July 2007 Key Metrics Released

Hello everyone, I am finally publishing our Public Metrics in Excel and OpenDocument and Google Doc formats through July 2007. I apologize about the delays in getting these posted, but hopefully the addition of some new data should make this worthwhile.

The new data, as highlighted by Phillip during his keynote at SLCC last weekend, now includes client crash rates, and both server and client performance statistics in terms of Frames Per Second. I have taken the step to publish these figures as Daily statistics for every day since January 1, 2007. As Phillip indicated, we will also be including metrics on Inventory Loss shortly- I expect to include it in August’s data. We will be focusing on discussing these new metrics during my office hours in Beaumont at 10 AM this Friday morning during my weekly office hours- please come ready to discuss them, these meetings have been very fruitful lately and I appreciate everyone’s input. Talk to you tomorrow!

More explanations of the new key metrics below the fold.

(* Edit: Links to XLS and Open Document versions amended- thanks! )

There has been lots of coverage of Phillip’s SLCC keynote and related topics in many blogs and interviews and as he pointed out, there is a huge focus for us on system stability and performance, and these new reports are the beginning of us being more transparent about how we’re doing on those fronts.

The First new tab is “Client Crashes” and it includes 3 new metrics per day – each a percentage of these events occurring per session on the grid. The first metric is “% of sessions that ended abnormally” – which is the percentage of sessions that did not end with an explicit logout event. The second is “% of users that were kicked off a simulator”, i.e. the outcome of the “you have been disconnected from the simulator” error message, which can occur because the sim went down (rolling restarts, planned or unplanned grid outage, or sometimes simulator crashes) as well as communication problems between your client and the simulator you are using, and the third is “% of Logged-in users reporting a crash” which gets populated if you say “yes” to the crash reporter tool and send us details about how/when the crash occurred. The crash reporting is very useful for us to find trends of particular problem code, client hardware, software drivers, and other details that are very useful in debugging. Please continue to send us your crash reports, thanks!

The second new tab is “Server FPS” and it has two metrics regarding simulator performance – “% of daily usage with sim > 35 FPS” and “% of daily usage with sim > 20 FPS”. Server performance can vary anywhere below 40 Frames Per Second, depending on number of avatars using the sim, number of active scripts, use of physics on the sim, and several other factors. Extremely busy sims that are exceeding the number of avatars that can comfortably use it at a single time will see their server FPS decreasing. Any number between 35 and 40 is pretty much maximal server performance, and any number below 20 is a simulator experiencing problems and guaranteed most of the users are experiencing strong “lag”. Typically simulators experiencing lower than 20 FPS will also be experiencing suboptimal “time dilation” figures.

The third new tab is “Client FPS” which will be affected by the Server FPS for the simulator you are using, but much more so it is controlled by your local machine hardware (and network speed) and your client configuration settings. Client FPS is the most direct measurement of your specific experience in second life, whether it be laggy or speedy. It is definitely possible for 2 users on the same server (that has server FPS of 40) to each have drastically different Client FPS due to hardware and settings differences. For residents with slower or older hardware, *please* take the time to turn down your draw distance and change your other graphics settings for second life so that you can experience less lag- you have more control over this than you might expect! We record the average FPS for all client sessions and bucket them into statistical “quartiles”, and report on 3 metrics. “Highest Viewer FPS for lowest quartile” is essentially the best performance that our lowest-performing 25% of users are experiencing. “Median Viewer FPS” is the overall average client FPS. “Lowest Viewer FPS for highest quartile” is essentially the worst performance seen for our highest 25% of users – these are definitely our users with the newer video cards and well-tuned client configuration settings. I know we have users that are experiencing *much* higher than 25 FPS with very fancy new hardware. If you are experiencing great FPS, talk to your friends about your settings and your choice of hardware- you might be able to help them out! In this Client FPS data you can see a great trend of better performance that occurred around the end of March when we release our new texture pipeline code.

I look forward to talking to you more about this new data tomorrow morning, thanks for reading!

About Meta Linden

Data Warehousing and Metrics Analysis for the world of Second Life. A Linden Lab employee in California.
This entry was posted in Economy. Bookmark the permalink.

82 Responses to July 2007 Key Metrics Released

  1. Ann Otoole says:

    i don’t see any client crash information in the excel version.

  2. Suzi Phlox says:

    WTG Meta! Keep the info coming….The more we understand the better we appreciate all the hard work that goes into SL. Thanks.

  3. Ann Otoole says:

    Also the crash metrics don’t account for the current production release client freezing the O/S, disconnecting all open network connections, forcing a manual abend. Therefore the metrics are only as good as the software in respect to handling abends properly. There are a lot of client crashes that are not detected.

  4. Rooke Ayres says:

    The numbers in the client crashes are disturbing.

    On average between 20-25% or 1/5-1/4 of the users crashes daily.

    These are terrible numbers. It’s no wonder everyone complains so much.

    Linden Labs has *GOT* to improve those statistics, or people are going to loose patience with that level of performance and leave.

    Given those numbers, it’s amazing that as many people put up with that on a daily basis. I know I’m getting pretty tired of it.

    Please improve your performance.

