Wikipedia opened, Wales assumed they would be forced to lock everything down within a week or two, once people found out about it. He decided to keep it as open as possible for as long as possible, and it remains very open today.
“I had assumed at the time that we would be you know forced to lock everything down within a week or two”
Hayek's 1945 American Economic Review essay 'The Use of Knowledge in Society', which Wales read as an undergraduate, deeply shaped his game-theoretic thinking about rulemaking. One basic lesson he drew is that making a rule against something does not stop it from happening; it just changes the costs and benefits.
“simply making a rule against something doesn't mean it doesn't happen it it just changes across benefits”
The Hayekian question maps directly onto encyclopedias. One option is to gather all the world's information, send it to a group of experts, and have them make editorial decisions. The other is to push decision-making out to the people who actually have the information. This analogy shapes how Wales thinks Wikipedia functions and ought to function.
“is it better to gather all the world's information have it sent in to a group of experts and then make editorial decisions or is it better to push the decision-making out to the endpoints”
Wikipedia is not just a billion random pages on the internet. Volunteers follow rules, standards, and guidelines and voluntarily coordinate their efforts. Although it is not a market system and is not run by money, this tells us a lot about how markets work, and Wales hopes it opens people's eyes to bottom-up coordination.
“they're actually able to coordinate their efforts and do that in a voluntary fashion I think it does tell us a lot about how markets work even though it's not a market system it's not run by money”
After launching in 2001 as a very open system, Wikipedia got more work done in two weeks than Nupedia had done in nearly two years.
“we got more work done in two weeks and we had gotten done in nearly two years”
Hayek reframed the central-planning versus market debate as a question of information flow. The choice is whether to send all information inward to a central authority to decide (e.g., a government official deciding how much bread to make) or to leave the information where it is and push decision-making out to the endpoints (individual bakeries).
“Hayek's points in that essay was that really in many ways this is a question of information flow”
Because contributors skew geeky and geeks are less into sports, Wikipedia's coverage of certain sports is limited relative to their popularity. Getting sports watchers to make a few edits could contribute millions of hours of useful effort.
“Wikipedia's coverage of certain sports is pretty limited compared to the popularity anyway of those sports because keke people aren't always into sports so much”
Nupedia, Wales's first free-encyclopedia project (conceived 1999), failed. It was a top-down project with a complex review system: contributors had to apply and prove their qualifications before writing on a subject. The process was intimidating and not much fun for volunteers.
“nupedia was a very top-down project a very complex review system people actually had to make an application to be allowed to write an article on a particular subject and they had to prove that they were qualified”
The arbitration committee is the last stop in Wikipedia's dispute resolution process and functions somewhat like a Supreme Court. It is very aware of precedent, and the principles in its findings are later cited by editors to resolve disputes without escalating to the committee.
“they hear cases in a similar way to the Supreme Court ... the arbitration committee is very aware of the importance of precedent”
Hayek distinguished law from legislation. He argued that a judge's job is not to decree what the law is but to discover the norms and customs that have evolved and rule accordingly, which parallels Wikipedia's descriptive governance.
“it's also very Hayekian he made the distinction between law and legislation and suggested the judge's job isn't to figure out what the law is in de cría but rather should discover what norms and customs have evolved”
Converting TV time into Wikipedias rests on a fallacy. The 100 million hours of Wikipedia effort come from self-selected people drawn to intellectual projects (basically geeks). Many TV watchers would be glad not to be writing an encyclopedia, so 1% of TV time (2 billion hours, about 20 Wikipedias) is not straightforwardly available.
“there are some fallacies though of course in doing this because there 100 million hours of human thought come from people who are drawn to an intellectual project”
A traditional encyclopedia like Britannica inevitably presents one expert's view of a figure like Milton Friedman: glowing if written by an acolyte, insulting if by an antagonist. Britannica's economist entries get basic facts right (birthplace, Chicago, Stanford) but lack nuanced discussion of why the person matters intellectually. In that respect Wikipedia dwarfs Britannica in many areas.
“the encyclopedia has inevitably the Britannica written or encyclopedia inevitably has one experts view of Milton Friedman”
People generally should not cite Wikipedia as a source in university-level work, for the same reason one should not cite Britannica. An encyclopedia's proper role in research is broad background to get oriented; serious work should go to the references and original materials. This holds independently of reliability; it would be true even if Wikipedia were perfectly accurate.
“I don't think people should use Wikipedia as a sighted source in general”
What surprised people about the study was Britannica's roughly 3 errors per article. It showed that great reference work is really hard. We should try to minimize errors and never be complacent, but a perfect encyclopedia is not within human possibility.
“we're never going to have a perfect encyclopedia that's just not within the realm of human possibility”
English Wikipedia had about 2.7 million entries at the time. Large gaps remain: for example, Wikipedia probably still lacks an article on every member of India's roughly 600-member Parliament, so there is still a long way to go.
