Do 1/4 of sessions really crash?

There were some interesting and rather candid metrics for SL recently posted on the blog.

Out of curiousity, I did a correlation analysis of LL's reported "% of sessions that ended abnormally - which is the percentage of sessions that did not end with an explicit logout event" and found two surprising things. The first is that about one quarter of all SL sessions crash and that there has been no significant change in that number throughout 2007.

I'd love if someone could show me that I've misinterpreted the report.

I know I get far fewer crashes than that.

Not getting an explicit logout could be an SL crash, a PC crash, a power failure, the ISP going belly-up for a minute, people with open source/libsl clients that don't work very well yet, etc, etc, etc..

Lately, it has been close to 50% for me, so that doesn't surprise me. I'd say 3/4 of the crashes are during normal activities (camera manipulation, usually) and 1/4 are from the sim crashing or being reset.

I crash maybe one out of every hundred sessions. Kudos to LL for being so open with their statistics.

Like Galena, I have lately been plagued with regular crashes when I'm alt-zooming. But I think that Sindy has also hit the nail on the head, not all unexpected events are 'crashes' surely?

crashes are based on were you are, if your in mainland/low class sims you will be prone to more crashes because it can't handle what people are doing or have on them. if everyone was in the newb ava and nothing attached to them you could fit more than 300 people in a class 5 sim without any problems

Except they track "client kicked from region" separately than abnormal ends (crashes). Unlike the crash data, the kicking has trended downward significantly over the year. As far as I can tell, these two phenomena are unrelated.

I've had lengthy periods where the crash rate was close to a hundred percent. Multiple month long periods. Mercifully this was a while back.

These numbers probably include a fair percent of users that have marginal or submarginal equipment, such as low ram, integrated video, and have systems infected with viruses and other malware, lots of programs running in the background, etc.

I say that because I think that is pretty common for computer users. There's lot of folks that use old computers, they let their kids use them and the kids don't use any judgement about avoiding malware and the adults not much better, they never update their drivers, etc.

Of course I have no idea how much a factor this would play in SL crashes.

I don't find the idea that a quarter of SL sessions end with crashing to be at all implausible.

I wonder if there is any way to determine what the crash rate is for people that have used SL for different periods of time. Those numbers might be including the large number of people that leave SL for good shortly after entering. The crashing might be the cause of a lot of the departures. I wonder though what the crash rate is for folks that have been in SL, committed to it, had time to get as good a video card as they want. had time to learn in the SL forums if they had no prior knowledge of such things as how to update drivers, install anti-malware software properly so their computers aren't crippled, etc.

I've only crashed during attempted video playback (which I never considered important) although I've not played much since 2005.

I too think that it is very good of them to release the data that they have. It is my fervent hope that someone at LL is using these metrics to determine if things are improving. It is certainly a step in the right direction.

Since it is late and I'm editorializing, I used to think that LL botched almost everything they tried; recent news is causing me to re-evaluate that.

At a rough guess I'd say 20-30% of my sessions end in a "crash". But 9/10 of those crashes happen after I click Quit.

SL often freezes/hangs while in the process of logging out, and I have to use task manager to get out. It usually offers me the option of sending a crash report next time I log on, so I'd guess those "crashes" would count towards the 25%. But they are not a problem when playing the game, and although my session may show up in the stats as a crash, in fact it happened after I'd already decided to log out, so my session did not end because of a crash.

"abnormally explicit logout even"t could be the result of people running multi avies 5+ or more. Since Bots are very population these days in money making happenings.Then having system "crashing "abnormally" when their system can`t handle the load on their computers. "Explicit logout event" where island owners are throwing people off their land for reason of littering the island with ALTS.

I have to agree with the two points made here. I don't know for sure, but I suspect that the ones with the fewest crashes are the early residents. I can't say that all of the oldbies are lucky, but it just seems that whenever people are having problems, its mostly the ones who just got here, or have been here less than a year. In my time here, the only times I've really had issues, everyone had the same problem.

I've noticed that too, especially in my Live Help days. I'd come up with two hypotheses about it. The simplest is that if you are an oldbie, SL has had to work reasonably well for you or you'd not be an oldbie.

The other - very hard to prove - is that older inventory items (and thus the inventory directory itself) seem to load more quickly than newer ones.

I'm uncertain what data you'd need to support or reject these guesses.