  5. DanOfWA Flanagan says:

    I also didn’t see any crash metrics in the OpenDoc version
    mostly population and economy stats

    Being able to see the ‘crash metrics’ would be kinda interesting

  6. Can you please provide the median daily concurrency figure ? That would be very informative. Thank you.

  7. Broccoli Curry says:

    I’m getting 67fps with 400 draw distance on a class 5 private sim with little scripting running … quite delightful 🙂

  8. Ann Otoole says:

    dang broc gimmee some o dat fps!

  9. The copy i got from the link was the old June one no July data.

  10. Simil Miles says:

    I’d like more details for Countries by active such as Quebec inside Canada so that we can better know what are the top languages (I made a graphic for last month, on my site).
    Or maybe you can give the language preference of the client ? that would be much more accurate.

    For population I’d like to see the difference between residents who left Help Island and those who didn’t. According to me 50% of new residents don’t go further than Help Island which means that you have a huge problem, or maybe it’s on purpose and it’s better like that for us.

  11. Is this asumption/calculation for July correct? Looks pretty high:

    24.3% of the active SL avatars (Total: 1,038,501) are female, and they spent 42.1% of all online hours. This means female avatars spent in average 40h online a month, and males only 17h.

  12. Gee and I was just grousing yesterday about no July Metrics, Thanks for releasing it and adding the new data, that is all very helpful. However . . . . . . . is it that much trouble to show the number of premium accounts that have lapsed into unpaid, or been outright cancelled? I know I keep asking for that figure . . . I see under premiums the date and number of premium accounts, but I have no idea how many you are losing in addition to your gains. I don’t know maybe I’m the only one interested in this figure, anybody else feel like that kind of data would be helpful?

  13. I have some client crashes on logout which dont disturb game experience. Some are power failures, most of them are Network Reconnect (new IP after 24h). So the numbers are not as bad as they look. However I did had quite a lot Graphics related bluescreens with older versions.

  14. Cappy Frantisek says:

    Wow ThaBiGGDoGG,

    Nice rant on a blog post. Good luck getting anything back.

    On Topic, I have to agree with Rooke. not a very promising stat about client crashes. Dunno if I would have included that one in a bad month. LOL

  15. 6000 Premium Residents down in July… that’s likely ~8000+ not renewing.
    That doesn’t look very good…

  16. Andrea Faulkner says:

    Good.. less strain on the grid from the less premiums and more room for those underage kids getting free unverified accounts! Yay LL! You’re so efficient. Bring back the Credit Card requirement and let’s have an actual adult grid again. Identity Verification.. not even manditory.. is a useless gesture.

  17. tony comet says:

    And how about paid premium accounts by country…..

  18. Maura Beresford says:

    The client crash numbers are absolutely hilarious. Does anyone actually think that even a tenth of the value of those figures would be acceptable?

    On the other hand, I’m only a couple months old and I didn’t realize that SL had this horrendous population explosion recently. I guess an overwhelming tidal wave of problems is to be expected when the number of users goes on an exponential rampage.

  19. Bradley Bracken says:

    I agree with you Andrea #13. I pray for the day Credit Cards are required on all accounts. Not all problems will go away, but the drop will be huge I’m certain of it.

  20. The XO says:

    Many client crashes can also be attributed to the utter poop hardware some people run on.

    I don’t have that many crashes, and I’m running the Linux Alpha client with all the “scary” options turned on.

  21. Neural says:

    client crashes? Inventory loss statistics?

    Can an excel spreadsheet hold data of such size?

    As for premium member cancellations: As much as I’d like to blame it on client issues, I’m sure some of the cancellations are due to that, I think there are a lot of cancellations because people are catching on to one very important feature. Private Islands.
    1) They are often cheaper. Offering more land, more prims, at a lower cost.
    2) part of the lower cost is that they don’t require premium membership to own on an island
    3) In many cases performance is better
    4) In many cases the covenants are *enforced* (a word that LL doesn’t seem to comprehend when it comes to the ToS)

    Overall, I think there are a growing number of people who like to have their home and/or store on an island run by someone who will actually ban individuals for breaking the rules. And they perhaps are finding stricter regulations acceptable because many island owners want their islands to look nice, and to avoid the “ghetto sim” look that some huge percent of the mainland has.

    LL has driven away premium accounts due to their own folly. Some theorize perhaps that it was LL’s intention to drive people away from the mainland, but somehow I find that only marginally believable.
    One thing for sure, LL does, at face value, appear to be doing a rather good job of shooting themselves in the foot.

  22. Meta Linden says:

    Fixed the links to the Excel and the Opendoc links – thanks Ann (@1) and Nikki.
    @Ann (@2) – those crashes would definitely be included in the overall crash statistics as there was no “logout” action for the session, whether the client process crashes or the machine abends. I personally experience those machine freezes on my mac client also and verified that mine show up in these stats. We are gathering these from the server side so they’re comprehensive – note that these are a percentage of *Sessions* so if you machine configuration is working poorly and you experience, say, 3 crashes in a day, whilst your neighbour stays logged in with 1 session, that averages 3 crashed sessions out of 4 total or 75% crash rate between the two of you.

    @Steve and @Insky – please create requests on the public JIRA under “Web>Public Metrics and Reports” so that we can prioritize them against the other requests I already have? I have created some JIRA tasks off of some of the common blog comments here, but haven’t been able to keep up with all of them. We already have some requests for daily *max* concurrency. Given that our concurrency follows a regular daily pattern, I’m not as certain that median figures would be that helpful, but I’d love to discuss at my office hours.

