2008 Interview with Clay Shirky

This interview originally appeared March 31, 2008 at the now-defunct Worldchanging.com website.

Clay Shirky is an influential writer, consultant, and teacher focused on the Internet as a social platform. He's one of the smartest thinkers I know about how people live, love, and work online. His new book, Here Comes Everybody:The Power of Organizing without Organizations, was just published by The Penguin Press. As an intro to Chapter 11, on "Promise, Tool, and Bargain," he says "There is not recipe for the successful use of social tools. Instead, every working system is a mix of social and technological factors." Clay and I had the following conversation early in March. We'll follow up with an asynchronous conversation on the WELL for two weeks starting May 28.

Jon Lebkowsky: My first very general question for you is about how the web started changing around 2000. What are your thoughts about what was driving the changes, and how the changes have affected our experience of the web?

Clay Shirky: This is the sort of ancient history that got me doing the book. Here Comes Everybody is, in a way, a do-over. I wrote an earlier book – a very different kind of book, about online community – and I had the grave misfortune to have it come out in April of 1995. The book was all about Usenet, the WELL, Echo, and it was about all of the social components of the pre-web Internet. And in April of '95, no one wanted to hear about that stuff anymore.

In fact, I got pulled into the web, too. I taught myself HTML, like a lot of people. I ended up being Production Manager of one web shop, and Chief Technology Officer of another. In that period, '95 to 2000, the template for the social use of the web was really under-optimized. Everybody was excited about using it to distribute information, and everybody was excited about ecommerce. We were basically recapitulating these older patterns: point to point transactions, replicating newspapers, magazines and so forth on the web.

I think that the change that started in 2000 came about for a couple of reasons. One – HotMail brought us all to the realization that the web could be a new interface for existing social platforms. It wasn't like email was one thing, and the web was the other. The web, in fact, was a general purpose interface.

The second thing is so many people were online by 2000, that you could actually start to get real social density, you didn't have to do everything just point to point.

And third, critically: the money ran out. Instead of entrepreneurs saying "I'm going to start this new little web service, and I'm going to go raise $5 million in venture capital, and I'm going to have this big business plan," people had to to ask themselves, "What's a cheap way to do this? What's a cheap way to accomplish my goal?" And, very often, the cheap way was to get the users involved. And once we started down that path, the possibilities just opened up.

Jon Lebkowsky: It was interesting to me that people didn't just throw up their hands and walk away, when there really was no money flowing. People who wanted to innovate, and who wanted to publish content online, all hung in, and were finding ways to do it. They were passionate about it.

Clay Shirky: Absolutely. And certainly a lot of people who rushed in in the late 90s, when it looked like there was free money, rushed back out again. But the people who were left cared enough about some other goal than being dotcom millionaires that they stuck with it. And very often the goals that were left, when the people who were seeking a quick buck were gone, were goals that had real social ramifications. These were people who wanted to make the world better in some way or other, rather than just figuring out a cheaper way to deliver plane tickets.

Jon Lebkowsky: You mention how much higher adoption was by 2000, and of course we've seen it increase persistently since then, so that pretty much everybody's online now. How does that change things, having this pervasive adoption of the web?

Clay Shirky: This is actually one of the things that first led me to try to describe the social patterns that ultimately ended up in this book.

There's a big difference between having some people online and having most people onine. That's a difference that appeals mainly to businesses, now the audience is larger. But there's another difference between having most people online and having everybody online. The advantage of having everybody online is that in your social group, if everybody is online, then you can take it for granted that you can use online tools to coordinate the life of that group.

Small social groups have very high density. In a group of five or six people, pretty much everybody has an interface to everybody else. That's a lot of interface. If even a couple of those interfaces can't be bridged by email or instant messaging, then people will default to the most inclusive possible technology, which prior to the Internet was the phone.

If you were under 35 in the year 2000, and you made more than $35,000 a year, you were almost certainly online and so were your friends, and you could start to take it for granted that you could use the Internet to coordinate your business life and your social life. You could use it to coordinate visits to church, group buying pools, anything that involved a group. Suddenly it became possible, and not because the technology was in place; the technology had been in place for years. It was because the social density had finally caught up with the technology.