“we don't have an article on every single member of Parliament of India so that's a very large Parliament I think around 600 people”
We have traditionally thought of culture as two things: commercial culture (pop music, movies, TV, producing both the Beatles and much rubbish) and subsidized fine-arts culture (opera, museums) paid for by government or wealthy patrons rather than direct consumers. Wikipedia and the broader wiki movement are something different, a revival and strengthening of folk culture. People create not for subsidy or for sales like a Harry Potter book, but in a grassroots, decentralized way.
“it's what I call folk culture it's a reimburse and a strengthening of the idea of folk culture of ideas being spread in a very grassroots and decentralized way”
Counterintuitively, Wikipedia has a bigger bias problem on less controversial subjects than on controversial ones. Controversial figures like Friedman attract both supporters and detractors who check each other, producing good articles. Obscure topics attract only fans (or haters), so there are not enough critical eyes on each side; obscure Japanese anime articles, for example, are written by unabashed fans and skew mildly positive.
“we have a bigger problem of bias of non neutrality on the less controversial subjects than we do on more controversial subject”
The biggest 'study' of Wikipedia's quality is the market test of 280 million monthly visitors, which suggests people are getting value from it, though they could be led into error.
“two hundred and eighty million visitors in a month I'd say it's a pretty good market test that people are getting some value from it”
Wikipedia's English content is growing at a slowing pace. There is a natural limit to the number of entries because titles are consistent: there will be one article on Colin Powell, not 35. Wales refuses to forecast the eventual article count because any answer will look funny given enough time.
“the nature of a Wikipedia entry is such that there is a limit to how many entries that we can possibly have ... we're not going to have 35 different entries on Colin Powell”
About 280 million people visit Wikipedia each month, per comScore numbers that match Wikipedia's own logs. Yet the Wikimedia staff numbers only 25, and everyone else, including Wales, is a volunteer.
“280 million people a month are using the site and yet we have a staff now of only 25 people and everybody else including me is all volunteers”
Clay Shirky, a previous EconTalk guest, estimated that Wikipedia represents about 100 million hours of human effort, and that Americans watch about 2,000 Wikipedias' worth of TV a year (roughly 200 billion hours).
“Clay Shirky who is a guest on the show while back suggested that Wikipedia at least when he was looking at it is the product of about a hundred million hours of human effort”
Wikipedia's system of government evolved for many historical reasons, and the community has never tried a radical rationalization of it. Americans find this harder to grasp than the British: Americans look for a written constitution, while British law contains centuries of customs and traditions that are not written down anywhere.
“I think Americans have a harder time getting it then than the British because we Americans we like to think of we're always looking for where is the Constitution”
A study from about three years before the interview sent Wikipedia and Britannica articles to experts to count errors. It found about 3 errors per article in Britannica and about 4 in Wikipedia, so Wikipedia was not quite as good but was within striking distance.
“there were about three years for article in Britannica and about four errors per article in Wikipedia”
Wales realized the top-down approach would never work when he tried to write a Nupedia article himself and froze. He had a PhD-level background in finance and a published paper on option pricing theory, yet he hit complete writer's block writing about Robert Merton, the recent Nobel laureate in option pricing, because the process was so intimidating.
“I thought I could write an entry about Robert Merton ... and just found myself with a complete writer's block it was very intimidating”
The Hayek analogy came after Wikipedia's founding rather than inspiring it. It is more a tool for understanding what works and doesn't work in Wikipedia and why.
“I think that analogy came after the fact I would say but it's more about understanding what works and doesn't work in Wikipedia and why”
The two US minority religions illustrate the bias pattern. The Church of Scientology is litigious, sued people online over revealing its secrets, and has many online critics and supporters, so its Wikipedia entries are very good because both sides vet and negotiate them. Book of Mormon entries, by contrast, stated things like 'so-and-so was a prophet' because almost only Mormons knew or cared about the content.
“the entries on topics related to Scientology tend to be very very good they're very carefully vetted by people who are supporters people who are opponents”
Unlike most founders, Wales faces no real succession problem. Wikipedia is largely self-sustaining; if he vanished, the English Wikipedia arbitration committee would take over his remaining special role. He jokes that, like Colonel Sanders, he'll eventually be a cartoon on the box.
“I think the Wikipedia would be fun if I were to vanish the arbitration committee of English PT would take over”
Roberts argues the accuracy issue is the wrong one. He found that Wikipedia's Milton Friedman entry attributes 'inflation is always and everywhere a monetary phenomenon' to 'A Monetary History of the United States', whereas Roberts believes it came from a 1963 lecture Friedman gave in India. Such errors might earn a red mark on a high school paper but are not very important.
“it suggests that the phrase inflation is everywhere and always a monetary phenomenon comes from the monetary history of the United States I don't think that's where it's from I think it came from a lecture he gave in India in 1963”
The Britannica comparison study was imperfect. It looked at a handful of possibly cherry-picked articles, mostly scientific and technical topics where Wikipedia happened to be strong, so Wikipedia would have come out much worse on topics where it was weak at the time.
