The Mysterious 40,000 People Plateau of Second Life User Concurrency

Second Life concurrency data courtesy since 2009 by Tyche Shepherd's Grid Survey

Here's the latest Second Life concurrency stats direct from Tyche, the woman who knows more about them than anyone outside Linden Lab (and possibly even more than anyone still at the actual company):

"The trend in maximum daily concurrency is pretty flat for the past 3 months or so [see chart here], the same with Median and Mean Daily. It is down on from what we were seeing in the Spring but still above pre-Covid levels."

The COVID "spike", however, was just a few thousand more active users, and most of them were existing SL accounts coming back during the quarantine.

Which leads to the mysterious plateau of Second Life user concurrency. My takeaway from what Tyche tells me next:

"The average of the mean daily concurrency trend in the past 3 months was [roughly] 39,819.

"The value for 2020 and 2019 for the same 3 months was about 39,865 and about 36,850 respectively -- so no real drop since last year and still being higher than the same period in 2019. This is based on the underlying trend after removing the weekly seasonality effect (which is pretty constant over the period in all 3 years and accounts for about +/- 1100 per day of week)." In other words, the median concurrency varies by about 1,100 on average, depending on the day of the week, with weekends being typically highest in CCU.

So for the last 3 years, Second Life's concurrency has been in the ballpark of around 40,000, give or take a few thousand here or there... and that's roughly been the range since at least 2009, when Tyche started tracking it! (Embiggenate her chart above, to see.)

This despite --

• Linden Lab claims 300,000+ new accounts are created every month.
• Most (all?) virtual worlds and MMOs gradually lose active user numbers over time, after peaking.
• Virtual worlds with similar functionality, such as ROBLOX, IMVU and Avakin Life, have far outpaced Second Life since then, especially in terms of monthly active users.

It's seriously the strangest stat I've ever seen in tech, and I don't have a satisfying answer for why it exists. But I'm working on a project that hopefully uncovers that mystery (he said mysteriously).

Comments

[Sally Crofton]

Just wondering out loud... Do the "online" figures take account of scripted agents/bots?

There still seem to be lots of them doing group mailings for rent or personal bots like ours & there's still a few of the arabian sims that have bot platforms - this would mean there will always be a "floor" through which figures will never fall (though I seriously doubt that would be more than 2-3k?).

It would be interesting to see regional access figures e.g. logged in from which region/country and what time? It would help folks running ventures that can't get these figures accurately (i.e. we don't use a vendor system or MP to track sales) to better rationalise their visitor levels.

It is interesting though that in spite of the "new signups" figure bandied about that concurrency bounces along around 39k - surely must be cause for concern at the Lindens about new user retention? It would mean the new sign in screen, new snapshot features, actively making venues pursue "new member friendly" policies and so on are not having the impact they hoped or that the issues aren't what the Lindens presume them to be?

[Sara Nobody AGAIN]

Has LL hired a new CEO yet? and as for the numbers one has to ponder how many bots do LL operate?

[Adeon Writer]

> Just wondering out loud... Do the "online" figures take account of scripted agents/bots?

If they have opted to flag them as scripted agents, no, they are not reflected it the online count, and is discounted from parcel traffic as well. According to TOS, you must do this if you run a bot.

But if you don't? You will be counted in traffic and total online counts like this. And... nothing will happen to you, really. I don't think LL has any way of checking who is a real account and who is automation.

What percentage of people run bots without properly flagging them? Who knows. I'm sure there is a decent chunk. People really aren't up on these kinds of things.

[Pulsar]

"for the last 3 years, Second Life's concurrency has been in the ballpark of around 40,000, give or take a few thousand here or there... and that's roughly been the range since at least 2009".
I'm not sure if I'm seeing the same thing. In 2010 the mean daily user concurrency hit ~60,000+. The average of the mean was maybe 57,000. It looks like so from the "Raw Data" graph in the posted image. This one may be clearer (although it is from 2019):
https://i2.wp.com/www.gridsurvey.com/charts/historicalconcurrency.png

This would be a ~50% decline from the peak.

It looks like it declined less from about 2016 onward, though.
It could be that

• Hard-core users tend to stay despite everything.


• When you invested so much time (and spent money) or you have a virtual business, lots of friends... all that keeps you in SL further and it won't be so easy to abandon everything.


• Other virtual worlds have indeed outpaced Second Life, but SL specific niche has specific interests or would miss some feature or aspects that in SL are openly accepted (*); therefore, so far, there isn't a full replacement.


• You can still use both, SL and the new ones.