    @SLEarther : actually the most significant contributor to the decrease in premium accounts was the institution of a process to put on hold accounts that are in arrears with payment to us, and accounts on hold are not counted in the totals. Many of those have since resolved their billing problem, and the overall percentage of accounts affected is still under 6%, but we do never like to see a decrease in premium accounts.

  23. RC Paderborn says:

    Meta: “but we do never like to see a decrease in premium accounts.”

    I understand that. And based on my experience, you may see another decrease when the August metrics come out.

    I had three premium accounts. I now have one. And that one will probably go away as well if Linden insists on charging me tier this billing cycle for land in the LAST billing cycle. Billing support told me that I’d have to pay it even though neither myself or my group owned any land.

    Seems like it should be automatic when selling group land to remove the contribution rather than me having to find out from support and the knowledge base the steps I have to do to keep from being charged for something not owned.

    On the other hand… if you’re billing me tier for land and you’re billing the new owner tier for the land for the same period, that’d be double billing which I believe is illegal even in California.

  24. RC Paderborn says:

    For those that don’t understand my last post…

    When you deed your land to a group, you “contribute” the tier for whatever m2 of that land is.

    When the group sells the land 1 day or 10 years later, your “contribution” remains on the books for the group. This is even though the group may not own ANY land.

    Because it’s still on the books and because the contribution counts against your holdings for the purposes of calculating tier, You will continue to be charged tier fees by the Lindens even though the land was sold long ago.

    That is until some kind soul in Support tells you about this when you complain, you look up in the knowledge base how to get rid of it and do it.

  25. Bobo Decosta says:

    How is a unique resident counted? I can’t imagine that their are only that much alts, as i see lots of people using over more than 20 alts to be able to leech campers.

  26. RC Paderborn says:

    Allowed holdings: 16,896 square meters
    Current holdings: 0 square meters
    Available for purchase: 16,384 square meters
    Peak square meter usage: 11,454 square meters
    Total monthly cost: US$75.00
    Next bill date: Friday, September 21, 2007

    That’s from “my account”. it says I held 11,454 m2 this billing cycle. All but ~3,000 m2 was sold July 14 and 15. This billing cycle started August 21. So my peak usage SHOULD show as ~3,000 m2.

  27. Lucinda Obviate says:

    I think specially as we move forward with the optional viewer updates. Some time line of when new viewers were offered, so we could see the performance comparison of different viewers would be useful. Might be able to then tell if perhaps I am having problems with the current viewer I could look at what was percentage wise for the greater number of user was more stable and performed better. Just thought.

  28. Alf Lednev says:

    Big Deal really, we all know how often we endure reduced or non services. “Lies, Dammed Lies and Statistics” sum it up. The Blog is full of the same complaints and same errors constantly. pretty pictures and columns of numbers are a diversion which may appeal tpo a certain geek element but do nothing to show any real work is being done actually to fix things.

    I know the cr*p “if it cant be measured” etc, the counter is, if you cared about customer service and input, you would already know the problems , they face.

    I see this exercise simply as an attempt to muddy the waters over the ID issue. Something that important to everyone should be sitting at the top of blog list permanently and not allowed to slide away and the 100 comments limit lifted. Its full of either Linden Alts praising it, or a few outraged people hogging the rest of the space. How is that fair and open debate?

  29. Cat Gisel says:

    I think the Second Life Crash Logger is stalking me. Can I file an abuse report? o.O

    Seriously, things that matter, people will whine about, or be diligent about. Things that matter to people they will have opinions about, dear Lindens, not because they think they are crap, but because they WANT them to be better so they can enjoy them more. You should be thankful (I hope you are) that people actually CARE enough. Yes you have to parse out (READ: ignore) all the whining and ***ching, just to get some decent arguements and real issues to look at. And yeah, that’s hard when you have to look at them ALL. (Cause you should, it’s good C/S).

    To that, I suppose it’s easy to get caught up in things, and it’s then that we forget that computers and networks are by nature complex and therefore just plain FLAKY. I am sadly getting used to it, but it doesn’t mean I enjoy it. I just don’t see the need to wast my time with complaining. I crash, I log back in. I crash again, I log back in again. Inconsistencies in the viewer itself I will NEVER understand, but oh well, I am no programmer. I saw a survey for about 2 seconds one day, and before I could even click on it it vanished.

    Hang in there, everyone. On both sides of the fence 🙂

  30. Elwin Jacobus says:

    Statistics are a remarkable tool. They can be used to ‘prove’ just about anything you like. I should know, I’m in marketing and this makes me naturally suspicious of published statistics.

    I applaud LL for including data on client crashes. I get many of these on my almost brand new Compaq notebook. I average 4 per session and this is too high. It is only the social networking, or in other words, my SL friends that keep me in SL.

    Those crashes and the, in my view, prohibitive cost of a premium account are what stop me upgrading my account. Other MMORPGs charge a one off fee for a premium account rather than a monthly recurring one.

    Overall the published statistics put a good face on what, judging by the feedback in the blog and my own experience, is a SL that still has many fundamental issues to sort out. Or to put that in other words; the statistics give the appearance of an attempt at window dressing.

    One of the numbers I look at every time I log in (and with the number of crashes I have that is often) is the number of residents online. I haven’t recorded the numbers but it seems to me that those numbers are declining.

    LL should give serious consideration to the reasons why these numbers might be reducing. I, for one, don’t want SL to fail and am happy to help make sure that SL stays online and continues to be a vibrant international community.

