
Clay Shirky on Coase, Collaboration and Here Comes Everybody 10/20/2008
What this covers
This conversation between Clay Shirky and host Russ Roberts explores how internet-based tools have upended the economics of coordination and organization. Shirky draws on Ronald Coase's theory of transaction costs—particularly the "floor" below which valuable activities never happen because organizing them is too expensive—and argues that digital platforms have demolished that floor. The discussion ranges across how Wikipedia, Flickr, and open-source software harness what Shirky calls cognitive surplus (the roughly trillion hours per year of leisure time in the industrialized world) to create enormous social value without formal institutions, hierarchies, or management overhead. The core claim is that new tools don't just make coordination cheaper; they invert the economics entirely, enabling self-governing collaboration at scales that would be impossible under institutional rules.
The conversation spans several distinct territories. Shirky explains why wiki software in particular shifts the dynamics of vandalism and preservation—making it easier to undo damage than to inflict it—and contrasts Wikipedia's success with the complete failure of Nupedia, which tried to graft traditional editorial gatekeeping onto the web. He describes how networked communication collapses the old division between broadcast and personal communication, allowing content to scale effortlessly from private to public. He also addresses criticisms that the internet encourages passivity and social decay, locating those complaints more accurately in television culture. The discussion turns toward governance and democracy, touching on how networked tools might enable hybrid forms between direct and representative democracy, and closes on broader questions about whether social science can meaningfully predict or model these collaborative phenomena at all, or whether they remain clouded by feedback loops and historical contingency that defy precise forecasting.
Shirky argues that internet-based tools collapse the 'Coasean floor' of transaction costs, enabling large-scale organized activity without formal organizations—aggregating cognitive surplus and social value that no institution could profitably or feasibly produce.
- Coase's transaction-cost theory has a neglected 'floor' below which valuable transactions never happen because organizing them costs too much
- Tools like Flickr and Wikipedia create social value as a side effect, without institutional coordination, exploiting humans' surplus leisure time
- Wiki software shifts the economics so that defending good content is cheaper than vandalizing it, enabling self-governing collaboration at massive scale
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.
Wiki software shifts the economics of collaboration so that undoing damage is easier than doing it—'imagine it was easier to take spray paint off a brick wall than to put it on'—because radical transparency and full edit histories let any editor reset an article to its last known-good state in a couple of clicks, giving defenders a structural advantage that, for the first time on the internet, favors those who want to preserve value.
“here's the amazing thing that wiki's do they make it easier to undo damage than they make it to do damage imagine it was easier to take spray paint off a brick wall than to put it on”
Cognitive surplus is the reconception of leisure time—the ~20 hours/week of TV watching per person, aggregating to roughly a trillion hours/year in the industrialized world—as an aggregate raw material that can be designed around; Wikipedia consumed only ~100 million hours total (less than Americans spend watching ads on a single weekend), making such projects explainable as harnessed surplus requiring only invitation and reward rather than recruiting and management.
“the idea of cognitive surplus is the shift from seeing leisure time... that we can all voluntarily commit to activities that make us happy it's the shift from seeing that as an individual problem of what do I do with 20 hours every week to seeing it as an aggregate that we can design around”
The birthday paradox illustrates that social complexity scales with the number of pairwise connections, which grows with the square of the number of participants, not linearly—so 36 people yield over an 80% chance of a shared birthday; this is why large organizations are qualitatively different from small ones and must be run differently.
“the number of pairs the number of potential connections between people rises with the square of the number of participants not not at the same rate as the number of participants and as a result by the time you get 36 people you've got over an 80% chance that they share a birthday”
The web privileges effectiveness over efficiency, allowing organized behavior at scales impossible to manage institutionally: Wikipedia's editorial culture (~2 million English contributors, ~5 million across languages) is roughly 100,000 times larger than its managerial culture—an inversion of a newspaper, where the managerial culture (all employees) is about twice the editorial culture.
“on Wikipedia the editorial culture is a hundred thousand times as large as the managerial culture”
Nupedia's failure demonstrates that grafting the institutional transaction-cost model onto the web doesn't work: its seven-step expert workflow created seven points of potential stalling, and as a voluntary effort it produced only about a dozen articles in nine months—whereas Wikipedia succeeded by replacing organized process with massive perennial peer review where everything is provisional.
“a seven step process means there are seven places where things can simply grind to a halt and because this is a voluntary effort that is exactly what happened and after nine months of existence nupedia had you know a dozen articles”
A key reason institutional projects stall is that gatekeepers with veto power enjoy that doing nothing means nothing happens; Wikipedia's success came from NOT giving experts veto power—a deliberately simplistic article ('asphalt: material used to cover roads') triggers others to say 'I can do better,' so articles improve from a provisional seed rather than waiting indefinitely for the ideal expert.
