Actually, one reason it's stagnated technically because Linden Labs aren't nearly as good at UI design as they think they are and lots of users are on forked older versions of the client software that predate their last UI redesign and don't support new features. It got so embarassing that they disabled the ability for users to advertise what viewer version they were running.
Sad. I worked at linden from 06-08, and later, when I found out that Viewer 2.0 had been developed in secret and the source thrown over the wall after release, I was stunned by the lack of awareness about how upsetting this kind of behavior is for open source developers who were working on the previous codebase. Even worse is the fact that those developers are working for no pay.
Forks matter, and dilute the mindshare of the already small number of people that are actually capable of contributing to an open source project. I don't understand why there are so many companies (google with android, for example) who think it is ok to develop major new versions in secret and then just toss the code over the wall. It is hard enough to follow the changes to a large open source project, but at least keeping up with commits in real time as they happen makes digesting the changes easier. Lots of small bites to swallow spread out over a long time. With the huge source dump style, I would argue that it is basically impossible for even an extremely competent developer to understand all the changes, and the ramifications of all those changes.
Also, it's Linden Lab. Linden Lab. Not Linden Labs. That used to drive me crazy, I guess I shouldn't really care any more. :-)