    From the statistics it looks like client, sim and server stability are the key areas of concern. I would strongly suggest that LL invests most of it’s resources in resolving those issues rather than adding even more unnecessary, and unwanted, bells and whistles.

    And, please make robust age verification part of the sign up protocol. Do that and I happily accept the small inconvenience of proving my age retrospectively. As to my identity, that is actually irrelevant. SL is a role playing environment and as such true identity is irrelevant so long as people behave within the rules.

    I sincerely hope that LL remembers why it is that so many of us come to SL. We are here to put RL behind us for a while and be who we want to be rather than who we actually are in RL. LS and RL impact on each other enough as it is without adding even more crossover.

  31. Justkickzazz Taurog says:

    Slightly disturbing numbers in respect to the crashes and the premium accounts not being renewed.

    I am still glad that some Lindonian actually had the ‘balls’ (yes, that’s you, Meta) to reply to the comments as normally it’s a ‘post and run’ affair ^_^

    @13 & 16 Sure…let’s make it YOUR private sim so YOU can be happy, shall we?

    @20 The problem of billing two customers for the same land is not only restricted to group land. It also applies to any land being sold after your billing date (what exactly does the phrase ‘for a considerable amount of time’ mean?) When I gave up my land on main – after being a very unhappy puppy with LL – it was 10 days after the billing date. I had to pay for the full month as well as the new buyers. This can be regarded as a very bad business attitude from LL, so say the least and stay on the friendly side. It remains to be seen when/if this will hit the U.S. courts and how this passage of the T.O.S. will stand !

  32. Alan Bamboo says:

    -hopefully they were all gamblers.
    -Time to make way for the new, serious, more creative or business minded.
    -SL is currently re-inventing/redefining itself

  33. Jdtrue Writer says:

    @at 19 Nick Soyer

    Well not really AT you but your comment lead to mine….

    “Little by little attacking what made some many people join, money & sex… gambling and fantasy.”

    1) Money: Looking at the L$ stats it looks like that is still going.
    2) Sex: Even with the new VOLUNTARY identity validation it can still be had but only by validating or doing it on your own land and restricting it to friends only.
    3) Gambling: Anyone that was coming to SL just for that really needs to call 1-800-Gambler because they are really addicted. (And even if the other 94% of the 5,810 premium members that we lost are former casino owners or addicted gamblers I estimate from the stats trend that their numbers will be replaced in less than a month. Heck give it three months and call it a Christmas gift!)
    4) Fantasy: Every time I see a furry, vampire, person flying or even just skinny people I know that is still going strong…

    Ohh and by the way — the first 3 can all be done in RL. Don’t tell anyone but the sex even feels better LOL.

    The stats that look like ¼ are crashing were scary. But when I think about it stats like that could just be people that insist on running SL on a machine that can’t handle it and keep signing back in over and over and over…

  34. Inigo Chamerberlin says:

    ‘Any number between 35 and 40 is pretty much maximal server performance’

    That’s total BS I’m afraid. Sim servers are SUPPOSED to run at 45fps with script and physics performance, if necessary, automatically throttled to permit this.
    THAT was well thrashed out between us and LL immediately after the introduction of 1.7 (for those who can remember).

    IF there has been an alteration of sim performance parameters it would be nice if Linden Lab technical staff made it clear and explained the new measurement system and parameters – as they were obliged to do immediately following 1.7.

    We can’t accept this sudden ‘move the goalposts’ tactic to explain away less than stellar metrics.

    Nice try though.

  35. Astarte says:

    Nice to get these stats. But I did notice the decrease in Premium Accounts, I know this can be for several reasons including those you mention, but it is still a decline.

    Another question on your stats. seeing as europe especially is the area that is always affected by downtime. is that factored into your calculations or just simply ignored ??

    I also note we still only have 10% of all unique registrations actually playing the game. What happened to the other 90%.

    Finally Do you seriously think a 25% crash rate is acceptable, because I don’t. I experience it on a day to day basis, and if it wasn’t for business interests, I for one would no longer be putting up with it. I would be one of those lost 90%.

  36. S. Voom says:

    oh, oh, oh.
    Downturn of premium accounts – not realy good, its the poorest value ever seen.
    Year Month Premium Residents grows %
    2005 Jan 5.273
    2005 Feb 5.710 8
    2005 Mar 6.287 10
    2005 April 6.669 6
    2005 May 7.039 6
    2005 June 7.548 7
    2005 July 8.231 9
    2005 Aug 8.953 9
    2005 Sep 9.826 10
    2005 Oct 10.772 10
    2005 Nov 11.633 8
    2005 Dec 12.433 7
    2006 Jan 13.403 8
    2006 Feb 14.178 6
    2006 Mar 15.162 7
    2006 April 16.535 9
    2006 May 18.320 11
    2006 June 19.685 7
    2006 July 21.441 9
    2006 Aug 24.702 15
    2006 Sep 28.253 14
    2006 Oct 32.526 15
    2006 Nov 42.430 30
    2006 Dec 49.776 17
    2007 Jan 57.661 16
    2007 Feb 66.838 16
    2007 Mar 75.714 13
    2007 April 83.063 10
    2007 May 89.845 8
    2007 June 94.607 5
    2007 July 88.797 -6

    The grows of the residents slows down on the level of 2005-05:

    Created Unique Residents
    Year Month Population grows %
    2002 Mar 4
    2002 May 5 25
    2002 Sep 17 240
    2002 Oct 19 12
    2002 Nov 36 89
    2002 Dec 67 86
    2003 Jan 109 63
    2003 Feb 129 18
    2003 Mar 167 29
    2003 April 227 36
    2003 May 338 49
    2003 June 527 56
    2003 July 758 44
    2003 Aug 924 22
    2003 Sep 1.088 18
    2003 Oct 1.219 12
    2003 Nov 1.426 17
    2003 Dec 1.663 17
    2004 Jan 2.160 30
    2004 Feb 2.582 20
    2004 Mar 3.024 17
    2004 April 3.816 26
    2004 May 4.713 24
    2004 June 6.101 29
    2004 July 7.042 15
    2004 Aug 8.153 16
    2004 Sep 9.247 13
    2004 Oct 10.343 12
    2004 Nov 11.358 10
    2004 Dec 12.524 10
    2005 Jan 13.817 10
    2005 Feb 15.124 9
    2005 Mar 16.559 9
    2005 April 18.185 10
    2005 May 19.757 9
    2005 June 21.971 11
    2005 July 26.164 19
    2005 Aug 28.725 10
    2005 Sep 38.802 35
    2005 Oct 52.698 36
    2005 Nov 64.776 23
    2005 Dec 77.392 19
    2006 Jan 95.513 23
    2006 Feb 111.846 17
    2006 Mar 127.815 14
    2006 April 155.303 22
    2006 May 179.979 16
    2006 June 235.007 31
    2006 July 298.154 27
    2006 Aug 403.585 35
    2006 Sep 529.076 31
    2006 Oct 778.465 47
    2006 Nov 1.095.959 41
    2006 Dec 1.431.144 31
    2007 Jan 1.990.849 39
    2007 Feb 2.632.041 32
    2007 Mar 3.256.371 24
    2007 April 3.779.566 16
    2007 May 4.370.810 16
    2007 June 5.224.582 20
    2007 July 5.706.958 9

    Is there any thought why?
    Are that the unpopular decisions LL made in the past?
    I personaly think its the gambling-thing.
    From the distance of a non involved market-watcher, it looks like a earthquake. A bit like the real-estate-crises we see at the moment. A business model crash and pulls other businesses with it down. Some expected extraordinary and investet and the first did but the rest payed.

    LL please try to find ur balance again.

  37. Hmm. Trinidad and Tobago completely fell off the ODS sheet…

  38. Nick Soyer says:

    @ Jdtrue Writer

    All I have to say is, Where’s my post, seems to have vanished lmao.
    Who’s kidding who now.

  39. Premiums

    2007 April 83,063
    2007 May 89,845
    2007 June 94,607
    2007 July 88,797

    Land Size mainland
    5/31/2007 160.06
    6/30/2007 166.59
    7/31/2007 178.76

    So while while the market for land “SHRUNK”(no wonder these statistics were delayed) by 6.14% to smaller than it was in May Linden Lab INCREASED the supply of land by 7.31% to 11.68% more than what it was when the market was comparable size AND continues to pour on my land. Lets never be so silly as to assume they actually are interested in stability or know what they are doing. lol

  40. Deltango Vale says:

    Ouch! The decline in Premiums says it all.

    In my opinion, LL has made several major strategic errors.

    First, the anonymous (no payment information on file) accounts. These enabled anyone to enter SL and left LL wide open to a media spotlight on potential abuse. The current ‘age verification’ fiasco is a direct result as LL seeks to plug the hole it created in the first place.

    Second, the failure to develop a system of land creation in line with population growth. From October 2006 to January 2007, when thousands of new residents were pouring into SL, new land was created at an appallingly slow rate. Limited supply and increasing demand drove up land prices accordingly. Then LL dumped land on the market, badly hurting those who joined between January and April. Just as the market began to stabilize, LL dumped more land on the market and another tranche of residents was hurt. This rollercoaster ride of supply mismatched to demand simply generated instability and mistrust.

    Third, the ban on gambling and the anti-sex campaign. This struck hard at the core of SL’s popularity in terms of fantasy and personal liberty (not to mention undermining business confidence as an entire industry was wiped out overnight). LL’s slogan is “Your world. Your imagination.” Virtual, voluntary and adult. Many residents, myself included, are bitterly disillusioned by LL’s change of direction.

    Fourth, continual grid instability. For SL to succeed, it must run as smoothly as Air Traffic Control. Nothing should go live until it is tested to destruction on a parallel grid.

    In conclusion, I suggest that LL hire competent strategists and managers before it loses control of its own creation.

  41. Astarte says:

    Just a short addendum. seeing as I am currently out of game having crashed.. and interestingly the crash logger has crashed as well again.

    But if you think these figures are down, what shall we expect with all the unusability of SL during August. Somehow a think we will see a continued downward trend. And it will also be interesting to see what % figure the crashes reach for the month.

    And Elenora..#10.. Yes if you check back females have spent far more time per month on SL than males for some time.

  42. None of the loss in premium accounts would have been due to the gambling ban, and folks leaving the game in the aftermath? The gambling ban came in 7/25, as I remember, and I reported on it for M2. We had a lot of people reporting that they were dumping their accounts, although that’s awfully hard to verify.

    Meta, is there any chance you can tell us how many of the premiums lost (not the net loss, but gross loss) came on or after the gambling announcement?