“one of the reasons things don't happen in institutional settings is because the people who were in charge enjoy the fact that if they do nothing nothing happens”
Public goods exist in part because, per Mansur Olson's logic of collective action, if they were privately provisioned at scale everyone has an incentive to opt out (free-ride), so they get under-created in purely voluntary arrangements—this is the flip side of the birthday paradox's network economics.
“this is Mansur Olsen's argument about the what he calls the logic collective action which is public goods exist in part because if they're to be privately provisioned to past a certain scale everybody has an incentive to opt out and so the public public goods get under created in purely voluntary farms”
Maintaining an institution capable of lowering transaction costs itself creates costs (e.g. middle management as the overhead of coordinating large organizations), which is why some valuable activities remain below the Coasean floor.
“famously middle management is is is the requirements for managing down transaction cost and a large organization but the transit the overhead is sensitive made maintaining an institution that can lower those transaction cost itself creates costs”
Coordinating preferences across a group grows combinatorially: getting ten friends to agree on a movie requires forty-five pairwise preference comparisons, so beyond a small scale groups must adopt formal mechanisms (voting) or unilateral governance ('I'm going to this movie, join if you want')—not because of human malice but because it is the only way to manage large groups in institutional settings.
“ten friends is forty-five comparisons of what do you like what do you like and then you've got to roll all of those up and very soon right you get to a point where there has to be either some formal mechanism like voting or some governance strategy like I'm going to this movie”
Different sciences accept different standards of proof: a biologist accounting for a phenomenon to the 95th percentile celebrates, whereas a physics experiment that is only 95% correct 'is wrong on its face'—implying social science, dealing with complex feedback loops, must accept lower predictability standards.
“if you get a biologist to account for something to the 95th percentile right Eureka right there they're popping champagne in the hallways right a physics experiment that don't think as things the 95th percentile is wrong on its face”
Networked communication fuses the previously distinct modes of broadcast and personal communication into a single channel, so content can scale effortlessly from private to public—meaning something no longer has to be interesting to more than two people to be worth putting online, a shift that disorients people raised in the prior broadcast/communication divide.
“the big change here right to change in a way underlying the rest of the the technological change... is the fusing of the old pattern of a broadcast network and a communications network”
Lowering transaction costs becomes a side effect of using networked tools (internet, mobile phones, apps): finding people with shared or complementary interests and coordinating with them becomes far easier, mirroring how search and e-commerce eased commercial transactions—enabling a 'ladder' of sharing, collaboration, and collective action.
“the interesting thing to me about the internet mobile phones applications built on top of them is that lowering transaction cost becomes a side effect of using the network”
We are witnessing 'the end of cyberspace'—the idea that the internet is a separate alternative space one retires into is being replaced by people using the internet to augment real-world activities, including coordinating face-to-face meetups.
“I think we're witnessing the end of cyberspace the idea that the Internet is a separate and alternative space to which you retire from this from the real world”
Existing examples of online collective action are all protest-oriented—relying on 'stop energy,' short-term, high-visibility coming-together to register protest against political or economic entities—rather than groups that come together to take constructive real-world action and succeed or fail as a durable group.
“all of the examples we have a collective action right now our protest oriented they all rely on stop energy their political protest to protest against economic entities but they tend to be short-term high-visibility coming together and registering some sort of protest”
Organizations exist to lower transaction costs: Coase observed that if markets are so efficient, the reason we have firms is that pricing and contracting for every small transaction (like buying pencils) would swamp the institution's ability to accomplish larger goals, so firms internalize those transactions.
“organizations exist to lower transaction call the question coach asked himself was if markets are such a good idea why do we have one we have organizations and all in particular why do we have firms and his answer was that if everything will market the cost of discovering opportunity and writing a contract for every little thing... would would actually swamp the ability of the institution to get larger things done”
Many criticisms aimed at the internet (alienation, passivity, social decay) are misdirected—they actually describe the culture created by television, a battle 'fought and lost by the 1960s'; people are now reading and writing more than before, and doing some of their ~20+ hours/week of screen time with a keyboard and mouse rather than passively, which is hardly a calamity.
“many of the criticisms that the internet is coming in for are not actually directed at the internet they're directed at the culture created by television the idea of a screen coming into the house and alienating people from from social society like that battle was fought and lost by the 1960s”
Behavioral economics has passed through the first three of Bill Hamilton's four phases of a scientific theory (not true / interesting but perverse / true for edge cases / everyone always knew it), reaching the stage where everyone accepts neoclassical economics fails to explain certain cases (ultimatum and dictator games), but we have not yet seen an interpretation that moves behavioral economics to the center of the calculation rather than the neoclassical model.
