by Clay Shirky
A firm is successful when the costs of directing employee effort are lower than the potential gain from directing. It’s tempting to assume that central control is better than markets for arranging all sorts of group effort. (Indeed, during the twentieth century much of the world lived under governments that made that assumption.) But there is a strong limiting factor to this directed management, and that is the cost of management itself. Richard Hackman, a Harvard professor of psychology, has studied the size and effectiveness of work groups in Leading Teams. Hackman tells a story about a man who ran a nonprofit whose board of directors numbered forty. When asked what he thought such a large board could accomplish, he replied, “Nothing,” in a way that implied he liked it that way. Because of managerial overhead, large groups can get bogged down, and whenever transaction costs become too expensive to manage within a single organization, markets outperform firms (and central management generally).
Activities whose costs are higher than the potential value for both firms and markets simply don’t happen. Here is the institutional dilemma again: because the minimum costs of being an organization in the first place are relatively high, certain activities may have some value but not enough to make them worth pursuing in any organized way. New social tools are altering this equation by lowering the costs of coordinating group action. The easiest place to see this change is in activities that are too difficult to be pursued with traditional management but that have become possible with new forms of coordination.
How Did All Those Pictures Get There?
On the last Saturday in June, Coney Island kicks off the summer with the Mermaid Parade, a sort of hometown procession for New York City hipsters. Hundreds of people show up to march around Brooklyn’s famously run-down amusement park in costumes that are equal parts extravagant and weird—a giant red octopus puppet, a flotilla of hula-hooping mermaids, a marcher sporting a bikini top made of two skulls. Thousands turn out to watch and photograph the festivities, taking pictures ranging from a couple of snapshots to dozens of high-quality photos.
A handful of these pictures end up in local newspapers, but for most of the history of the Mermaid Parade, most pictures were seen only by the people who took them and a few of their friends. The sponsor of the parade didn’t provide any way for the photographers to aggregate or share their photos, and the photographers themselves didn’t spontaneously organize to do so. That is the normal state of affairs. Given the complexities of group effort, hundreds of people don’t spontaneously do much of any consequence, and it wouldn’t have made much sense for anyone to expend the effort to identify and coordinate the photographers from the outside. A couple of years ago, however, the normal state of affairs stopped operating.
In 2005, for the first time, a hundred or so of the attendees pooled thousands of their Mermaid Parade photos and made them publicly available online. The photos came from all sorts of photographers, from amateurs with camera-phones to pros with telephoto lenses. The group was mainly populated by casual contributors—most people uploaded fewer than a dozen photos—but a handful of dedicated contributors shared more than a hundred pictures each, and one user, going by the online name czarina, shared more than two hundred photos on her own. The group pooled these photos by uploading them to a service called Flickr, giving each of the photos a free-form label called a tag. As a result, anyone can go to Flickr today, search for the tag “mermaidparade,” and see the photos. This is a simple chain of events: people take pictures, people share pictures, you see pictures. It’s so simple, in fact, that it’s easy to overlook the substantial effort involved behind the scenes.
Flickr is the source of the sharing, but here’s what Flickr did not do to get the sharing to happen: it didn’t identify the Mermaid Parade as an interesting event, nor did it coordinate parade photographers or identify parade photographs. What it did instead was to let the users label (or tag) their photos as a way of arranging them. When two or more users adopted the same tag, those photos were automatically linked. The users were linked as well; the shared tag became a potential stepping-stone from one user to another, adding a social dimension to the simple act of viewing. The distinction between Flickr coordinating users versus helping them coordinate themselves seems minor, but it is in fact vital, as it is the only way Flickr can bear the costs involved. Consider what it would have taken for Flickr to organize hundreds of amateur mermaid photographers. Someone at Flickr HQ would have had to know about an obscure parade on the other side of the country. (Flickr is based in California.) They would have had to propose a tag for the group to use in order to assemble the uploaded photos. Finally, they would have had to communicate the chosen tag to everyone going to the parade.
This last step is especially hard. When you are trying to address a diffuse group, you are locked into the dilemma that all advertisers face: how do you reach the people you want, without having to broadcast your message to everybody? People in the category “Potential photographer of the Mermaid Parade” aren’t easy to find. Flickr couldn’t have known in advance who would go to the parade. Instead, they would have to send messages out to many more people than would actually attend, in hopes of reaching the right audience, advertising to photographers, hipsters, New Yorkers, and so on, in hope of getting the tiny fraction of those groups who would actually go. Most such ads would be seen by people who weren’t going to the parade, while most of the people who were going wouldn’t see (or pay attention to) the ads. Given those obstacles, no business in the world would take on the job. The profit motive is little help; no one could sell enough pictures, even the skull-bikini ones, to be able to pay the photographers, much less leave any profit afterward. Likewise, no nonprofit or government agency would touch the problem; even the porkiest of pork-barrel projects isn’t going to cover publicity for hula-hooping mermaids. The gap between effort and payoff is too large for any institution to span.
