Second Life's Real Growth and Pretend Growth

This is like deja vu, for me, since I seem to talk about it at least a few times a year to people.

Linden Lab released its latest numbers for the Second Life economy. On the surface, the headline is "ZOMG SL ECONOMY GROWS 65%!". In fact, the blogosphere has already picked up this headline, in the usual "Gee, I don't bother to understand the platform, technology, or relevance." that I so recently griped about.

This is Linden Dollar (L$) transactions between users. Sure, it's the virtual economy, but this is not a meaningful number. Allow me to:
- Explain why this is not a meaningful number
- Explain the various ways this number can be easily manipulated
- Explain what metrics are meaningful

L$ transactions is not a meaningful number

If I'm in Second Life, and I have L$ play money, and I spend it, I'm spending play-money. It has the *potential* to become real money, but only when cashed out, such as with Second Life's Lindex system. The examples of real items that can be purchased are extremely rare; since the power of a currency is dependent on the buying power, then we can't consider this a real currency in-and-of-itself. Naturally, Linden Lab would be stupid to have it any other way - if it were considered a real currency, they'd have a dozen government agencies of each of a few dozens countries banging on their office doors demanding regulation as a bank, or perhaps its own country. So, as a side-note, it's ironic that Linden Lab touts the SL L$ economy, but absolutely needs to ride the other side of the legal line in terms of validity of its currency.

Now, even not as play-money, we have the issue that we're talking microtransactions. The average transaction size is tiny - as most things in Second Life cost the equivalent of pennies. People are much more likely to spend their money on super-cheap items on a whim.

Gambling A Problem

A few years ago, Linden Lab used to post, on the Secondlife.com website, the last 10 or so transactions done in world, with what what purchased. Back then, Linden Lab was doing the same thing - publishing the L$ transactions like it's the greatest statistic in the universe. I took a sampling of the transactions, and found that 7 - 9 of each 10 were gambling related. For example, the transaction might say "50L$ - poker table" or "10L$ - slot machine". If 70 - 90% of transactions are gambling related, then the number is hugely inflated.

If I were to sit down at a SL slot machine - well, one that isn't horribly rigged to the benefit of the house - and, in my experience, since gambling in virtual worlds is completely unregulated, they all are unfair - I might spend an hour at a machine. Let's say I break even. I would leave the table with no more or less *net* money. However, I could have racked up hundreds, thousands, or even tens of thousands of L$ in transactions in that hour. In a real economy, we don't count each transaction at a casino for the same reason. People report their net gain / loss, unless you hit something like $10,000 or more, in which case tax withholding is taken immediately.

But gambling is banned in Second Life, so you say!

No, not really. As the Alphaville Herald (formerly Second Life Herald) has reported, gambling in Second Life is still alive and well under the guise of "games of skill". While some of it has gone, I did my own searching, and it's pretty easy to find a variety of places to go. Also, some of them use a third, imaginary currency to do transactions, however, this is a minority case.

The Number Can Be Easily Manipulated

Let's write a hypothetical script. I'm not the first to suggest this is possible, and, *any* scripter in Second Life will tell you how easy this is - I'm not revealing any sort of secret. It probably could be done in 10 lines of code or less. EDIT: And as I suspected, it had been done already, as covered by the Herald. (Thanks Pixeleen of the Alphaville Herald for the link.)

Anyway, this script would pass a small amount of L$ - let's say 10,000L$ - roughly $40US. This script would be run on 3 or 4 avatars, and constantly be passing money back and forth. Let's look at the numbers one might generate using a sluggish estimate of 1 transactions every 10 seconds.

10,000L$ x 6 per minute = 60,000 / minute
60,000 x 60 per hour = 3.6 million / hour
3.6mil x 24 hours per day = 86.4 mil / day
86.4mil x 365 days per year = 31.536 billion L$ / year.

According to the Linden Lab statistics:
"User-to-User transactions in 2009 totaled US$567 million in 2009, growth of 65% over 2008."
Assuming $4/1000L$ as a decent approximation, that's about 141.75 billion L$ transactions. Note that Linden Lab reports only 500 million L$ in XStreet sales, less than 1% of total transactions.

So from this simple script, with a modest 1 transaction per 10 seconds between all the avatars, one might generate 22% of the entire L$ transactions in all of Second Life's millions of users. If the transactions were faster, or more L$ used, a few avatars could easily shatter the numbers.

That's how unreliable these numbers are.

I'm not saying Linden Lab manipulates it - let me be clear. I just want to point out that this number is essentially unreliable.

