I played SL since 2006 and until 2009 or so, and I still keep up with it from time to time. I think one problem it has is that it tries to do a lot of things, and does none of them well. Building was fun for a while, but you hit a limit with what you can do with their in-world tools, and now they allow users to upload assets created in 3rd party 3D modeling tools (anything that can export to Collada format)
Users are slowly making SL better but it almost seems in spite of Linden Lab, not because of them (the improvement from LL are few and very far between)
I think its because SL is not growing wildly. Yeah, they generate $700M from the virtual economy, but what do they do with it? Without a focus on growing, there isn't as much of a focus on innovation. Plus, as the article states, they've tried a lot of things to get past the 1M active players number and haven't cracked it yet. This discourages them from trying anything expensive to stimulate growth, like improving building or implementing a more realistic physics engine to draw people into a virtual world that looks real and immersive.
Simply put, it looks like they're unwilling to take any risks to make significant changes to SL.
Users are slowly making SL better but it almost seems in spite of Linden Lab, not because of them (the improvement from LL are few and very far between)