“they focused on mostly scientific and technical topping ranks where we happen to have a strength so we would have come out much worse had they you know done in trees that we were very weak on”
Wikipedia articles get longer over time, but only up to a point. The community feels overly long articles become unusable, so when an article gets subjectively too long it is split into sub-articles, e.g., the main China article plus 'Economy of China', 'Human rights in China', and 'Demographics of China'.
“if you look at you know China well there's the mean article on China which is several pages long but then there will be many cite articles like the economy of China human rights in China”
The bulk of Wikipedia's costs has traditionally been technology-related, mainly bandwidth and new servers, with staff a smaller share.
“the bulk of the cost traditionally have been technology related”
Deciding to split an article does not require administrator powers; anyone can propose it and create the new article. In practice, though, administrators do most of it.
“there's nothing about that process of deciding whether we should move something out into a separate article that is really requires administrators”
Harnessing spare time is already happening because people go through an internet life cycle: they start as readers, eventually become participants, and may end up blogging, commenting, or editing Wikipedia.
“there's sort of a life cycle of you on the internet they get on the internet and there are readers at first but eventually they become participants”
For much younger students, such as a 10-year-old writing a short class paper, citing Wikipedia is fine. We should be glad the child is writing and thinking about giving credit to sources that helped.
“a 10 year old is writing a little short paper for class and they want to say that they got some information from Wikipedia that I think we should be just glad”
A surprising feature of Wikipedia is that its neutrality concept works reasonably well. Problems occur, but an entry on an important figure like Milton Friedman will generally not be extremely biased in one direction.
“one of the things that I think is surprising about Wikipedia is is the neutrality that the neutrality concept actually works reasonably well”
Wikipedia's governance is community driven. The community elects administrators and members of the arbitration committee. Wales decides when elections occur and formally makes final appointments to the arbitration committee, though traditionally he follows the community vote.
“there are people who are elected by the community to be administrators then people can also be elected to the arbitration committee”
Wikipedia is having an astounding cultural impact worldwide. It is the number three hit in most search engines for almost anything, and people rely on it for information. Responsible citizens of the world, not just Wikipedians, should therefore think hard about its strengths and weaknesses; it is a socially important question.
“it's the number three hit in most search engines for most anything you type in”
In theory one need not be an administrator to be elected to Wikipedia's arbitration committee, but that has never happened, and a non-admin making it is extremely unlikely.
“in theory you don't need to be an administrator be elected the arbitration committee but that's never happened”
For the Wikipedia organization, the biggest challenge is always fundraising to keep the servers running. Although there have never been really serious fundraising problems, it must be taken very seriously.
“for the organization the biggest challenge of course is always fundraising and we've never had any really serious problems with fundraising”
Wales says he is a bigger fan of people watching TV than most people in his position, and that TV is not as bad as people make it out to be.
“I'm actually a bigger fan of people watching TV than most in my position I don't think it's quite as bad as people make it out to be”
Wales had no knowledge of the OED's history when he started Wikipedia. He only read 'The Professor and the Madman' several years later at someone's urging.
“I did not I did not in fact I read the the the famous books about it the professor and the madman only really several years later”
Wikipedia's impact hits Wales most when traveling. Visiting community technology centers in some of the poorest areas of the Dominican Republic, he found children living in shacks with tin roofs and with very poor access to education using Wikipedia every day.
“these are kids who are living in shacks with tin roofs ... but they're using Wikipedia”
Top-down expert review was costing a lot of money and progress was very slow, which is why Wales had long complained about it.
“it wasn't didn't seem to be working very well and it was costing a lot of money and progress was very slow”
Wales's new company, Wikia (pitched as 'building the rest of the library'), hosts about 12,000 active wiki communities, ranging from political activism and humor to extensive video gaming content reflecting its computer-savvy geek base. The aim is to expand the citizen-participation, collaborative-authorship model into as many niches as possible.
“we've got about twelve thousand active wiki communities building all kinds of projects”
The English Wikipedia arbitration committee has about 15 members, all volunteers, with typically 9 to 11 reasonably active at a time.
“there are total of 15 people on the committee but it's all volunteers ... we normally run with probably 9:00 to 11:00 a reasonably active”
English Wikipedia had 1,622 administrators at the time of recording, with around a thousand active.
“now here we go 1622 administrators and English Wikipedia”
The lack of academic research into Wikipedia's quality is a shame for two reasons. Most research would likely find Wikipedia's quality fairly decent, and research identifying where Wikipedia is flawed, where its error rates are higher, and what its strengths are would be very valuable for the community.
“I would love to see research actually helping us to understand better where is Wikipedia flawed where does it tend to have a higher error rate”
Wikia had just launched an experimental question-and-answer site in a wiki format, letting people ask and answer questions collaboratively, and it was going well. Wales says he never develops things in secret.
“the thing we've just launched is the question and answer site which we're still experimenting with”
Wales says he never develops things in secret; he randomly does things as he finds them interesting. He jokes that his next project might be to 'kill the cell phone carriers'.
“I never develop things in secret I just sort of randomly do things as I think find them interesting”