“bill Hamilton the biologist once said every scientific theory goes through four phases... that's not true that's an interesting if perverse way of looking at the world I see the experimental evidence that counts for certain edge cases and I have always believed that as always everyone knows it”
Working direct democracies (Swiss cantons, New England town meetings) are bounded by scale: past a certain size the transaction costs of participation become prohibitive, so beyond that scale representative democracy is required—but networked tools now make it possible to experiment with hybrid forms that more directly assess public preferences without relying solely on periodic elections.
“all of the examples we have of working direct democracy the Swiss cantons system ten meetings in the in the in the northeastern United States and so forth are specifically limited by scale”
When we observe organized behavior we wrongly assume an organization is directing it, but phenomena like Wikipedia pages are organized without an organizer—an instance of what Hayek called emergent order.
“when we see organized behavior we assume an organization is behind it right someone is directing people to write that Wikipedia page and that's not in fact the case”
Social science explanations of collaborative phenomena must become largely probabilistic ('cloudy with a chance of collaborations') because social life involves many complex feedback loops making it more like predicting weather than gravity—even with full knowledge of inputs, predictability is far lower than the neoclassical model hopes, and such phenomena get stuck in special cases (you couldn't recreate Wikipedia with identical inputs because Wikipedia already exists).
“the physics of this stuff is much more like the physics of weather than the physics of gravity which is to say it's a degree of predictability even when you know all the inputs is much lower than... the neoclassical model hopes for”
The signal non-technological enabler of harnessing cognitive surplus was the creation of the open-source license (GPL, by Richard Stallman ~25 years prior), which injected 'the right amount of inefficiency' into the IP system—using copyright law to subvert rather than extend copyright's traditional goals by granting more rights to users and fewer to creators, so contributors no longer fear being alienated from their work.
“what that license did is it injected the right amount of inefficiency into the intellectual property system... it created something that that actually subverted rather than extending the traditional goals of copyright... it created more rights for the user and fewer rights for the creator”
Federalism enables institutional innovation: because there are 51 decision-makers rather than one, even if a single national decision on virtual companies would likely be 'no,' a single state like Vermont can experiment, and if it gains traction politicians will weigh increased profitability against decreased power and favor profitability—potentially making Vermont 'the Delaware of cyberspace.'
“the great end terrible things that happen the United States happened because of federalism right they happened because we have 50 competitive systems trying all fit models if there were a single decision yes or no should we allow virtual companies to exist quite likely the answer would be no but there are 51 decision”
Neoclassical economics is not as naive about altruism as the straw-man critique suggests: economists, including Adam Smith, have long understood Homo economicus is not purely selfish and have studied charity for centuries—the real failing is that the formal models of how much altruistic activity would occur were 'quite sterile,' and embodying non-financial motivation properly in models may be beyond the scope of social science.
“a number of people including myself actually ironically have tried to incorporate altruism into the formal models of so-called Homo economicus... I think Adam Smith understood that was wrong... it's a bit of a straw man to say that people weren't aware”
Flickr's aggregation of 3,000+ Mermaid Parade photos via freeform tagging accomplished something no institution could do: it created large-scale social value as a side effect of self-interested individual actions, without advertising costs, coordination, or any profit motive—because the value existed but was insufficient to overcome the transaction costs of for-profit assembly.
“what Flickr did was it did something no institution can do which is to create just as a side effect of these self-interested actions this large-scale aggregation of a kind of social value”
Per Danah Boyd's research, the rise in teenagers' social-network use correlates with a reduction in their freedom to go outside: parents and authorities increasingly keep teenagers indoors and apart (a 'mass house arrest'), and teens respond by maintaining social lives through the media they're permitted to access.
“Dana Boyd has done the Berkeley sociologist has done phenomenal work on showing that the rise in social network use among teenagers correlates with the reduction in the freedom to simply go outside”
The internet turbocharges homeschooling: contrary to the stereotype that homeschoolers withdraw into isolation, they form deep networks—sharing educational materials, advice, and even a used-textbook market that exists only within the homeschooling network—reproducing the economics of public goods (infrastructure previously only provisionable by the state) without state provision.
“they've dropped out of a bit of state-sponsored society but are in deep and long-lasting conversation with one another and they're sharing theirs they're sharing educational materials are sharing advice there's a used textbook market that exists only within the homeschooling Network”
Economists pre-2000 would have predicted Wikipedia could not work because of the free-rider problem—everyone would happily read it but no one would create it without profit or glory; they were wrong because they misunderstood the pure pleasure (plus small private non-monetary returns like comments and recognition) that overcomes free-riding, making such voluntary contribution a massive positive social externality unprecedented in human history.
