Second Life Concurrency (Now Below 70K) is Not the Best Indicator of SL User Activity, Here's Why

Daniel Voyager has a good summary of Second Life concurrency rates in 2011, citing data from Tyche Shepherd's invaluable Grid Survey site. As Mr. Voyager notes, SL's concurrency remained below 70,000. That's an interesting data point, but here's an important qualifier that some often miss: Concurrency is not the best indicator of SL activity. Here's why:

• As Linden Lab CEO Rod Humble announced during his SLCC keynote, SL has seen growth in new returning users in recent quarters (not much, but perhaps about 10-25 thousand.) However, as he pointed out, this won't necessarily be seen in the concurrency rates: Most of these users tend to be from a younger 20-something demographic (historically, active SL users have skewed older, from the 30s to 50s), and these new, younger users have lighter in-world activity patterns.
• Because SL has such a very international userbase (less than 30% of the total are in the US), concurrency will naturally rise and fall as one time zone's users log in, while another time zone goes to bed.

Because of these two factors, it's very unlikely that we'll see SL concurrency rates notably rising until there's a very significant jump in total monthly active users. (I'd say approaching 100,000 or more.) And even if SL were to gain this amount or beyond, the concurrency rates may still not grow much, as these new users will probably log in for shorter time periods, from all parts of the globe. So if you want a truer survey of SL, look to its monthly active uniques.

Comments

[Hitomi Tiponi]

I'd really like to see the figures for monthly active uniques from avs more than a month old. I think that would give us a good indication of any upturn in user retention. Currently the figures are easily affected by the noise of those who log in just one, two or three times then leave.

[shockwave yareach]

Sorry, but that does not compute. If concurrency ebbs and flows with the different time zones as you say it does, then it did the same from 2003 - 2011. Likewise claiming that young people today don't spend as much time on games or the net is easily disproven simply by looking at the age distribution of WoW players (late teens early 20s for males, 30 and 40 years old average for females.)

The virtual world no longer attracts the big money and the interest like it used to do. Those willing to pay top dollar have come, been treated horribly by LL, and gone. Even those of us stubbornly remaining aren't on as much as we used to be. The reasons and the solutions have been spoken about so much that we have blisters on our vocal cords. But the bottom line to LL's bottom line is simply that the noobs aren't going to plop down 300$ a month for virtual land.

And no amount of smokescreen or trying to make us look at the wookie is going to change the fact that people are leaving and taking their money with them. If Concurrency was a stellar benchmark 3 years ago, it still is today. And with so much data being hidden from us by the lab, it's one of the few reliable datasets we have left to work with.

[Hamlet Au]

"If concurrency ebbs and flows with the different time zones as you say it does, then it did the same from 2003 - 2011. Likewise claiming that young people today don't spend as much time on games or the net is easily disproven simply by looking at the age distribution of WoW players"

Actually, SL only started gaining its highly international distribution around 2006-2007 during its media hype period (which also artificially boosted the concurrency.) And MMOs like WoW generally have very different usage patterns than light social worlds like IMVU.

[Tateru Nino]

If you're looking for a reliable indicator of overall Second Life usage, pretty much the best number available is median concurrency.

Peak concurrency, while interesting from a load-management perspective, isn't really particularly indicative.

[Eddi Haskell]

I can't agree Hamlet. Younger users do not spend less time online than older users -- if anything, they spend more time online. And your international argument does not computer either, since Second Life usership has always had a strong international flavor --- Europe and Japan have always been well represented. It is very easy to inflate the number of sign-ups who then drop off quickly. It is very hard to inflate the number of real Second Life users online at once.

Median concurrency has, and will be continue to be, the best way to look at success in Second Life. No ifs, ands, or buts.

[Henri Beauchamp]

It's funny how LL *always* attempts to bias the numbers, of just hides them away (like the excellent and reliable "Connected last 60 days" figure that was removed from the viewer login screen now over one year ago, when the figure clearly showed already the fact that SL was on a slow decreasing curve after a short plateau), so that they can still brag and pretend everything is wonderful in the best virtual world on Earth...

Let's face it: the exponentially growing curve that we witnessed back in 2006-2007 will NOT return, *unless* LL rolls back all the mistakes they have done over years:

- Return to a SANE viewer UI (no, v1 was not perfect, but it is still today WAAAAAY better than v2/3 !), and a renderer that doesn't require a super-computer to run smoothly.

- Lower the land fees and stop being GREEDY: better having more and more users paying less than less and less users paying more !

- Stop segregating Adult contents: SL *is* an ADULT game. the whole main land shall be ADULT but for around 30% of PG areas that would be reserved for teens and marginal "work-safe" stuff.

- Reinstate the original motto that has been long forgotten by LL: "You World, Your Imagination". LL should not be the one to decide what the contents of SL should be: it's to the resident to make SL. LL must "simply" provide the infrastructures and tools.

[Alex]

I'm not sure if this is just horribly misguided by the author's personal expectations/hopes/opinion or if this is intentional nerd-baiting.

Concurrency is (together with L$ spent) pretty much the only useful statistic that's still being released. A decrease in concurrency can be directly translated to "Second Life is now emptier then it used to be"

[Hamlet Au]

L$ spent is definitely an important metric, but concurrency, not so much for the reasons given above. In the free-to-play space, monthly uniques are the most important industry standard metric.

