Why Second Life Numbers DO Matter
Over at Valleywag, where sanctimony can sometimes reign, a professor named Clay Shirky has been getting some attention for criticizing how Linden Lab reports the numbers for its fast-growing Second Life virtual world. But along with his acidic criticism of Linden Lab's numbers, he seems even more eager to direct vitriol at the business press, including myself. His contempt is aimed at articles in various publications and websites, including a column I wrote in November at Fortune.com.

Shirky is outraged that writers should report without comment the number of registrants Linden Lab has accumulated for Second Life, the figure the company calls "residents". He is especially offended by references like one I made November 10 to the "1.3 million members" of the service.

To give him the minimal due he deserves, he's correct in noting that my use of "members" was inaccurate. In fact, the "residents" number merely reports how many people have tried to sign up for an avatar they can use in the world of Second Life. As I said in my initial column, the system is very difficult to use. In fact the majority of registrants, it would seem, either give up immediately or else very infrequently return. While millions may have registered, only 15-20,000 are in the world at any given time. Linden Lab CEO Philip Rosedale said in November that three months after registration, about 10% of registrants still log in at least weekly, a figure that has remained consistent even as the service grows.

However, the growth of the resident figure remains astonishing. On January 1 it surpassed 2.3 million, an increase of one million - or 77% - since my piece less than two months ago. Shirky is entitled to think that is not noteworthy, but it is.

More significantly, actual use is rising at a comparably rapid rate, albeit at much lower levels. In fact Linden Lab publishes extensive data about its own growth. These charts show that user hours, number of paying members, land creation, and in-world economic activity are all growing exponentially. For instance, the number of paying residents has more than doubled since June to over 40,000. (You don't have to pay to use Second Life, and many who go there merely to sightsee do not.) User hours per month grew from about 3.5 million in June to 7.5 million in December.

Incidentally, the company publishes enough data there and elsewhere on its website to render Shirky's complaints completely illogical. Linden Lab now routinely reports more data than any other social networking or commercial community site I know of. His umbrage might be more usefully directed at companies like MySpace, which claims something like 135 million user accounts. How many of those are in use? Would MySpace reveal as much data as Linden Lab does?

There's no question that Linden Lab faces a near-crisis as interest in its product so obviously soars. The product is unusable by most casual users. If you aren't lucky enough to encounter a generous guide on your first visit, who helps you understand the basics of movement, navigation, and avatar etiquette, you probably won't come back.

Yet on the other hand, with usage already growing so rapidly, the company may not really want all those newly-interested hordes to get into its virtual world anyway. It already maintains about 5,000 servers, having added more than 800 in the last six weeks or so.

There is a limit to how quickly any system can grow, and Second Life is such a complex product that it may not be able to handle all the users who are interested. The product's source code exceeds 5 gigabytes today. What makes Second Life unique and important is the ability of its residents to create the content there. But a system in which 20,000 simultaneous users can all be moving around making things like houses, trees, hats and even flying penises takes a lot of computing horsepower.

Shirky and other complainants at Valleywag and elsewhere may not agree, but the reason so many business and other journalists throughout the world are writing about Second Life is because they recognize it as something fundamentally new. I will continue to take note of Second Life's novelty and significance.
Posted by David Kirkpatrick 12:53 PM 11 Comments comment | Add a Comment

David, you say However, the growth of the resident figure remains astonishing.

There's a clue to the story in here that I think you are missing -- the growth of the Residents figure _is_ astonishing, yes. One possible explanation is that is that many, many people are interested in Second Life.

Another possible answer, however, is that it that with the dropping of limits on avatars per account and accounts per person, it is the ratio of Residents to users itself that is changing. In addition, with the creation of a free basic account, the number of people who sign up but never log in or log in and quickly bail is also increasing. If these effects are what is driving the astonishing growth in Linden's reported numbers, then the second million Residents represents considerably fewer actual people than the first million Residents did (itself already a small number.)

More importantly, the second million Residents also represents many fewer eventual repeat users.

Reporting on Linden's population growth by measuring a change in the Residents number only makes sense if a) there is a Residents-to-people ratio that is stable across the pool of Residents and b) if that ratio has been stable over time. Neither thing is true.

Here are three questions I'm certain you have never asked Linden Labs:

1. How many people (not Residents) have created an avatar in Second Life?

2. Of that number, how many have returned 30 or more days later?

3. How fast is that latter figure growing?

The fact that Linden exports a lot of numbers about its service is meaningless, since none of their numbers answer those three fairly basic questions, and no one in the press has actually asked.
Posted By Clay Shirky, Brooklyn, New York : 3:02 PM  

I believe that as in everything new, there are people who agree or disagree.

As noticed, registered users (which doesn't show return users) are used mainly by all web sites.

At least Linden Labs shows always the concurrent users, which is a more reliable metric. And the fact is that in the last months this number of concurrent users has increased (I think now it peaks more than 20k users concurrent online). And we should notice, that is not easy for one person with many accounts to login at the same twice (unless they have 2 computers).

So the soar numbers is a fact. If they are due to the media hype or not remains to be seen. My personal opinion though is that unless Linden Labs fail to face their success-infrastracture and scalability problems-, we'll see even more increased numbers.

Beyond all these, Second Life or another virtual world that will come up, is here to stay. That's why so many companies try to adopt the new thing. It might be something that will give new direction to internet. It might not, but nobody wants to miss the train when it leaves the station.