“if you'd asked some economists in 1950 1960 1970 1980 90 even 2000 could Wikipedia work they'd all say no... because you don't get any any glory... there's no profit everyone's gonna freeride”
Wikipedia's success in lowering transaction costs simultaneously opens it to system-gaming (e.g., editing of the Sarah Palin article before her VP announcement), forcing it to back into governance models (semi-protection, edit-war handling) that balance open participation against abuse—making the Wikimedia Foundation effectively 'the experimental wing of political science.'
“Wikipedia is having to back into certain kinds of governance models to prevent people from taking advantage of the system and it's it's it's in my mind the Wikimedia Foundation now now says it makes up the experimental wing of political science”
Majority voting can manufacture an illusory 'will of the group': if six prefer a gory action film and four prefer a romantic comedy, declaring the action film the group's will misrepresents preferences, so where collective action is not strictly required, such decisions should be removed from the political marketplace and decoupled in a live-and-let-live way consistent with liberty.
“if six of them want to go to a romantic comedy... so the will of the group is gory action is just an illusion it's not the will of the group it's what mmm no majority wants”
The next breakthrough will be a 'license for collective action' analogous to the GPL: a legal vehicle (like Vermont's experimental virtual company—no paper filings, physical headquarters, or hierarchical management) that lets purely online groups be deferred to by the state to hire, fire, hold bank accounts, and create bylaws, which would explode collaborative activity beyond information production into real-world action.
“I think that what's coming is a license for collective action it's analogous to the way the GPL provided a license for intellectual production... someone is going to figure out how to take the vehicle of incorporation literally embodiment of a group... and be deferred to by the state”
Beyond Coase's well-known 'ceiling' (where firms stop being efficient and markets take over), there is a neglected 'Coasean floor': a set of valuable transactions that no one engages in because they require organization, yet no organization will undertake them because the transaction costs are too high relative to any extractable profit.
“there's also cozy in floor right there are a set of transactions that no one engages in because you have to be organized to engage in them but no organisation is willing to eat the costs because in this case the transaction costs are simply too high in traditional settings”
Calling Wikipedia a 'positive externality' reveals a theory not working hard enough: the externality of Wikipedia is in fact the business/economic value it creates, while the social effects are internal to the transaction—so internalizing non-financial motivations would better explain what is going on, and Wikipedia will stop being viewed as an externality.
“the externality of Wikipedia is in fact the business value it creates right the positive economic effects of the Wikipedia are the externality and the social effects are what what what are internal to the transaction”
When measuring changes in the American standard of living, we neglect changes in the workplace—in 1900 about 40% of Americans worked on physically dangerous, brutal farms, whereas today's workplace is remarkably safe and, notably, allows much more leisure on the job (people listen to podcasts at work).
“in the workplace in 1900 most Americans about 40% were on the farm that's a pretty unpleasant life is brutal it's physics dangerous today's workplace is remarkably safe... the amount of leisure that's available while you're under the roof”
Markets exist (rather than organizations) where the level of transactions and the complexity of interacting actors are both large enough that you cannot coordinate within an organization—famously what 'did in communism' at the end of the last century.
“markets exist on the other hand where the level of transaction is large enough and the complexity of the interacting actors is also large enough that you can't have been organized for a famously the thing that did in communism in the in the end of the last century”
The aggregate view of shared photos is not merely a post-hoc gathering but actually creates the incentive for more people to take more pictures.
“that aggregate view isn't just a sort of post hoc gathering at these images because it actually creates the incentive for more people to take more pictures”
Words and reading are becoming less important while images, skimming, and browsing become more important, evidenced by the scale of image platforms (Flickr ~2-3 billion images, Facebook over 5 billion)—a largely unnoticed social phenomenon enabled by bandwidth and access.
“words are becoming less important and images are becoming more important reading is becoming less important and skimming and what browsing is more important and Flickr I think has two or three billion images”
The threshold to learning a craft like photography has fallen from a gap into a spectrum: where improving meant joining photo clubs and subscribing to magazines, the socialization of sharing now exposes everyone to others' photos and lets them ask photographers directly how a shot was taken, lowering the threshold of learning something new.
“that gap has now become a spectrum because everybody's exposed all the time to these these photos and can go in and ask the photographer hey that's great we know how'd you take that shot”
The New York City Council overriding the term-limits law that voters approved twice—reportedly so the mayor can run for a third term, with a quid pro quo letting council members also run again—is a large 'bug' in the representative system, overriding the clearly expressed will of voters simply because they can.
“the City Council simply overriding the clearly expressed will of the voters because they can strikes me as a fairly large bug in the representative system”
Public goods like garbage collection and schooling are provided publicly to everyone all the time because it is cheaper to do so that way.
“why does it have to be publicly provided to everyone all the time because they're cheaper they're cheaper to do that way”