---
EDIT: Massively blogger Tateru Nino published her own take on Linden Lab's 65% figures just now. She's a bit more critical than I was, but pointed out at least one major way that L$ transactions are inflated - management of accounts. Shop owners, land barons, and a variety of groups in Second Life often times have money shared in trust. This money may be passed around fairly commonly. Some specific examples might be:
- A shop sells popular item X made by multiple creators, who each need a cut of sales. Buyer A pays shop owner B. Shop owner B pays multiple avatars their cut of sales. Suddenly that 100L$ sale became a total of 200L$ in transactions.
- A land baron has 10 sims that they rent. They use one avatar to collect rent. They have staff to pay, as well. Similar to the store, the actual rent cost is, at minimum, doubled, perhaps more if there are any intermediaries.
- A charity group has a trust account for L$. Donations to the group pass through multiple avatars before reaching the US$ cash-out. Since Relay for Life this year earned over $250,000US this way via Second Life L$ donations, I'd say this is a real issue as well.

----

Let's Talk About the Reliable Numbers

There are 3 reliable numbers that I'd like to point to:

1. L$ sold for real currency on the Lindex (and 3rd party L$ transactions sites)
2. square meters of land owned in SL
3. number of active users (non-bot)

1. L$ sold for real currency

This is a highly interesting statistic because it represents actual commerce. This is real money being exchanged. According to Linden Lab's statistics:
"The total US dollar value of all Linden dollars traded on the LindeX™ currency exchange in 2009 reached US$115 million in value, 7% growth over 2008"

That is absolutely not a trivial number. You'll notice a few things:
- The number is approximately 1/5th of the transactions number.
- This number is still on the rise. A 7% growth in a global recession is pretty dang good.
- The growth percentage is much lower than the growth of raw transactions.

2. Square meters of land owned in Second Life

This is extremely important as it represents actual monetary investment into Second Life, in purchasing real servers, real bandwidth, and real support time from Linden Lab. According to Linden Lab:
"Resident-owned regions reached 23,900 in December 2009, up 6% over December 2008"

To give you an idea of the scope, 23,000 sims equals approximately $6.9 million in raw monthly sim revenue. It's probably lower than this since there are low-density sims which cost less than regular sims.

This number has grown 6%, which seems on line with the Lindex L$ to real currency transaction growth of 7%. Again, in a global recession, nice work to Linden Lab.

3. Active Users (Who are not bots)

There are no "users who left Second Life", since there is no account termination unless User accounts are purged specifically requested by the user, or by administrative punishment for wrongful use of an account. Additionally, Linden Lab has purged old, inactive accounts at least once. I can see the reasoning to remove unused alts, bots, and failed account creations, however don't see the reason to remove accounts that spent any significant amount of time in Second Life. But I digress. Raw users is not a good metric of populace and usage of Second Life, and thankfully it it one that Linden Lab has long abandoned as touting.

Active users is a much better metric. Active users can be defined in many ways, but ultimately it involves a certain number of visits and a certain numbers of minutes use over a certain amount of time.

Linden Lab's latest report doesn't have this data. Instead, it reports usage data, some of it broken down into bands of use. It's a huge loss not to have more data on these users. When people sign up for accounts, it asks for some info, but no where does it ever ask, "Why are you trying Second Life?"

I'd love to see stats on:
- Number of people who log onto Second Life once for a specific event and/or location advertised outside of Second Life.
- A totally revamped Registration API that solution providers can use to better sign up users and provide better statistics. Universities could then say things like, "We signed up 100 accounts, they were students who were assigned to do a class" and then cross-reference this with their usage. A marketing campaign by a Fortune 500 company could say, "We signed up 5000 users from our website promoting our Second Life attraction", and the same cross-referencing done with usage.

My current company provides metrics solutions - as do a few competitors - and these help illuminate usage, but only for the company or organization who sees the statistics. It would also be a great help for Linden Lab to provide more hooks in their code to grab more data. You know, like Apache does with a standard web-server. Grab the data, let us get into the data of a sim using a standard web / command line interface / database.

Conclusions:

Second Life grew at least 7% last year. Reports of it growing 65% are highly exaggerated. The truth, if I had to guess, is more like 10 - 15%. At the same time, I wholeheartedly applaud Second Life for growing the amount it did. Rumors of Second Life's death for two years now are also highly exaggerated.

Idiot bloggers who republish the 65% statistic without taking a minute of time to analyze and question the data deserve to have their blogging license revoked. Likewise, the same is deserved by writers who proclaim Second Life dead. ^^