Yet there the photos are. Without spending any serious effort on any individual set of photos, and without doing anything to coordinate or even identify groups of photographers, Flickr has provided a platform for the users to aggregate the photos themselves.
The difference between the value of the photos and the cost of aggregation is a general one. Flickr isn’t just for photos of dancing mermaids, family reunions, and the effects of that third margarita; it also hosts photos of broad public interest. Flickr provided some of the first photos of the London Transport bombings in 2005, including some taken with camera-phones by evacuees in the Underground’s tunnels. Flickr beat many traditional news outlets by providing these photos, because there were few photojournalists in the affected parts of the transport network (three separate trains on the Underground, and a bus), but many people near those parts of the transport system had camera-phones that could e-mail the pictures in. Having cameras in the hands of amateurs on the scene was better than having cameras in the hands of professionals who had to travel.
The photos that showed up after the bombings weren’t just amateur replacements for traditional photojournalism; people did more than just provide evidence of the destruction and its aftermath. They photographed official notices (“All Underground services are suspended”), notes posted in schools (“Please do not inform children of the explosions”), messages of support from the rest of the world (“We love you London”), and within a day of the bombings, expressions of defiance addressed to the terrorists (“We are not afraid” and “You will fail”). Not only did Flickr host all of these images, they made them available for reuse, and bloggers writing about the bombings were able to use the Flickr photos almost immediately, creating a kind of symbiotic relationship among various social tools. The images also garnered comments on the Flickr site. A user going by Happy Dave posted an image reading “I’m OK,” meant to alert his friends who had subscribed to his images on Flickr; he received dozens of comments from well-wishers in the comments. The “Do not inform the children” image generated a conversation about how to talk to kids about te
rrorism. The basic capabilities of tools like Flickr reverse the old order of group activity, transforming “gather, then share” into “share, then gather.” People were able to connect after discovering one another through their photos.
A similar change in the broadcasting of evidence happened after the awful destruction caused by the Indian Ocean tsunami at the end of 2004. Within hours of the tsunami dozens of photos were available on the Web showing various affected places, and within days there were hundreds. As with the London bombings, there was no way to get photojournalists on the scene instantly, but here the problem was not just the speed of response but the spread of the damage, which affected thirteen countries. And as with the London bombings, the photos weren’t used just for evidence; people began uploading photos of missing loved ones, and various weblogs began to syndicate these photos to aid in relocation. The most visited photo tagged “tsunami” is a picture of a little boy, age two at the time he went missing. The picture originally went up with contact information to aid in the search, but as time went on, it turned into an ongoing memorial; viewers posted hundreds of comments of support and prayers under the photo, and many commenters came back months later to check in and conversed with one another in the comments. When the boy’s body was finally recovered and identified, months later, several people posted the sad news on Flickr, and the community that had formed around the photo posted expressions of grief and condolences for the family, then dissolved.
Photo sharing also helped provide the world with documentation of the 2006 military coup in Thailand. Immediately after the coup the military placed restrictions on reporting by the media, but it didn’t (and probably couldn’t) place similar restrictions on the whole populace. As a result, many of the earliest photos of tanks in front of Government House, the parliament building, came from individuals posting images from ordinary digital cameras, and they were discoverable by their tags (Bangkok, Thailand, Military, Coup). One of those users was Alisara Chirapongse, then a fashion-obsessed college student, who blogs under the name gnarlykitty, who posted the coup photos to her weblog, along with running commentary on the cause and immediate aftermath of the army overthrowing Thaksin Shinawatra, then prime minister. As the army announced that it wanted to take control of communications and ban public political speech, her posts took on a new urgency:
One new little change that this law brought us is the whole new level of censorship. No political gathering, no discussing politics, and of course no voicing your opinions whatsoever about the whole mumbo jumbo coup. (Oops did I just do that?)
Alisara posted links to Wikipedia, the collaboratively produced encyclopedia, which was acting as a clearinghouse for breaking news of the coup (as is now usual). She also pointed her readers to a petition to restore freedom of speech and to a proposed demonstration, which she later attended and photographed.
Then, as the initial disorientation of the coup gave way to the new normal, Chirapongse went back to her life as a fashion-obsessed student. As she put it,
This blog is my personal blog where I usually write things concerning my life and things I like. Since my life is lived here in Bangkok Thailand, it should come as no surprise to anyone that I sometime blog about it. So blogging about the Coup is merely blogging about something that’s currently happening in my country.
The rest of that post was about a night she spent at a club, and the post after that was about how much she likes her new camera-phone. She wasn’t a full-time journalist, she was a citizen with a camera and a weblog, but she had participated in a matter of global significance at exactly the time when the traditional media were being silenced.
