Second Life private simulator data, December 2006 to May 2009
by Tateru Nino 
Jul 4th 2009 at 6:00PM
Filed under: Economy, Opinion, Second Life, Virtual worlds
It's been widely grumbled about that Linden Lab has stopped publishing some key Second Life economic and service statistics, such as the number of non-mainland simulators (that is, those that are privately owned). Well actually,as it happens, that isn't entirely true, thanks to the wonders of automation.
The data was culled from the Second Life Web-site, where these things go, quite some months ago but automation is a funny thing. Linden Lab's systems kept publishing the data, and my own automated systems kept collecting it. So, despite a months-long blackout of the data it all continues to pile up in my databases, with the reliability and patience that only machines possess.
Here's a larger version of the thumbnailed image. You can click on it for the largest size.
Here are our key labels along the timeline:
A – this is where Openspaces simulators were announced as a generally available product. Simulator purchases immediately begin to rise, and the following month, they take off. There's barely any slackening before
B – This is where Linden Lab announces the change to products and pricing in October 2008. From a peak of 26,539, the first ever monthly drop is recorded, with the fall finally bottoming out at
C – The decline stops here with 21,246 in March 2009, followed by an immediate jump of a full 4,281 simulators for April, and a further 955 for May, the last point on our chart.
Clearly the 2008 Openspaces product was popular. The curve looks like it would have easily topped 30,000 simulators by the end of 2008. The slight slackening in the rate was due to the announcement of specification and pricing changes partway through 2008, followed by the sharp decline in total simulator numbers, as simulators were abandoned or consolidated. Because of the announcement, October 2008 didn't have as much sell-through of simulators.
April 2009 is something of a mystery. With a net loss of 304 simulators in March 2009, suddenly there's a jump to a net gain of 4,281 for April 2009, before returning to a pre-Openspaces growth rate for May 2009.
That large jump for April is certainly intriguing. It isn't at all clear what it indicates. We're pretty sure that it isn't a data error, but we're really not sure what it actually represents. Looking at the raw data, 3,608 new non-mainland simulators suddenly appear on or about 17 April, following a lackluster fortnight. (June will end almost 10,000 simulators up from the end of March)
It's possible that some of those sims belong to Zindra, the new adult continent, and aren't being counted correctly as mainland sims, but that can't be the whole story. Zindra – we believe – just isn't that big.
Linden Lab could be mirror-testing simulators, running a second copy of the simulator on different software/hardware, but the same inputs and validating new systems that way, but again, the number seems far too large for that.
Our best guesses at the moment are that there is either a significant change to the way simulators are being categorized, and some or all of the simulators formerly counted as the mainland estate are no longer being counted as mainland sims, or that there is some very large deployment of simulators going on that we don't know about, and perhaps cannot yet see.
What cannot be reliably inferred is that the data represents a recovery ... at least not yet. The anomalous surge in mid-April pretty much squashes positive interpretations of the data, until more information comes to light. The data could represent a real recovery, or it could just be QA or Systems Engineers mucking about, or some secret project that the Lab has going.
Got your own theory? We'd love to hear it!
 |
Are you a part of the most widely-known collaborative virtual environment or keeping a close eye on it? Massively's Second Life coverage keeps you in the loop. |
Reader Comments (Page 1 of 1)
Shockwave Plasma said on 6:21PM 7-04-2009
Possibly setting up a new datacentre? They might be doing some load testing.
Reply
Eris said on 6:42PM 7-04-2009
Could this explain (some of) it? If you look on the world map there's an array of empty 'Landstore' sim's sitting waiting to be claimed and placed. I don't remember ever seeing these before, but maybe they were always there? Presumably having them already on the grid speeds up the allocation process when they're actually bought?
http://slurl.com/secondlife/Landstore%20Pool%20250/128/128/20/
Reply
Tateru Nino said on 7:17PM 7-04-2009
Some of it, but from April through June, we're talking roughly 3600-9800 or so sims. I doubt the pool accounts for more than a thousand.
Land Barron said on 7:26PM 7-04-2009
According to Tyche Shepherd the region count is around 28, 322. Not sure how official the count is but she's consistent on a weekly basis. http://www.sluniverse.com/php/vb/business-land-economy/8523-new-second-life-sims-past-16.html
Reply
Tateru Nino said on 8:14PM 7-04-2009
The Lab's numbers list just over 31K up to Friday. It's entirely possible we can't see the extras. Or maybe their data is just plain wrong. The sudden jumps in April (and later in June) though seem to suggest that something's going on.
Nightbird Glineux said on 10:28PM 7-04-2009
Could LL's announcement of grandfathering for Openspaces sims have anything to do with the acceleration after point C?
Reply
Tateru Nino said on 10:36PM 7-04-2009
I'm pretty sure it didn't. The announcement took place in the last week of May, and from looking at the raw data, it caused a bump, but too small a bump to show on this chart.
invincible.metaphor said on 6:28AM 7-05-2009
1. Land Store is exactly 100x5 sims and has always been since beginning of my observation, i.e. 24th of March. It's size and position never changes.
2. According to my map, most of the growth during the period of observation is accounted for by large checkerboard-patterned clusters of sims, most of them clearly residential - i.e. relatively few landowners bought relatively large numbers of sims each.
3. By my count, since 24th of March, 2391 islands sank and only 4194 were created. That includes new Linden mainland.
Data is collected by grabbing all the sim coordinates for which map tiles exist and then querying Map API to see if they're actually still there, it is limited to only include sims actually known to be accessible - i.e. things too far off the map are not included.
