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.

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

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.

You talk quite a bit about public vs private, and the way we're using the web for everything

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.

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.

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 and Nancy White with communities of practice?

Clay Shirky: That's exactly right.

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.

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.

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...

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.

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.

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.

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.

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.

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.

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.

So the simplest thing is sharing,

The next pattern up is collaboration, where there actually is some more coordination required between me and other people.

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.

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.

The big question for us was emergent leadership.

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?

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.