Today I see the Wiki page on Third Party Viewers (TPV) was updated last night. We now have more information on how the Lab ranks the viewers. Plus they make a good point about variables I did not consider.
Oz Linden added that a viewer’s crash rate is the percent of that viewer’s sessions that end in a crash. So, whether there are a thousand people using the viewer or ten thousand, the rate is representative of that viewer. This means there is no popularity factor in the number. More users of a given viewer will make the viewer’s rate more accurate and representative.
Then there is the confusion about which version of the viewer has that crash rate. Oz made some changes to the wording rewriting a section to read, “The rate used for each viewer is that of the highest version number for that viewer for which there are at least 200,000 minutes of use in the sample period (a one week sample from Monday to Sunday).”
A paragraph states that the crash rate for any given viewer may vary depending on the operating system one is using. So, one viewer may have a very low crash rate on Windows and a high rate on Mac or Linux, or the other way around. I take it that the published crash rate for the viewer is the percentage of crashes for a given viewer across all operating systems.
If you are looking for the least crash prone viewer, the Linden TPV list is ONLY a hint. It is a good hint, but only a hint.
I also want to point out that many viewers in use are Beta versions. Kirstens and the newer Imprudence (named experimental) are both considered experimental viewers by those developing them. The Linden Lab viewers for projects are experimental viewers used to test new in-development features. Others, Like Firestorm, and the SL Development versions are alpha level software that residents tend to use as if they were beta versions, or at least I do. The stability of these viewers can change drastically from version to version.
If you listen in on the various support groups, you’ll generally hear, at least, a couple of people a day swear off a viewer because they are having problems with it. Then announce they are going to such and such viewer. If you read the Answers forum you’ll see some people post everyday that for weeks they have been crashing and having problems. These are incidental empirical experiences by residents that often never consider what THEY may be doing that contribute to their viewer problems. For instance, running Outlook or Seti/Einstein/Boinc and the viewer concurrently is just asking for problems. Failing to keep one’s video drivers updated also creates problems. When the viewer crashes not rebooting the computer will lead to all sorts of strange problems.
I think it should be obvious that numerous factors contribute to the stability of one’s viewer. All those factors are contributing to a viewer’s crash rate. So, the Linden crash rate is NOT a pure representation of the viewer. The number is an aggregate of all the factors that affect viewers.