Jon Lebkowsky: With Metcalfe's Law and Reed's Law, you're really talking about an increase in potential value that can be realized as real value every day.

Clay Shirky: And the funny thing about the relationship between social applications and Metcalfe and Reed's Laws is that social applications actually trailed them early on, because people don't want to adopt technologies that cut out some members of the group. Why would you use something that excludes some members of the group? But once social density kicks in, social applications actually overperform Metcalfe's Law, as predicted by Reed's Law, because the Internet isn't just about point to point connections, the way Metcalfe's Law is. It's also about group connections.

There was a famous example of this in the attempt to put MetroCards – to put digital card readers – in the New York City subway system. There was a very grim interim report from the Department of Transit, because they were using the token system and the MetroCard system at the same time, saying we've wired 80% of the stations, but we're not seeing 80% of the users use MetroCards. "Oh woe is me, woe is me, this whole thing is potentially a disaster."

And then you read on a little farther, and you realize they hadn't put the MetroCard Readers in Times Square or Union Square yet, which are two of the busiest subway stations. So as long as anybody had to use a token in any station, they weren't going to switch to the MetroCard. Social applications work exactly like that. Merely getting 80% of the people in your business on email meant that there were still significant conversations that you couldn't have online. And so people wouldn't make the switch.

Jon Lebkowsky: Well, sure. If you have a key member of your team or your group who just can't or won't adopt, just can't get it, it just can't work. You see this a lot with wiki. People want to use wiki for collaboration, but out of a dozen people in their group, three people are just totally wiki-resistant, just don't get it.

Clay Shirky: That's exactly right. And you bring up another important point. It's not just the availability of the technology, it's the mental availability of the user. If you've got the web, you can get access to a wiki, but if you've decided you are, as you say, wiki-resistant, it doesn't matter. This is one of the many reasons that groups of young people overperform groups of older people, even given the same technology. In addition to access to the tools, just the set of the functions that go into doing the job – it's more present among people who are more familiar with the tools.

Jon Lebkowsky: You talk quite a bit about public vs private, and the way we're using the web for everything – we all have the same tools to publish in a fairly sophisticated way and we're publishing in public, but not everybody is publishing with the same intention.

Clay Shirky: This is really a reply to all of those media outlets who are writing disparagingly about user-generated content, saying that the content of a weblog is dreck that no one would bother to publish in the print world. All of which is true, but irrelevant, because, of course, the people who are publishing the little observations about their trip to the mall in LiveJournal – they're not talking to you.

The really big change here is that we've got a medium which scales from small groups – me talking to a group of my friends – all the way to "now I am making a public declaration." And because previously, we had a world where, if somebody said "I love you" on the phone, you knew it was meant for you. And if somebody said "I love you" on the TV, you knew it was specifically not meant for you, because the mode of carriage lets us figure out how that message should be interpreted.

And that's now broken. There are people having relatively personal conversations with their friends, yet they're doing it in a public medium. But that's no different from sitting around talking with friends in the food court at the mall. If you want to go down and find a group of teenagers chatting to each other at the mall, you can sit at the next table over and listen in, but then it's pretty clear in that situation that you're the weird one.

What we don't yet have is a set of social norms for figuring out – in a medium like the web, which scales from intimate personal address all the way to full publication – which messages we should be paying attention to and which messages we should be ignoring.

Jon Lebkowsky: When you mention friends, it makes me think about how we've started to use "friend" as a verb...

Clay Shirky: Yeah, I'm going to friend you – yes, exactly.

Jon Lebkowsky: So are we changing the meaning of that word, of what it means to be a friend.

Clay Shirky: I don't think we're changing it so much as we're adding to it, which is to say that I think people still have a sense of the old meaning of friend, as someone you would do a favor to if they were in some real trouble. We still keep that meaning around. I don't think that sense has been denatured, but I do think that the word friend now includes someone who sent you a message on Facebook, and you friended them because why not?

There was an interesting period during the dominance of Friendster where people would talk about their friends, and then their friendsters, and their friendsters were people who they were friends with only on that site. So we may see some growing subtlety in people being able to signal, "Yeah, this person is actually a friend of mine, whereas that person is only a contact I have on Facebook."