My guess is, the raw data disclosed by LL is somehow bogus. :)
Reply
Tateru Nino said on 7:35AM 7-05-2009
There's either a flaw in the data, or there's a rollout of simulators that we can't see. They can't be Nebraska sims, as those wouldn't be counted (no connection to the grid), but it could be some sort of sub-estate. Quite a sizeable one.
aliasi.stonebender said on 8:08AM 7-05-2009
Hm. I dunno, looking at that graph, the 'recovery point' seems to be right about where it would have been without the artificial "Openspace" spike. A little lower, perhaps, but not too far off.
So my out-of-the-rear guess is just the artificial growth - by every avatar with a bit of cash going "OH WHEE I CAN GET MY OWN SIM FOR LESS" and proceeding to load every fancy script they could onto a void sim meant for scenery - has been factored out and now we're about back to where it 'should' be.
Reply
invincible.metaphor said on 3:10AM 7-06-2009
Actually, map tiles on slurl.com include all sims on the Agni grid - even the teen grid. If it's a sub-estate, those invisible sims are well hidden indeed.
Reply
Louis Platini said on 11:43AM 7-05-2009
More detailed information about gain and loss of regions is available at http://www.metaverse-business.com/regionweek.php
The most recent regions (names) are available at http://www.metaverse-business.com/regionlst.php
There is indeed an increase in new sims the last weeks and less regions disappear.
Reply
Jay said on 5:42AM 7-06-2009
Have you considered that the Lab used to have all these servers running openspace regions... but then Jack Linden at the behest of Management decided to piss everyone off.
With all those empty computers now running nothing... and with the need to be able to say in the June financial reports about to hit us... "there are x,000 new simulators this quarter easily making up for the few people who left, SL is strong and forging ahead" when in fact the opposite is true.
A drug dealer never tells those who buy his crack that things are bad.
Reply
JeanRicard Broek said on 2:18PM 7-07-2009
LL stats are always difficult to understand. I am sure internally they have better metrics.
Let me first say, there are two reasons to know the size of the SL grid.
1. To understand the size of the world as most users know it for personal or business reasons.
2. To get some idea of how big the Lab's business & system is.
These are two different things and the metrics/counts would be different, considering hidden private/corp/gov regions, maybe even whole private grids, types of regions; full, openspace, homestead, dev, test or prod servers. even the duplication of regions across datacenters.
I am most interested in the "public grid", The one the average user/resident can access and find on the map and TP to.
I did find this from Tyche Shepherd as of 7/6 very informative
"Total number of Main Grid regions is now 28523 ( 23028 private estates & 5495 Linden owned)" http://www.sluniverse.com/php/vb/business-land-economy/8523-new-second-life-sims-past-17.html
* Estate - Adult: 218
* Estate - Mature: 19474
* Estate - Offline: 2
* Estate - PG: 3334
* Linden - Adult: 190
* Linden - Mature: 3875
* Linden - PG: 1430
My question is can Signpost or someone else that has knowledge of the map API, count the regions on the map and compare the count with other methods. I tend to be a visual person ~L~
Reply
Tateru Nino said on 11:55AM 7-07-2009
I am inclined to agree. 23028 sounds like the right number. What these figures seem to show are simulators that have been added in the last couple of months that are invisible to the map API (and to the rest of the main grid).
Linden Lab has more than one estate that is concealed from both. Even though *most* of the special estates (like the TG) are visible to the map API, some remain concealed.
The totals *could* be mainland sims being counted as non-mainland, but the pattern of jumps in the numbers don't seem to correlate to that in any sensible way.
Tyche Shepherd said on 6:37AM 7-13-2009
I have also seen the the ongoing publication island data by LL and did noticed that jump in April and May. The Jumps more or less were to levels which agreed with my own surveying of the size of the total main grid. i.e. Private Islands AND Linden Owned . I'm sure that this is what the newer figures represent and thus are not directly comparable with previous published figures as they only included private islands and Linden owned regions were excluded.
My own daily surveys agreed with the original publicly available figures - there was a slight discrepancy of +/- 10 regions which was probably explainable by my daily surveys being conducted around 8am SLT , whilst LL's figures were based on the situation at midnight SLT.
My figures are based on in-world surveying using a modified client. The Map API method will also detect c.220 teen grid regions and c.500 landstore regions which are not visible inworld on the Main grid.
Yesterday I reported 28725 regions of which 23181 are private estates (or private islands) and 5544 are Linden Owned (predominantly Mainland, with c100 for Help islands and special projects etc)
http://www.sluniverse.com/php/vb/business-land-economy/8523-new-second-life-sims-past-17.html#post742005
Reply
Tyche Shepherd said on 7:14AM 7-13-2009
I would add that as of 10th July the LL islands added stats are back in line with mine as are the latest results from 11th July
2009-07-11 islands added
July 2009
Islands Owned (end of Month) 23160
Islands Added During Month 270
June 2009 -
Islands Owned (end of Month) 22890
Islands Added During Month 1322
Whatever the blip was - the figures are back to reporting Main Grid publicly visible private islands.
Reply
Tateru Nino said on 8:49AM 7-13-2009
Well, Linden Lab staff are among my most avid readers going by the logs. Most of the company tunes in regularly. What we report hardly goes unnoticed, and if we report anomalies publicly, they get a good hard look by the Lab staffers in pretty short order.