Massively^™

A curious thing happens when you chart Second Life concurrency as a percentage of the total signups. Across the last two years, there's been an irregular, but steady decline, which is what you would expect. With retention of signups to 90 days running at about 10%, and concurrency being a function of both active users and session lengths (among other things), a continuing decline in the curve is the most obvious result.

About six months ago, however, that suddenly changed, and something very interesting has been taking place since.

Peak concurrency as a percentage of total signups suddenly flattened out in the vicinity of 0.48%. It's measurable, verifiable and consistent. Median concurrency follows the same essential curve.

With overall logins over a sliding 60 day window on the decline, daily signups slowly edging downwards, and 90 day retention stable, initially we were at something of a loss to explain the result. Bugs, policy changes, nothing much seems to have affected the visible proportion, though it takes temporary sharp dips where the grid stability was egregiously poor.

We talked over the results with an assortment of Second Life users and pundits for about a week and then spoke to Linden Lab's staff statistician, Meta Linden, about the results. Meta generally agreed with our collective hypothesis that the result indicated a stronger commitment among committed users.

"I'm seeing that those who become active are becoming slightly more active," she told us.

Meta Linden tracks user activity internally in a series of overall monthly activity groupings, discarding outlying data points that indicate short-lived 'mayfly' (or browser) accounts and always-on bots and automata.

"Those who are really active (in excess of 160 hours per month, for example) are actually on a steady growth curve," observed Meta, "but the amount of time spent by the over-one-hour-per-month and over five-hours-per-month groups are also on the rise."

Put simply, while the conversion rate remains relatively constant, overall user commitment among the conversions is increasing. Why that increase should track the overall signups so closely for six months is harder to work out.

The best answer to that seems to be network effects: every committed user of Second Life essentially adds some value to Linden Lab's virtual world as a whole, increasing engagement and interest for those who are inclined to find Second Life a good fit for them. This might be considered a payoff of the overall principles of user-generated content.

Demographically, Second Life went through some significant shifts in its first four years. From that point on, however, it seems that the world has finally stabilized. At present though there is very little actual correlation between the age of the account and the amount of time spent each month, there is a stark correlation between the age of the user and the time spent each month.

Older users spend more time than younger ones, with the under 25 age-group both the least well represented and representing the lowest average use per person.

Nevertheless, all of the active users are spending increasing amounts of time using Second Life, and the boosts to weekly median concurrencies and peak concurrencies have been consistent and stable for the last six months.

If the concurrency trend continues as it is, then it appears that Second Life will overtake Everquest's peak concurrency around Christmas-time this year (reports put that figure at just over 90,000), and start moving towards the concurrency ballpark [link may require Internet Explorer] occupied by World of Warcraft.

While Second Life might be out of its infancy, it's by no means mature. This generation of virtual worlds is only just getting going. In a sense, the party's just starting.