___

(*) = I'm sure someone is thinking about sexuality. While sexuality happens elsewhere too, sexuality and erotic stuff is indeed openly accepted in Second Life (in the appropriate places) even dark fantasies. But sexuality aside, you can also edit everything inworld (as long as you have mod permissions) and tinker with minute aspects of your avatar and accessories. Also most of the grid is a single giant world: you can travel for hours, it feels like a world. And so on.

Vice-versa, some of those aspects that appeal the SL niche (and you have that niche for a reason) are weak points to other people.

Second Life is banned from Twitch and I think it is not very well suited for minors. Roblox users (or they parents) may be concerned, but also not every woman is happy to be approached by horny guys so often. And several are into BDSM groups. You may not notice this if you just stay with your friends and you don't move much around; but as you explore a lot and you met many people, you can see that every day, sometimes even in "PG" rated regions. Skimpy clothes, horny guys, naked noobs with a visible erection, up to more explicit stuff.

The "complex" user interface lets you do so many things but it also turns off several newcomers, who are used to change their game characters or avatars easily, with previews and a couple of clicks (I mean, your inventory in SL is like a file manager, awesome for "power users", less for other people, not counting alpha slices and alpha layers).

And so on.

So the use cases don't fully overlap.

[Pulsar]

I wrote "SL and the new ones" while I had Avakin Life in mind, but of course Roblox and IMVU aren't new. Anyway, you can use/play with these other popular virtual worlds (or also with VRChat), and still enter in Second Life other times.

[Wagner Au]

"The average of the mean was maybe 57,000."

Good point, thanks! That's still not much of a drop from a mean of 40,000 when you consider that there's supposed to be 350,000 new sign-ups every month. Even if only 1% of those new users (3500) become regular users over the last 120 months, the mean concurrency should be well into the six figures by now!

[lkosov]

Are we sure that 300-350k/month number is still correct? It's going on four years old now.

The incredible consistency in the concurrency count almost makes me wonder if those numbers are incomplete - not showing people on private sims maybe, or adult-rated sims? Just sort of spitballing.

[Lewis Luminos]

"Linden Lab claims 300,000+ new accounts are created every month."

I suspect that the majority of those are people creating bots to spam the forums (who don't even know that the grid exists), and existing users creating alts, which generally aren't going to be online at the same time as each other. Looking at the new user islands on the map, I rarely see more than three or four green dots there at any one time, so the actual number of real new users, who are genuinely logging a new account inworld for the first time and going through the tutorial, is very small.

[Pulsar]

You are welcome! Glad to contribute. And indeed user retention isn't great. In 2007, according to Rosedale, about 10% of the newly created avatars were still logging weekly after 3 months. I don't know how much that improved in these years. Meanwhile there is always who quits, for a reason or another.

Lewis Luminos may be right. Also someone who isn't a bot still gives up early (changing their mind, turned off by something, etc). Indeed not every signup translates into a new avatar rezzing inworld.

300k+ per month would be about 10k+ per day or more than 400 new avatars every hour on average.
Even if someone rezzes and poofs after few minutes, giving up, you would still have a new avatar every 8-9 seconds on average; moreover the orientation path takes a while to complete. Thus the the welcome islands should look populated every time, but in fact they aren't.

Of course there is also more than one welcome island (the first orientation regions, not the Social Islands), but not so many to spread the new avatars apart so much.
Learning Island has a couple of prototyping/testing clones (Learning Island Build and Learning Island QA), but I think the only one used to receive new avatars is Learning Island 1.
Firestorm Orientation region intercepts some new signup, but I guess most of the actual newcomers would start from secondlife.com and follow that route.
Even if they were distributed evenly among, let's say, 4 regions, you would still see a new avatar every half minute.

Then, after the orientation path, you are sent to the Social Islands (or the equivalent for Firestorm). Part of the genuine newcomers may give up before that and several Social Islands do exists, but indeed you rarely see more that few green dots among the whole of them.
The only populated one is Social Island 10, because there are oldbies gathering there, like in the old infohubs. Only few avatars there have less than one month and rarely are rezzed today.

[lkosov]

Pulsar's comments got me curious, so I sat there with the world map open and watched the green dots on Learning Island 1 for an hour yesterday, from 1235-1335 SL time. From what I can tell the player dots update once a minute, so it's an imperfect way of monitoring things, but was easier than creating a new account and camping on the island.

Anyway, best as I can tell, twelve accounts joined in that one hour. Saturday afternoon in the Americas, Saturday evening in Europe. If that's normal (and that's a big assumption to make from one observation of one hour's possibly flawed data) that's something like 9,000 new users in a month. Even if I'm somehow an order of magnitude off, that's still a *far* cry from "300,000+".