And it's a terrible mistake to compare the current established web sites with something so new in terms of users. To be fair let's remember how static web sites were 15 years ago.
Posted By John Kotinis, Athens, greece : 6:20 AM  

The NY Times baby! This must be for real.
Posted By Dave, New York, NY : 1:16 PM  

Thanks Clay, for your response.

I posed your questions directly to Linden Lab CEO Philip Rosedale yesterday. Here�s what he told me, via email:

�#1- 1,525,670 unique people have logged into SL at least once. This is considering distinct email/payment info as distinct people, rather than IP addresses. [He is checking the unique IP address numbers but suspects they will be comparable.]...So in comparing that to the overall signup number, the difference is created by two sources: alt accounts (cases where one person has multiple accounts), and cases where the person signed up but has never logged in (possibly because of firewall or computer problems).

#2. 252,284 people have logged in more than 30 days after their account creation date.�

For the answer to #3, Rosedale provided a spreadsheet which lists the number of new registrants by month and the number of those who were still returning longer than 30 days later. That data shows the astonishing growth I referred to�registrants in January 2006 were 20,000. In October they were 254,000.

While the percent of registrants still active after 30 days has, predictably, declined a lot since early 2004 when it exceeded 45%, it remains a substantial 15%. Of those 254,000 who registered in October, 39,575 still were active after 30 days. The absolute number of those still returning after 30 days grew 23% for October�s registrants over those who registered in September.

It is hard for me not to be impressed with any service whose active new users are growing 23% a month.
Posted By David Kirkpatrick, Fortune New York : 1:04 AM  

I find it so useful that David Kirkpatrick, far from being the lazy, credulous journalist depicted by Shirky, simply did his homework, then picked up the phone, then asked Philip Rosedale himself to deliver the numbers.

And indeed he concedes that no other social software or game company provides anywhere near this degree of transparency and wealth of information on the statistics page.

Clay Shirky, meanwhile, is still ranting about the misleading statements of 6 weeks ago. Even in that time, SL has grown fantastically.

I was trying to impress on Shirky the importance of the 36,000 paid subscribers who own land a few weeks ago; it's already up to 42,000.

Philip Linden has honestly told residents in recent weeks that he doesn't think retention is over 10 percent; here it is 23 percent for October.

As I've said in the debates with Shirky on Terra Nova, I go by my own number of customers, and I and others with businesses have seen them grow dramatically, double, tripling in the last 3-4 months with real people who really log in every week.

The figures of sign-ups are always going to be misleading because for some kinds of people, SL is a place they go to on vacation twice a year or drop in only to hear a certain famous speaker -- it's like the number of times they might have used TurboTax or RealAudio in the last few months.

For others, they use it as much as Word or Outlook Express or Excel.

Shirky wouldn't deny the makers of TurboTax the right to say they had sold XXX million copies even if their customers don't use it every day; yet he becomes this fastidious about SL. Why?

I think it's mainly about the problem of Web 1.0 yielding to Web 2.0 or even 3.D.

Prokofy Neva
Posted By Prokofy Neva, New York, New York : 8:45 AM  

A Prokofy Neva post with reason, and without venom! I am going to have to tell all my friends. :-)

1.5 Million plus people have taken the time to download the client and try it at least once! When did 1.5 M become insignificant?
Posted By Bob, Seattle, WA : 10:34 AM  


Thank you for doing this. You are, I think, the first person to get actual numbers from Linden.

You've done us all a favor by removing the earlier figures from the debate and replacing them with real numbers -- 2.3 million people and 77% growth in two months would have been 70M people by December of 2007. The new figures of 250 thousand and 23% suggest 3 million users by December of 2007, and one which, finally, gives us something to base our reasoning on.
Posted By Clay Shirky, Brooklyn, New York : 2:58 PM  

1,525,670 unique people have logged into SL at least once. This is considering distinct email/payment info as distinct people, rather than IP addresses. [He is checking the unique IP address numbers but suspects they will be comparable.]

I would be surprised if the number of unique IPs logging in is comparable to the total distinct email addresses, since the 1.5M includes all the accounts registered using fake email addresses as "unique people". I'm interested to learn what the unique IP number is when Philip Rosedale gets back to you.

I know people who have registered upwards of 10 accounts using different fake email addresses. Without even having to use a real email address to register a "resident", counting accounts by unique email address is next to useless.

I'd be interested to find out how many accounts have logged in over the past 30 days *that were created more than 30 days ago*. Any chance of getting that info in a followup from Rosedale?
Posted By Thomas, New York, NY : 3:39 PM  

Hi David,

It seems silly to congratulate someone for a job well done when all they did was pick up the phone. Nonetheless, congrats! You did more than most who are in a similar position to do the same.

Best regards,
Posted By IAmEric, Los Angeles, CA : 9:27 PM  

Prokofy, I'm sure Clay and everybody else, wouldn't have a problem with Linden reporting its user figures as those who have paid it money, just as nobody is going to arguing with a claim from TurboTax that they have sold XXX million copies.

The analogy would be TurboTax offering a free demo and counting everyone who has started the download (whether completed and installed or not) as a customer. I'm quite sure they'd be called on that.
Posted By Matt, Mill Valley, CA : 4:26 PM  

So... when is Linden Lab going to stop calling it 'residents', displaying it in three different places, embedding it in the 'What Is'-paragraph, etc.

It is quoted in the context of being "number of what a normal person would consider a current player of the game" every day, every single day numerous times, and I don't think Linden is full of dummies...

It's a deliberate deceit.
Posted By Brenda, Austin TX : 11:05 AM  

To send a letter to the editor about The Browser, click hereTop of page

Got a news tip? Send it to The Browser