Jon Lebkowsky: Another major change I noted around 2000, when I first started using Ryze, and for all those years before that – I had been online by then for a decade or more – and I couldn't see my online friends. And then Ryze created a social network platform where anyone could easily upload digital photos, and at the same time digital photos were more available, because digital cameras were coming out. Suddenly you had visual reference, and today nobody really thinks about whether they know what their "friends" that they never met face to face actually look like, because everybody has a pile of pictures online at Ryze or Flickr or Facebook.

Clay Shirky: Yeah, what we know about those people has been transformed.

Jon Lebkowsky: The experience seems to have more depth now than in the nineties, even though we had really powerful experiences that were text-based. Now we have so much more that we can do.

You said at one point in one of your chapters that our social tools are not an improvement to modern society, they're a challenge to it? What were you thinking about there?

Clay Shirky: For the last hundred years, the key organizational conversation was, are big challenges better taken on by the state, by the government, raising taxes and spending the money, or are they better taken on by businesses operating in the marketplace. But the dot dot dot at the end of that sentence was because obviously people can't get together and do these things for themselves.

There was a basic assumption, both in capitalist and communist theories of large scale action, that the complexities of ordinary life would defeat the ability of groups to come together and do things on their own.

It seems to me that what's happened is that this thesis has now been rendered false in a surprising number of cases, and, maybe more importantly, a growing number of cases. There are places now where people are coming together and creating value for one another without doing it in either the framework of government or the framework of business.

I gave a talk at Supernova, a brief talk on the Perl programming language. I was pointing out that the Perl programming language, which has been an absolute mainstay of the web from the earliest days, is held together by love. It's not held together either by government intervention or by corporate investment. It's held together because a bunch of people love Perl, and more importantly, they love one another in the context of Perl. They like being part of a community that makes this language work, and work better.

The idea that this could create a programming language as good and as powerful and as ubiquitiously-used as Perl is new. One of the big shifts, and one of the reasons I wrote this book – this is a non-techie book, instead of writing for my usual audience of folks, programmers and engineers, I've actually tried to write it for my Mom – to explain why this is a big deal. One of the things I think is happening, is that the pattern of groups being able to come together and do things for themselves is now spreading outside of the technical and geek communities, and is becoming a general social capability.

Jon Lebkowsky: You mentioned love as a motivator and social glue. Do you have a technical, operational definition for love?

Clay Shirky: You know, I don't. (Laughter.) I have the same definition that the supreme court used to have for pornography, which is I know it when I see it.

That's actually an interesting question, I should take that seriously. Right now it's defined largely by negation, which is to say, when people come together and do things together without obviously being motivated by either requirements or payments... if I'm doing something, and it's not because my boss told me to do it, or I'm doing something and it's not because I think I'll get more money at the end of the day, if I do it – then almost by definition I'm doing it for love.

That strikes me as kind of an unsatisfactory definition, and there is so much work yet to be done on motivation. In part it hasn't been done because neoclassical economics assumes that most human motivations can be backed into money, so that you can use money as this kind of universal calculator, even if there's no money involved in the actual transaction. And we now know that to be false, from a lot of research and behavioral economics. There are some jobs where people will do the job better if they're not paid, which is to say if they sense they're being asked for a favor and are participating in community building, they'll actually do a better job than if they're simply given money to do the work.

Jon Lebkowsky: Isn't this like the work of Etienne Wenger, Nancy White and John Smith with communities of practice?

Clay Shirky: That's exactly right. Communities of practice is one of these great patterns of demonstrating, to the consternation of many neoclassical economists, the degree to which people will go out of their way to help each other with no obvious return.

The community of practice that I love is the high dynamic range (HDR) photography people on Flickr. Back in the old days, if some new photographic technique came along, it would take 5-7 years to spread from someone's photo studio to photo magazines, and finally to widespread visibility in Popular Photography, and the average darkroom.

You could see the high dynamic range technique, where you take multiple exposures of the same scene and combine them to get the brightest brights and the darkest darks, rip through Flickr, where people were posting these photos, and someone would come along, and say "Oh, my god, that's the greatest photo I've ever seen, I love it. How did you do that?" And then you had these threads that were thousands and tens of thousands of words long with pointers to external software, and other people posting images in the thread that would help illustrate things.