  43. Blinders Off says:

    I have heard from most reliable sources that at one time SL had in excess of 600 FPS. That changed when SL opened up the grid to free, non-verified users. LL knocked the FPS per person down to an average of 45, people lagged like a fiend, and they called it a “service improvement” and an “alternate way of measuring FPS”. Uh huh. It just gets funnier and funnier.

  44. Tsu Goodliffe says:

    can we actually do something about these griefers and people who keep recording and harassing residents such as my self on a daily bases, even when you ignore them they find a way to interrupted your daily activitys in second life, Even if linden lab put credit card verification, are you aware that a 13-14 year old can have a credit card, or how many parents would actually put there credit card and real info on a game like this just so there kids could play ? Parents are starting to not care on a daily bases. but that doesn’t stop the fact that we still have people crashing sims and making Second life a more horrible place to be in. I think Linden lab really needs to put in some short of code to stop spoofer Technology, so that the perment bans actually stay, instead of banning them only so they can come back the very next second more Enraged.

  45. Wyald Woolley says:

    @38 Tsu

    Actually I think there is an valid reason to require payment verification beyond some level of age varification. If one were griefing, and were banned, they would need to find another credit card to log back in with. Burning credit card numbers each time you act up will certainly put the skids on behavior, no matter if over 18.

    Hewever, LL really wants the under 18 on the main grid, and wants the land owners to be held accountable if the kiddies see something that they shouldn’t, so nothing will come of using a credit card to raise the bar even an inch to keep the kiddies out.

  46. Tsu Goodliffe says:

    Some of the griefers aren’t kids like the “SA” Goons or m7x or PN there just abouch of people with nothing better to do, if you sit around and ignore them it only fuels them more to attack you.

    I Agree Wyald but the problem is, I could get a credit almost every day its very easy. There are many ways to do it.

    My Main concern is how often these griefers are attacking on a daily bases. Its becoming worse and worse, and all the lindens can say is AR, well I’m tired of ARing when its only getting them banned so they can come back with there alts on there hacked client that allows them to avoid hardware bans etc.

  47. kotivalo hirvi says:

    If SL requires credit card, I will leave. If some parcel owners can require credit card, they should have to pay at least 100L/month tax to LL for that filter.

    No entrance fees to SL-parcels and sims, but some parcel owners might want to require that visitors pay some tax payment to LL to show that they really want to come right there. Again, such parcel owners should have to pay some lindens to LL for that right to install such filter.

  48. kotivalo hirvi says:

    It would reduce griefer harm if we had, among other things:

    20 more sandbox sims, of which 10 are script 6 hour and 10 non-script 12 hour.

    Sandbox sims for 2 month olds, 6 month olds and 1 year olds.

    More script restriction options for owners. Not just sript or no-script, but something in between.

    Owner adjustable speed limits for objects. Combat sim owners might want to set it to 100-200, others to 30 meters per second.

  49. Bradley Bracken says:

    I’m shocked by the comments such as “Ouch! The decline in Premiums says it all.” The increase in premiums occured at a ridiculous rate for several months and one month there’s a drop and people act like it’s the end of SL.

    A 6000 drop in members is nothing to sneeze at, but of course it’s going to happen after a major policy change. It would be expected. Yes, it will probably drop more in August. Considering, however, that even if it drops to 70,000 the totals are still higher than they were just 5 months ago I’d say that LL can sleep pretty easily at night.

    There’s plenty that needs to be fixed but the max exodus the naysayers have said will happen has not happened.

  50. U M says:

    LOL and your suppuse to tell us those tht longed in and the time spent online are REAL! THAT IS BS! Can you people BS the numbers any worse this this? Please don`t insult us with these BS numbers nobody is that stupid!!

  51. U M says:

    Japan 44,847.00 7.99 % 1,563,899.40 6.66 %

    United Kingdom 38,876.00 6.92 % 1,535,157.83 6.54 %

    these number are suppose to be real? OMG there are ahell of alot more UK users then japanese OH PLEASE!!!!!!!!!!! If this is not a fixed and totally lie i dont know what is?

  52. Montana Corleone says:

    @37 Blinders Off : “I have heard from most reliable sources that at one time SL had in excess of 600 FPS”

    Get real. What’s the refresh rate on your monitor? 90Hz? 100Hz? If it’s 100Hz, then it only refreshes 100 times per second, ie skips out 5 out of 6 frames at that rate, and is completely unnecessary. In any case, anythhing much above 25fps the eye doesn’t see (that’s why movie film and PAL TV standard is 25fps).

    Kudos on the numbers again. I’ve been dying to see what has decreased this month, as most numbers were levelling dramataically.

    So 25% of people crash out of a session huh? 10% get booted off the sim… Personally, my numbers are probably higher for the first and less for the latter, but they are averages. 25% would seem pretty high though. If a commercial piece of software you bought crashed 25% of the time you used it, you’d stop using it pretty quick, right?

    Now, if you want this thing to be global, and it sort of has to be, since 8 million users can’t all have a piece of state of the art kit at the same time (and the manufactureres don’t make enough quick enough anyway), then you should be looking at increasing fps at the lower end.

    I used to bounce between 7 and 11 fps. Now I’m 4 or less to 7fps. And why is that? SHINIES…. And a bit more to come from the ID verification data you want to share, Windlight, and any other shinies you have up your sleeve. For many, rushing out and buying a new computer every month is not an option. For some, eg Mac users, it often isn’t part of the culture. Macs regularly put in 8 years or more of service, rather than be scrap heaped after 4 months like lots of PCs.

