Virtual world stats as misleading as a dating profile

CLAY SHIRKY — The conversation about Linden Lab's spurious population figures continues. If you want to see how hard it is to think clearly about how many users virtual world Second Life has, given the fog of its maker's numbers, consider the case of Kevin Murphy of ComputerWire.

In the memorably titled Bloggers Are All Ignorant Scum, Mr. Murphy launches a vigorous defense of reporters covering Second Life, in response to Naming Names: The Tech Reporters Who Flack for Second Life. In his response, Mr. Murphy explains how trivial the Naming Names objection was, what the Linden numbers actually mean, and quotes several reporters who got the story right.

Unfortunately, Mr. Murphy turns out not to understand the Linden numbers either. As a result, he, and all the reporters he quotes, make the very mistake documented in Naming Names.

There are only two things you really need to understand about the Linden story. First, any Residents figure from Linden is unrelated to the population of Second Life (because a Resident isn't a user) and growth in the Residents number is not directly tied to growth in users (because existing users can create new Residents). Second, any reporter who publishes a population figure for Second Life is snowing you on Linden's behalf, since Linden never tells anyone how many regular users Second Life has.

Murphy misunderstands the argument of Naming Names. He explains the core complaint of the piece this way: "Second Life says it has about 2 million subscribers, but that only about 800,000 of them have logged on in the last two months. About four reporters carried the first number without reporting the second. That's it."

Except that isn't it. Linden does not say it has two million subscribers. It is Mr. Murphy himself who helpfully but wrongly supplies subscribers as a synonym for Residents. And the issue isn't one Residents figure over another — they're all junk. Residents does not refer to subscribers, members, citizens, customers, participants, or users. It is a count of signups, not people, it includes unknown quantities of failed logins, double-counting, and abandoned accounts, and it has varied over the life of the company. As a result, it is worthless as a measure of adoption, even as a proxy for growth.

The gap between how little we know and how much Linden seems to be saying is what trips us up. (This is the same trick as the guy who cleans the boats putting "Has keys to the marina" in his Match.com profile. It's not lying, exactly...) The issue is not fraud but lousy reporting.

Had Linden labeled signups with some made-up word like Surnits, and then went around telling reporters they'd added a million surnits in 8 weeks, reporters would have said "Great. What's a surnit?" Now Linden doesn't want to answer that question (boy, do they not want to answer that question), but if they said nothing at all, they might have been asked anyway. So instead, they provide an evocative but uninformative unit, Resident, that forestalls the question by making the hearer think they've heard the answer.

Reporters want figures, and Residents sounds kind of like subscribers or members or something, so they replace the former with the latter. (The reporters Murphy quotes all make this mistake: Cade Metz/PC Magazine, Bill Thompson/BBC News, Daniel Terdiman/CNet, Don Tapscott and Anthony Williams/Globe & Mail, Shankar Gupta/MediaPost.) This happens unconsciously, so the reporters never realize that they are inventing statistics that Linden never supplied and which they never checked. Linden is neither transparent or opaque — they are camouflaged, and brilliantly so.

The key point of Murphy's piece is that the business press is being unfairly criticized for not reporting this story well: "We're apparently credulous, willing to prostitute ourselves for Linden Labs, because we're naive starry-eyed youngsters who haven't been around long enough to cut through obvious hype."

Given that he hasn't been able to figure out what the Linden numbers mean (and more importantly, don't mean) even when writing a story entirely based on them, he comes close to living up to that caricature.

Contact information for this author is not available.