This community sprung up around high dynamic range photography, and they essentially explained it to themselves in the course of about three months. HDR photography went from being something that a handful of people knew how to do to a general technique that any photographer who's willing to spend an afternoon on Flickr could pick up and understand. And the speed of that spread wouldn't work if money were involved.

The awareness and the growth in expertise actually happened faster because people weren't asking for payment in return for value. They were asking to participate in a community that loved this stuff. I think we're going to see a huge amount of experimentation with those kinds of advantages, which will appear in all kinds of new places.

Jon Lebkowsky: In my own work, I've been looking at and thinking about how these sorts of things happen, especially in business environments. And we know that they do happen, and now there's a body of work... like Verna Allee and the value network people, who are saying, "We don't really have a way to capture that value, or quantify it, so how do we do that?" Are you familiar with the value networks body of work?

Clay Shirky: Yes, and one of the really interesting patterns that jumped out at me, doing a book about large scale collaboration, is that very often really large-scale collaboration, whether it's a Wikipedia or Linux or what have you, involves a small number of people who care an enormous amount, and then a large number of people who only care a little bit, but who are participating, who are adding their value to the overall work product.

What the value networks work seems to be to point to is ways in which you can create some of this kind of benefit without having everybody participating in a formal community of practice, and also getting more heterogenous kinds of skills and values involved. Everybody who's in the HDR community of practice on Flickr is (a) a photographer and (b) experimenting with HDR. But once you get to something like Wikipedia, there are people who are fact checkers, and there are people who are sentence editors, and there are people who are content creators. You get a kind of division of labor that's really quite different, and makes the whole more valuable, in part because of those differences.

Jon Lebkowsky: There's a whole interesting question about kibitzing, about lurkers in a community and the extent to which they actually add value. And, of course, many lurkers are never 100% lurkers. Even if they don't uncloak in public, they'll email people who are having conversations, and drive things along. There was something in your writing, an idea that suggests the shape of a fried egg, where you have a cluster of real activity in the middle, and you have a sort of supportive community around it that's less involved, but still contributing.

Clay Shirky: I haven't used the fried egg analogy, but I love that. And the observing community is the pool from which the participants are drawn, even if a majority of the people in the observing community never become participants.

Jon Lebkowsky: We've been thinking about that in Austin, where there's an active community of bootstrap entrepreneurs. One thing we've been talking about recently, that I had been thinking about for a while, is the idea that you could potentially do the larger things that people normally grow monolithic corporations to do... that you could cluster and aggregate networks of smaller companies to collaborate to do these larger things. Instead of having a big company with departments, you just have a network of companies that have figured out how to organize so that they can really depend on each other. And that gets to the issue of trust, which you talk about...

Clay Shirky: What you just said is, in my mind, the key piece of economic analysis, which is when the transaction costs are down, then the ability of smaller groups to find one another and bind themselves to one another as needed goes up. And once you get those two things happening at the same time, you can actually start figuring out when you'd be better off decreasing the size of the group and increasing the discoverability of the interface.

Jon Lebkowsky: How would this relate to the question of trust, and how you get the group to come together and to work? How would that relate to your trinity of plausible promise, effective tool, acceptable bargain...?

Clay Shirky: A lot of it starts with the plausible promise, with telling people, if they come together, they can actually do something successfully. And very often modest success matters more than audacious goals.

If you look at the original document proposing either Wikipedia or Linux, the most striking thing is how incredibly modest the original requests were. But that was enough. It was enough to get people involved. And then, if you can do that, and in many ways that's the hardest thing to do... then you get to the problem of figuring out which tool to use, and what bargain to use.

The tool is relatively simple, which is to say there's a few classic misakes to avoid – if you want people to converge on some sort of shared work product, don't launch a mailing list. If you want people to diverge and generate lots and lots of competing ideas, don't launch a wiki. But fitting the tool to the job is in many ways a matter of looking out and seeing who else has got a problem similar to yours and what tools are they using.