    Inv loss figures sound good, but I’m perplexed here. If you know which inv is lost, why don’t you restore it? Or is this just the odd few items that are reported via Bug and JIRA, because if it is, pretty meaningless…

  53. Pingback: Aussie population update - 15K and rising

  54. kotivalo hirvi says:

    About fps: When speaking about physics fps, many would like physics fps that is not 40, but 400 per second. Such rate would solve a problem with combat areas, where bullets can go through walls from 7 meters. (Ok, there probably is some solution to that particular problem that would not require any more computational resources, maybe not even linden involvement.)

    I m not sure how tv and movies can be 25fps, but 25 is too low for SL’s non-game purposes. Maybe it has something to do with latency. Maybe 25 would be enough for non-game purposes, with more complicated and efficient software on both ends.

    With current system, I feel that combat areas would need 100 fps for sim.

  55. Deltango Vale says:


    Any significant drop in growth during the key expansion phase of an enterprise bodes ill for the long-term prospects of that enterprise.

  56. Ellen Spark says:

    One thing that puzzles me. I can never seem to find one of the most important Key Metrics. It should really be displayed in the viewer. The one I am referring to is:
    Logged on in the last 24 hours.
    Since we are Global this is an extremely important number.
    Can you tell me where I can find it please ?

  57. Those are fascinating statistics. The number of abends was even higher than I (as someone who crashes all the time) would expect. The gender numbers were interesting… but how do they arrive at them… and is it the avi’s gender or the gender of the person sitting at the ‘puter?

  58. U M says:

    Global status seems alittle strange. How can they measure in this fashion? Number just don`t addup?

  59. Farallon Greyskin says:


    Yes I’d love to see the loggen in in the last 24 as well, as long as we are getting the US$ spent in the last 24, the logged in in the last 24 would compliment that nicely.

    THough I think the metric I would like to see is logged in in the last 7 days. or 2 weeks at most. THe logged in in the last 60 days contains WAY too many of the signup and run people and is a figure that is “inflated” by probably about 10x over the ACTUAL consistant user base.

    THe number of peopl e that log in atleast once a week would be a really great metric. You would still have a fair amount of the sign and run people in it but it would be much MUCH closer to the real repeat connection population (Or maybe # of accounts that have loged in in the last week who have also logged in more than once maybe?)

    It would be really nice to know rather than guessing the ACTUAL population of SL (the current guess is about 200-300k) .

  60. pantaiputih korobase says:

    i guess, many of the problems are jsut due to the number of people online. I bought a new PC some months ago, it runs smoothly as long as less then 40.000 people are online. It crashes without any reason when> 40.000 people are online…. so, it can not be MY hardware.
    two possible solutions: reduce the number of people online for some given time – that would mean no free acounts anymore, let people pay for introducing a basic account with a small monthly fee of let’s say 1 US$, premium members like may thank you due to the momentarily reduced number of online users then. while user numbers increase again after a while you may think about option two and realize it in the meantime: get a stable system which can stand 100.000 online users 🙂
    everyone will be happy again!

  61. U M says:

    Also the statas reflects what the trend is. Or what LL want you to believe. But gesh people are not stupid to believe how some Global status numbers are higher for some parts of the world and lower for others. But then again maybe its all in a Non RL measure since this a VR world.

  62. Ric Mollor says:

    A few comments on the ‘statistics’

    The measurement of in-game hours it does not appear to have any adjustment made for time spent “camping” or AFK. Unless this is acknowledged and some sort of correction made any measurement based on hours usage is almost meaningless.

    25% of sessions ending abnormally? It might be interesting to know how often the crashes occur. There is quite a difference in user experience between a crash after 20 minutes and one after 5 hours. An ongoing measurement of mean/median time per session would be very helpful.

    Finally. Client FPS in the low teens!?! Simply unacceptable numbers and another poorly measured metric. Where and when are these measurements made? Almost everyone will get a high FPS when alone in an empty sim. Throw 50 blinged out avatars in a small area and it’s lag for everyone. There is no situation where users should be getting single digit FPS numbers.

    I recently tested “There.com” and noticed that when a large number of users are together the most distant avatars are automatically displayed with reduced resolution. It’s possible for over 50 people to be together with no perceptible lag. Perhaps LL might want to examine the competition?

  63. Justkickzazz Taurog says:

    why don’t you get yourself your own personal blog and spare us with your cr*p and annoyingly aggressive and biased comments. There are free ones out – as that would most likely fit your cheap style! You seems to be sitting all day long in front of the PC just waiting for a blog entry to be posted just to jump on it and tear it apart no matter what it is about.
    Get a life !

  64. les says:

    # 37 Blinders Off Says:
    August 30th, 2007 at 7:26 PM PDT

    I have heard from most reliable sources that at one time SL had in excess of 600 FPS. That changed when SL opened up the grid to free, non-verified users. LL knocked the FPS per person down to an average of 45, people lagged like a fiend, and they called it a “service improvement” and an “alternate way of measuring FPS”. Uh huh. It just gets funnier and funnier.

    what happened was they capped sims at 45 FPS when they released the 1.7 patch (perhaps the worse release in SL history). They capped sims for one very simple reason. To allow them to stack multiple sims on a single server. when i bought land in yora in 04 it was running 4000 FPS.