The content in these examples is quite varied—the gentle ridiculousness of the Mermaid Parade and the awful seriousness of the London bombings; the man-made intervention of a military coup and the natural destruction of the tsunami. The common thread is the complexity of gathering the photos. The groups of photographers were all latent groups, which is to say groups that existed only in potentia, and too much effort would have been required to turn those latent groups into real ones by conventional means. The mermaid photos were too unimportant to be worth any institutional effort. The London bombing photos were taken by the people on the scene. The tsunami’s destruction was spread out over tens of thousands of miles of coastland, and the uses of photos included finding missing persons, something outside the purview of typical newsgathering. During the Thai coup the military rulers were able to place restrictions on organized media, giving amateur photographers an advantage in providing views of tanks in the streets. In each of those cases the cost of coordinating the potential photographers would have defeated any institution wanting to put photos together quickly and make them available globally.
The task of aggregating and making photos available is nothing like, say, the task of putting a man on the moon. Prior to services like Flickr, what kept photo-sharing from happening wasn’t the absolute difficulty but the relative difficulty. There is obviously some value to both photographers and viewers in having photos available, but in many cases that value never exceeded the threshold of cost created by the institutional dilemma. Flickr escaped those problems, not by increasing its managerial oversight over photographers but by abandoning any hope of such oversight in the first place, instead putting in place tools for the self-synchronization of otherwise latent groups.
Making the Trains Run on Time
The structure of traditional managerial oversight is often illustrated by an “org chart,” a diagram of the official organizational hierarchy. This chart is the simplest possible view of an organization’s reporting structure. It is usually drawn as an inverted tree of boxes and arrows. The box at the top represents the head of the organization; the lines drawn downward from that box connect her to various officers and vice presidents through the layers of management, until, at the bottom, there are the rank and file, represented by boxes with lines connecting upward but not downward. The org chart diagrams both responsibility and channels of communication—when two boxes are connected on such a chart, the upper box is the boss; communication from the CEO flows down through the layers of management, while information from the workers flows up in the same way. Compared to the chaos of the market, the org chart draws clear and obvious lines of responsibility, and it is that very clarity that allows the firm to outperform a pure market for work.
The org chart is like institutional wallpaper—ubiquitous and not terribly dramatic. It’s funny to think of it as a specific invention, but its existence and form owe quite a lot to the environment in which it was first widely used—railroad management in the 1800s. The pioneering managerial methods were meticulously documented by Alfred Chandler in his book The Visible Hand. The principal problem in running a railroad was arranging for eastbound and westbound trains to share the same track, because it was prohibitively expensive to lay more than one track for a particular line. By 1840 Western Railroad, a pioneer in building longer rail lines, had to deal with a dozen trains crossing in opposite directions every day. That situation created obvious safety risks, risks that were not long in moving from the theoretical to the real: on October 5, 1841, two passenger trains collided head on, with two fatalities and seventeen injured. This accident alarmed both the public and Congress and forced the railroads to rethink their management.
For the next fifteen years railroads invested in better oversight. As a result, their safety record improved, but their profitability decayed. A big firm like Western could haul more people and cargo to more places than could a smaller railroad, but the cost of managing the enterprise had risen much faster; Western was actually making less money per mile of track than its smaller competitors. David McCallum, a railroad superintendent for the New York & Erie Railroad, proposed both an explanation and a solution for this decayed profitability. As he put it in his Superintendent’s Report of 1855:
A Superintendent of a road fifty miles in length can give its business his personal attention, and may be cons
tantly engaged in the direction of its details . . . any system however imperfect, may under such circumstances, prove comparatively successful.
In the government of a road five hundred miles in length, a very different state exists. Any system which might be applicable to the business and extent of a short road, would be found entirely inadequate to the wants of a long one.
More is different: a small railroad could function with ad hoc management, because it had so few employees and so few passing trains, but as the scale rose, the management problems rose faster. This is where the institutional dilemma meets Birthday Paradox math: not only does managing resources take resources, but management challenges grow faster than organizational size.
McCallum’s proposed solution to this dilemma included making a clear delineation of the responsibility for different segments of track. Central management would oversee regional divisions and supervise the trains passing through their region. McCallum introduced several formal innovations to New York & Erie: strong hierarchical oversight, including an explicitly divisional organization of the railroad with different superintendents responsible for different parts of the railroad. He diagrammed this form of organization with what may have been the first commercial org chart in history. This method was widely copied by other railroads, then by other kinds of firms.
In addition to revolutionizing management structure, McCallum wrote six principles for running a hierarchical organization. Most are what you’d expect (number one was ensuring a “proper division of responsibilities”), but number five is worth mentioning: his management system was designed to produce “such information, to be obtained through a system of daily reports and checks, that will not embarrass principal officers nor lessen their influence with their subordinates.” If you have ever wondered why so much of what workers in large organizations know is shielded from the CEO and vice versa, wonder no longer: the idea of limiting communications, so that they flow only from one layer of the hierarchy to the next, was part of the very design of the system at the dawn of managerial culture.