The bargain is the hardest one of all, particularly around this idea of subdividing into smaller groups that then interact with one another. Because the bargain really says, "what are the users' expectations of one another over the long haul? – as opposed to anything that the site's founder or host can promise.

Getting the culture right is really an art, and not a science... which is to say that your early culture is going to be set by the people who happen to come around, and you've got to work with that while, at the same time, keeping your eye on wanting to have a culture that can scale up over the long haul.

Kathy Sierra has a fantastic example from Java Ranch, which was a site meant to host friendly conversations among Java programmers. They wanted to get away from the kind of supercilious snarkiness that characterizes a lot of technical communities. So they have a terms of service you have to accept to be part of the community, and the actual terms of service, in its entirety, is "Be nice."

And that was their way of saying, "We can't enforce every little jot and tittle of user interaction. We know people are going to say things that may upset one another. All we're going to say is, our standard of behavior is that you should be nice to each other, and if we see that not happening, we're going to intervene."

It's such a beautiful rebuke to all the lawyerese of you can't do this or that, where people try to enumerate everything that could go wrong. Because what they did, I think, in that model, is that they managed to streamline the kind of thing that has to go into a long-term user bargain, into a very simple to understand concept, and I'd like to see more of that and less of the "we had the lawyers wrote the terms of service, and suddenly it's fifteen printed pages."

Jon Lebkowsky: We have everybody online now publishing with the same forms of media, everybody's got access to everything, and you've got mass communication on one end of the spectrum, and on the other end you have very intimate but still public conversations, which is kind of interestingly weird. Is that a gradual continuum? How much are people really confused about the kinds of conversations they're having?

Clay Shirky: This is an experiment I want to see run, but I think this is a very interesting question. Here is my hypothesis: that one of the things that people create some kind of really deep mental model for is modes of communication. People my age and older have a very good sense of when to call someone on the phone, and when to send them a personal letter, and when to go see them. But we don't have such a good sense of when to email them, or IM them, or Twitter or what have you, because all of that stuff was invented after we had already solidified our sense of the media landscape. All of those things are still new.

One way to test this would be to see whether fifteen year olds today have a literally more intuitive sense of when to call, when to SMS, when to email, and when to IM. And I think they do. I think that the confusion around media is largely with people who have grown up in the environment we grew up in, where television is one thing, whereas the phone is another thing. The medium that reaches groups isn't a communications medium. The medium that is a communications medium doesn't reach groups. When all that has gotten overturned, it looks strange to us that people having group communications in a public medium – you know, these half a dozen friends, are all Live Journaling one another about their trip to the mall, or the party last Friday. But to those kids I don't think it seems weird at all. And if that's true, then that's the kind of generation gap that came up around the use of the telephone or the use of the telegram, and I think it's something society will have to weather for thirty years. If I'm wrong about that, which is to say, if increased numbers and kinds of media actually lead to increased social confusion, then I think that society is going to have to develop some formal methods of etiquette in order to figure out how to manage all of this proliferation of new communications options we've gotten.

Jon Lebkowsky: Twitter has turned out to be a very interesting communication space. I really didn't get it, didn't have the right experience of it for the longest time, because I was just using the web interface. Occasionally I would activate it for my phone if I was stuck in traffic and bored, and wanted company.

But I recently started using Twitter via IM using GTalk. and that's an entirely different experience, in that you really get the flow of conversation, seeing comments as they're posted.

One of the interesting things about Twitter is that you have this continuum that we were talking about... you have some people who come to Twitter only because they want to broadcast, to announce something to the world, or at least to their network. So they'll show up and post a url, "this is my latest blog post" or whatever. But they don't really hang out and have conversations. More often, though, Twitter users have public conversations where they're talking either to everybody, or to a specific person through a public reply. And you have people who want fairly intimate conversations and will go to direct messages, which are private. So there's this whole spectrum of experience you can have on Twitter.

Clay Shirky: I think like everybody, when it came out, I started playing with it, but it seemed to me that most of the action and gone private, but I had not tried to use the GTalk interface. I'll have to give that a try.

Jon Lebkowsky: What is the problem of filtering, and how has it changed? You talk about a priori filtering in the publishing world, and how filtering is now more after-the-fact.

