CLAY SHIRKY — Second Life, the much-hyped virtual world backed by Benchmark Capital, is heading towards two million users. Except it isn't, really. We all know how this game works, and has since the earliest days of the web:

Member of the Business Press: "How many users do you have?"

CEO of Startup: (covers phone) "Hey guys, how many rows in the 'users' table?"

[Sound F/X: Typing]

Offstage Sysadmin: "One million nine hundred and one thousand one hundred and seventy-three."

CEO: (Into phone) "We have one point nine million users."

Someone who tries a social service once and bails isn't really a user any more than someone who gets a sample spoon of ice cream and walks out is a customer.

So here's my question — how many return users are there? We know from the startup screen that the advertised churn of Second Life is over 60% (as I write this, it's 690,800 recent users to 1,901,173 signups, or 63%.) That's not stellar but it's not terrible either. However, their definition of "recently logged in" includes everyone in the last 60 days, even though the industry standard for reporting unique users is 30 days, so we don't actually know what the apples to apples churn rate is.


At a guess, Second Life churn measured in the ordinary way is in excess of 85%, with a surge of new users being driven in by the amount of press the service is getting. The wider the Recently Logged In reporting window is, the bigger the bulge of recently-arrived-but-never-to-return users that gets counted in the overall numbers.

I suspect Second Life is largely a "Try Me" virus, where reports of a strange and wonderful new thing draw the masses to log in and try it, but whose ability to retain anything but a fraction of those users is limited. The pattern of a Try Me virus is a rapid spread of first time users, most of whom drop out quickly, with most of the dropouts becoming immune to later use. Pointcast was a Try Me virus, as was LambdaMOO, the experiment that Second Life most closely resembles.


I have been watching the press reaction to Second Life with increasing confusion. Breathless reports of an Immanent Shift in the Way We Live® do not seem to be accompanied by much skepticism. I may have been made immune to the current mania by ODing on an earlier belief in virtual worlds:

Similar to the way previous media dissolved social boundaries related to time and space, the latest computer-mediated communications media seem to dissolve boundaries of identity as well. [...] I know a respectable computer scientist who spends hours as an imaginary ensign aboard a virtual starship full of other real people around the world who pretend they are characters in a Star Trek adventure. I have three or four personae myself, in different virtual communities around the Net. I know a person who spends hours of his day as a fantasy character who resembles "a cross between Thorin Oakenshield and the Little Prince," and is an architect and educator and bit of a magician aboard an imaginary space colony: By day, David is an energy economist in Boulder, Colorado, father of three; at night, he's Spark of Cyberion City—a place where I'm known only as Pollenator.

This wasn't written about Second Life or any other 3D space, it was Howard Rheingold writing about MUDs in 1993. This was a sentiment I believed and publicly echoed at the time. Per Howard, "MUDs are living laboratories for studying the first-level impacts of virtual communities." Except, of course, they weren't. If, in 1993, you'd studied mailing lists, or usenet, or irc, you'd have a better grasp of online community today than if you'd spent a lot of time in LambdaMOO or Cyberion City. Ou sont les TinyMUCKs d'antan?

You can find similar articles touting 3D spaces shortly after the MUD frenzy. Ready for a blast from the past? "August 1996 may well go down in the annals of the Internet as the turning point when the Web was released from the 2D flatland of HTML pages." Oops.

So what accounts for the current press interest in Second Life? I have a few ideas, though none is concrete enough to call an answer yet.

First, the tech beat is an intake valve for the young. Most reporters don't remember that anyone has ever wrongly predicted a bright future for immersive worlds or flythrough 3D spaces in the past, so they have no skepticism triggered by the historical failure of things like LambdaMOO or VRML. Instead, they hear of a marvelous thing — A virtual world! Where you have an avatar that travels around! And talks to other avatars! — which they then see with their very own eyes. How cool is that? You'd have to be a pretty crotchety old skeptic not to want to believe. I bet few of those reporters ever go back, but I'm sure they're sure that other people do (something we know to be false, to a first approximation, from the aforementioned churn.) Second Life is a story that's too good to check.