Concurrency is less meaningful, as is median concurrency. Talk with Linden data folks, and they emphasize that point. In fact, I get the impression the Lindens roll their eyes when SLers obsess over concurrency stats as if that's the most important indicator.

[Aliasi Stonebender]

Hamlet does have a good point, and keep in mind 'being online all the time' doesn't mean 'being online TO SL all the time'; I've noticed younger sorts tend to not spend hours on hours in SL like I used to. (Okay, I still do spend hours on hours... sometimes. :) )

[Tateru Nino]

We'll happily use other metrics if the Lab will provide them. Until then, we'll make the best of what we've got.

Watching peak concurrency is rubbish, though. For obvious reasons that is going to fluctuate independent of actual usage.

As for median concurrency, it's actually very good as an indicator - unless you've got access to *better* data... which we do not.

If I were the Lab, though, I wouldn't be rolling my eyes when even the statistics that *are* published are erratic and published unreliably. Maybe the Lab should obsess over a little reliability in that area. :)

[foneco zuzu]

Henry, not only is the one that makes most of users still be online (Mesh on v1;) but raized fair and directly all that the Lab needs to do.

[foneco zuzu]

But i believe it has already starting to improve, Lea sim are the best example of what can be done if the Lab just creates and improves the tools and let the users do the rest!

[Hiro Pendragon]

Given the two reasons stated:

"Most of these users tend to be from a younger 20-something demographic ... younger users have lighter in-world activity patterns."

But the name of this article is "Second Life Concurrency ... is Not the Best Indicator of SL User Activity ..." so having lighter in-world activity patterns ... um ... is user activity.

"concurrency will naturally rise and fall as one time zone's users log in, while another time zone goes to bed."

Well that's easy to measure - if this is true, we should expect that the concurrency troughs will be less deep.

The bottom line (pun intended slightly) is that we should be able to take the area underneath the graph as, essentially, total user-hours. Is user-hours not "activity"? I've always agreed that we need to look at monetary factors of how many L$ are sold for US$ and actual land owned, but how long people are actually logged in not only seems standard to me, but Neilson ratings.

[Hamlet Au]

"We'll happily use other metrics if the Lab will provide them. Until then, we'll make the best of what we've got."

Linden Lab does provide other metrics: Monthly uniques and monthly uniques spending L%, among others. Have you asked them if median concurrency is a more meaningful metric for user growth and activity?

[Pussycat Catnap]

"Likewise claiming that young people today don't spend as much time on games or the net is easily disproven simply by looking at the age distribution of WoW players (late teens early 20s for males, 30 and 40 years old average for females.)"

Teh interwebz iz fulls of hot young studz seek'n cougars?

Oh my!

(Doesn't SL have nearly the same kind of age distribution? Or is it just my impression of male AVs that they're a bit 'younger'?)

[Pussycat Catnap]

"- Return to a SANE viewer UI (no, v1 was not perfect, but it is still today WAAAAAY better than v2/3 !), and a renderer that doesn't require a super-computer to run smoothly."

The world done passed you by...

The Amish are ---------That-Way----------->

Rest of us, even the old Emerald Crowd, now on Firestorm, have moved on.

[Pussycat Catnap]

"Because SL has such a very international userbase (less than 30% of the total are in the US), concurrency will naturally rise and fall as one time zone's users log in, while another time zone goes to bed."

Pay attention not to Concurrency peaks and valleys - but the mean for a day. The mean has been fairly static for years - ranging from the low 40k's to high 50k's day by day. There were 50k's a little more often a few years back, but even then there were a good number of 40k's, and even now 50k's pop in.

- This may be a reflection of shorter logins, meaning less data to 'uplift' the low points int he day... rather than actually less users. But not sure on that. Nonetheless, the shift down has been very gradual and rather minor over the last 3 years. When compared to the peak days of 2006/2007 it might seem like a disaster

- but, during the entire "the lindens are out to kill SL" era of M. Linden, we only shed maybe 10k people - if you believe this 'light usage theory' is bunk. If you subscribe to the theory, we probably gained people...

As for seeing so many popular sims close: that's just Disco.
- As in, fads and what is popular changes. For ever place closing, another opens. Its just that the old-hats aren't in the 'in crowd' anymore for the new spots... we're all still dancing disco, and the kids are doing break-dancing & glam-rock... but hip-hop's coming up right behind them too...
- times change, even in SL.

Now where mah mood ring and pet rock @?

[Tateru Nino]

I hate to be the one to point this out, but "Average Monthly Logged-In Users" isn't the same as "Monthly Uniques" - in fact, I wouldn't consider them to be comparable figures (even if there were a monthly-uniques figure to compare it to).

Also, that says nothing about relative activity levels, which is the key information we get from monitoring concurrency (the other, of course, is being able to measure actual service stability).

[Tateru Nino]

In fact, now that I think about it, Linden Realms skews the average monthly economic participants data as well, essentially concealing any patterns in overall economic activity within SL (to the point that we got much of any from that figure, which, alas, we do not).