Clay Shirky: The problem with filtering is, now that there's not bottleneck for production, there is no way to filter in advance. You can't filter the good from the mediocre in advance, simply because it's too expensive. No one has the cash needed to simply keep on top of everything that's coming down the pipe, because now everybody has a pipe.

So filtering has now gone to this post-hoc thing. As good as it has gotten, with things like PageRank and del.icio.us and Technorati, and so forth, we're still in a world where the average experience of wandering around the web is of being exposed to all kinds of things that are really kind of irrelevant. The searching and sorting problem hasn't yet settled itself down.

One of the things I try to explain to people when they say how much junk there is on the web is to use the analogy of a book store. You go into a book store and your experience of the book store is, "oh, I went right to the section on philosophy, and I went right to the books on Plato, and there they were." So I know that there's all this great literature in the bookstore.

But if you picked up that book store, and you shook the contents out into the street, and you waded in and started picking books at random, you'd find Chicken Soup for the Hoosier Soul and Love's Tender Fury, and all of this stuff. In fact, our experience of the book store as being a site of a lot of really good content is in large part because we're really good at ignoring 99% of what's in there. If you're not going to the book store for self-help books, you don't have to look at them.

And because the filtering problem on the web is so enormous, and because we're still in relatively early days of figuring out how to solve it, we can't yet get to that happy state where the stuff I'm not interested in doesn't show up. It takes a much more active stance in terms of searching and grooming and so forth to zero in on the good stuff.

So it seems to me that the problem of filtering is going to remain one of the key problems of the age, for the next few decades, in part because the volume of material people are producing is still going up. And once we get a relatively good solution for filtering the web, for example, along comes Twitter – here's this new medium that we don't have these filtering tools for. How do you figure out what to read and what to ignore and what to save and what to throw away, and so forth? That problem is coming up now, and is going to keep coming up over and over again for as long as we're on this ride. We keep going to a place where there's so much more content this year than last year, so a lot of our old strategies are broken.

Jon Lebkowsky: It seems to me that one of the real problems of filtering is that, to the extent we feel that we have to filter and set up filters, that we're liable to exclude things that we didn't know we would find interesting.

Clay Shirky: That's right. And designing filters with a certain amount of serendipity involved is a key part of this. But even then, even with some serendipity, it is so easy to have the amount of content radically overflow any strategy that we've got for sorting the stuff that we care about from the stuff we don't care about. Even with a serendipity meter built in, we still have to work hard to get this right.

Jon Lebkowsky: Where do you see things going? You've written a good analysis of where we are, but what comes next?

Clay Shirky: The ladder that I develop in the book is how much does the individual have to coordinate themselves with the group to get an effect. So the simplest thing is sharing, right? Flickr, del.icio.us, YouTube, Napster... my ability to share with millions of others and then for all of us to profit from that requires very little coordination from me. That pattern is very easy to bootstrap.

The next pattern up is collaboration, where there actually is some more coordination required between me and other people. This is Open Source software, this is Wikipedia, and so on.

The pattern that strikes me as being most radically different from what we've had before is collective action, the pattern where the group comes together, and stands or falls depending on the actions of the entire group. Every member of the group is affected by the action of the group as a whole. I spent a lot of time looking, in particular, at the political prostests in Belarus that are using the flash mob model for protesting. It seems to me that the collective action model, where the group isn't just a loose collection of individuals, it's actually a unit, has not yet seen a lot of traction. There have been some interesting experiments, but most of the interesting work there is still in the future. And that's what I'm watching out for – what's coming with the future of collective action, because I think there's a huge amount of work still to be done there.

Jon Lebkowsky: When we were doing the Extreme Democracy book, and as a precursor to that we were having the emergent democracy conversation, the Joi Ito thing. The big question for us was emergent leadership.

Clay Shirky: Yes.

Jon Lebkowsky: How does that work. How do we actually have leaders emerge, and how does the group know – how does a flock of birds, for instance, know which bird is in the lead at any given time.

Clay Shirky: One of the big surprises about the Open Source movement is how many of the projects had a benevolent dictator for life at their head. There are a few that don't, like the Apache Foundation. But Perl and Python and Ruby and Linux and on and on had the charismatic, technically adept founder at the head. How people find and identify those leaders, and what lessons we can take from the technical community to the nontechnical community, I think is a really big open question.