    Now they charge 1675 X 4sims just to turn on the server. Then they charge 300X4 for monthly. Pure Gold.

    This is the best Zee linden could come up with to make this company profitable. Linden Lab could really use a little vision…call me some time. Really.

    They should be charging for resources, not raping island owners to cover the free world fantasy.

    …and most lag is local and produced by the land owners because they have no way of knowing how laggy the junk they put out is….

    really LL…. call me and get a plan.

  65. U M says:


    Zee has how long been tying to make LL profitable over the last few years? How many trail and errors of changing rules and other factors in the game? They are making money that is for sure. 1.7 client was a turning put that caused alot more problems even worse the 1.5 client did. But Zee Linden wanted population higher reguardless of the out come how stable thegame was. Well he got it and now look?

  66. TigroSpottystripes Katsu says:

    I hope I can write this comment before the comments are closed….

    is it possible to have each person have (even if local) access to their performance logs that are being transmited to you?

    and a tip for people experiencing frequent “flash-freezes” (sudden spikes of little to no fps at all that then as suddenlly as it came it fades away), I’ve just recently bought more ram for my machine, and now haven’t experienced such phenomena yet (probably just a few times and barellly noticeable), aparently, together with the rest of the system my SL was freezing whenver all the memory physicly avaiable got consumed, and then started responding again once enough memory has been moved from the chip to the hd)

  67. “The measurement of in-game hours it does not appear to have any adjustment made for time spent “camping” or AFK. Unless this is acknowledged and some sort of correction made any measurement based on hours usage is almost meaningless.” — almost but not quite meaningless… Camping and AFK are as integral to the SL experience as lag, being Ruthed, and crashing. And there is no way to tell if the campers are chatting, uploading textures, listening to the stream, etc. Nor is there anyway to tell if the AFK person is truly away from the keyboard or is just doing something which doesn’t involve the keyboard.

    Also, I don’t know that all that many people are camping at any given time given that there are now few incentives to do so… the only reason now is if you want to get your traffic levels up.

  68. U M says:

    56 had this problem then i upgrade to 4 giga DDR2 and no longer have the problem. But the FPS advance is not going to help this thats a simualor problem ( if your not refering to the BASIC status bar results.

  69. U M says:

    @57 how does that explain the ranking of Global users ranked in country logging in? Look at the numbers and tell us if its possible for this to occure? logic tell us UK has more possible users then Japan does.

  70. U M says:

    Is there a stat on Premium Residents for each country? This I love to see.

  71. Pingback: Sula Pinta podcast #010: Katse itään päin « Sula Pinta

  72. The Realist says:

    Here is a message to the decision makers at Linden Lab:

    Some of us had high hopes that SL would become a Web 3.0. It still has the potential, and there are still very few competitors around.

    However, if the Lindens cling to the idea that they should run SL as a monopolist, and a moralist, SL is definitely doomed.

    In my view, the only way out for SL is that the Lindens give up the idea of controlling the contents. Did Netscape try to control the contents of every web server ?

    They should see their system as what it is: an excellent 3D web server and client solution. The way to profit and growth is easy: license out the server code on the open market and thrive on license and support charges. Let hosting companies sort out the legal issues with their clients. They’ve been doing it for ages with HTML content.

    Linden could become another Microsoft if they do it right.

    Think about it, Lindens !

  73. DanielRavenNest says:

    Anyone that wants to find out the average concurrency can find it by dividing the total player-hours in a month by the number of hours in that month (ie 24 x [28-31days depending on the month]).
    Possible reasons for average hours of female players to be higher than male:
    (a) men have more alternative games to play that involve killing things or blowing shit up, thus their time is split between SL and other games.
    (b) women like to socialize and/or shop more
    (c) escorts have an incentive to stay online, since they are making money, once their customer cums at home, he’s done for a while

    I have no clue if any of the above are right reasons, its just what I was able to think of on the spur of the comment

  74. Ricky Zamboni says:

    Hey Meta. You might want to make sure that the economic stats page is pointing at the latest version of the key metrics file. As of right now it’s still linked to the June stats…

  75. McCabe Maxsted says:

    Am I the only one who noticed that the only drop in SL’s premium accounts since its inception was in July, by 6,000 users? And that July introduced the god-awful communicate window? And voice?

    I hate to say it, but we residents told you so.

  76. Nighthawke says:

    One of the problems i’m seeing in the market trends is the post-gambling ban era. The trend in the YTD is downward and growing at an alarming rate. There MUST be another major source of cash that would give the sims a shot in the arm. Right now, I give SL 1-2 years until the P/L ratios make chances of survivability of the company, very poor.

  77. Vivienne says:

    @ 62

    It was not really in July. It was the last weekend of July, when LL messed up the grid with this 1.18 voice server/client crap. The August data will most probably show an alarming decline of premium accounts. And they still are not really back on track with the software. I only hope they´ll get it straight.

  78. Pingback: SL demographics (July 07) « netsifter

  79. kariette says:

    Hi! Was wondering where I can find past key metric files? Thanks!

  80. Pingback: Introducing new Service Quality Metrics « Official Linden Blog

  81. Pingback: September 2007 Service Quality Metrics Posted « Official Linden Blog

  82. Pingback: Wikinomics » Blog Archive » Second Life and government

Comments are closed.