Second, virtual reality is conceptually simple. Unlike ordinary network communications tools, which require a degree of subtlety in thinking about them — as danah notes, there is no perfect metaphor for a weblog, or indeed most social software — Second Life's metaphor is simplicity itself: you are a person, in a space. It's like real life. (Only, you know, more second.) As Philip Rosedale explained it to Business Week "[I]nstead of using your mouse to move an arrow or cursor, you could walk your avatar up to an (AMZN) shop, browse the shelves, buy books, and chat with any of the thousands of other people visiting the site at any given time about your favorite author over a virtual cuppa joe."

Never mind that the cursor is a terrific way to navigate information; never mind that Amazon works precisely because it dispenses with rather than embraces the cyberspace metaphor; never mind that all the "Now you can shop in 3D efforts" like the San Francisco Yellow Pages tanked because 3D is a crappy way to search. The invitation here is to reason about Second Life by analogy, which is simpler than reasoning about it from experience. (Indeed, most of the reporters writing about Second Life seem to have approached it as tourists getting stories about it from natives.)

Third, the press has a congenital weakness for the Content Is King story. Second Life has made it acceptable to root for the DRM provider, because of their enlightened user agreements concerning ownership. This obscures the fact that an enlightened attempt to make digital objects behave like real world objects suffers from exactly the same problems as an unenlightened attempt, a la the RIAA and MPAA. All the good intentions in the world won't confer atomicity on binary data. Second Life is pushing against the ability to create zero-cost perfect copies, whereas Copybot relied on that most salient of digital capabilities, which is how Copybot was able to cause so much agida with so little effort — it was working with the actual, as opposed to metaphorical, substrate of Second Life.

Finally, the current mania is largely push-driven. Many of the articles concern "The first person/group/organization in Second Life to do X", where X is something like have a meeting or open a store — it's the kind of stuff you could read off a press release. Unlike Warcraft, where the story is user adoption, here most of the stories are about provider adoption, as with the Reuters office or the IBM meeting or the resident creative agencies. These are things that can be created unilaterally and top-down, catnip to the press, who are generally in the business of covering the world's deciders.

The question about American Apparel, say, is not "Did they spend money to set up stores in Second Life?" Of course they did. The question is "Did it pay off?" We don't know. Even the recent Second Life millionaire story involved eliding the difference between actual and potential wealth, a mistake you'd have thought 2001 would have chased from the press forever. In illiquid markets, extrapolating that a hundred of X are worth the last sale price of X times 100 is a fairly serious error.

Like video phones, which have been just one technological revolution away from mass adoption since 1964, virtual reality is so appealingly simple that its persistent failure to be a good idea, as measured by user adoption, has done little to dampen enthusiasm for the coming day of Keanu Reeves interfaces and Snow Crash interactions.

I was talking to Irving Wladawsky-Berger of IBM about Second Life a few weeks ago, and his interest in the systems/construction aspect of 3D seems promising, in the same way video phones have been used by engineers who train the camera not on their faces but on the artifacts they are talking about. There is something to environments for modeling or constructing visible things in communal fashion, but as with the video phone, they will probably involve shared perceptions of artifacts, rather than perceptions of avatars.

This use, however, is specific to classes of problems that benefit from shared visual awareness, and that class is much smaller that the current excitement about visualization would suggest. More to the point, it is at odds with the "Son of MUD+thePalace" story currently being written about Second Life. If we think of a user as someone who has returned to a site after trying it once, I doubt that the number of simultaneous Second Life users breaks 10,000 regularly. If we raise the bar to people who come back for a second month, I wonder if the site breaks 10,000 simultaneous return visitors outside highly promoted events.

Second Life may be wrought by its more active users into something good, but right now the deck is stacked against it, because the perceptions of great user growth and great value from scarcity are mutually reinforcing but built on sand. Were the press to shift to reporting Recently Logged In as their best approximation of the population, the number of reported users would shrink by an order of magnitude; were they to adopt industry-standard unique users reporting (assuming they could get those numbers), the reported population would probably drop by two orders. If the growth isn't as currently advertised (and it isn't), then the value from scarcity is overstated, and if the value of scarcity is overstated, at least one of the engines of growth will cool down.

There's nothing wrong with a service that appeals to tens of thousands of people, but in a billion-person internet, that population is also a rounding error. If most of the people who try Second Life bail (and they do), we should adopt a considerably more skeptical attitude about proclamations that the oft-delayed Virtual Worlds revolution has now arrived.