Jon Lebkowsky: We had a sort of laboratory for thinking about this with Howard Rheingold's Electric Minds, the business he created around an online community. When Howard realized that he needed to do something with Electric Minds, that it really wasn't working as a business, the question was, where does it go? He got a buyer who agreed to honor the community. Then the question was, if Howard was going to become just another community member, who was going to lead? It's a long story, but in the end, the community found that the benevolent dictator model seemed to work very well.

Clay Shirky: Yes, absolutley. And it locks the "benevolent dictator" out of participation. Stacy Horn, who founded Echo, had this problem. She could not go out and socialize with her own users, because she was the owner, and everybody kind of behaved weirdly around her. So she ended up having to mainly consign herself to conversations that were only populated by people who remembembered when Echo was just a few hundred people, so that they wouldn't treat her so weirdly. But she couldn't, in fact, be just an ordinary member of the community.

Jon Lebkowsky: That's interesting. Howard's next thing, of course, was his semi-private Brainstorms community, where he's the door. Everybody comes through him, so he knows everybody who comes in. That weirdness that Stacy Horn experienced may have been there to some extent with Electric Minds, but it's absolutely not there at Brainstorms.

Clay Shirky: No, because you're already going through Howard on the way in, so you're sort of aware of that.

Jon Lebkowsky: Yeah, and even though he's still kind of the benevolent dictator, he's a member of the community. The problem you run into is when you have some people in the community that feel you need to throw a person out, because they're misbehaving – this has been a big deal on the WELL, for instance. In one case, there was a guy who was trashing the commons on the WELL in a big way, but because of the strong tradition of free speech on the WELL, the managers didn't want to just throw him out, and there was a quandary – what do you do about this guy? Because you didn't really have a strong benevolent dictator who would just throw him out. You had to have a process, and the process extended the pain.

Clay Shirky: It's a dilemma, deciding when the needs of the group trump the needs of the individual. And it's a tough moment, nobody likes that moment. It's anti-democratic in one way, and yet all groups require that, because all groups acquire the kinds of trolls that you're talking about here.

Jon Lebkowsky: The tragedy of the commons.

Clay Shirky: Yeah, exactly.

Jon Lebkowsky: And you made a strong case, I think, in your writing, for the need for governance. Obviously there is always some governing principles in any group, whether they're formal or informal. It's a problem when we try to put those principles aside.

So in closing, how do you think governance is going to play out in the future? The Internet is a big laboratory for governance models. What impact could that have on our actual, formal mechanisms for governance?

Clay Shirky: The biggest impact will be if we find some way to defer to groups, to allow groups to come together and make some choices for themselves that the government defers to. Or, if we start regarding the output of groups as being legitimate expressions of the will of the people.

Many people have floated this idea of a policy wiki, or the notion of doing the national budget using the wisdom of crowds. Those experiments would be, I think, the most radical. On the way to that, even before the really radical stuff, I think the big change is going to be just the number of times that people start to pull together and have success, as with this airline passengers bill of rights – after the industry fought it off for eight years, suddenly in eight months a little group came from nowhere with no budget and no staff, and actually succeeded in rewriting the law. [Author's Note: Since this conversation, the 2nd Circuit Court has struck down the NY State Passenger's Bill of Rights. Now the test of the people vs the airline industry moves to Congress and the Supreme Court.]

I think the big change in government is going to be with people getting some sense that if they come together, they can actually do things for themselves.

Jon Lebkowsky: I think that's really important. I think the problem that we have, even within the Democratic party, is that there's a set of people who'll say, say screw the will of the people, the people don't know what they're talking about. We know what's best for them.

Clay Shirky: The superdelegates, in a way, were set up specifically to keep people from sending unelectably liberal candidates into the general election. But it's such a bad fix for that problem. Now that that system might actually kick in, I think everybody in the Democratic National Committee is trying to find a way to back away from it. Because I think the amount of attention, and the number of new voters they turned out... if they were to actually have the election go to someone who hadn't been ratified by the people, I think it would be a catastrophe.