Hmmm.. open sourcing SL server code... it would probably take a significant rewriting, since right now, Second Life operates on the assumption that both the regions and the central servers are run by a single operation, and a lot of stuff unique to Second Life is hard-coded in. In addition, there are the architecture issues I've mentioned above. Plus, why bother when there's already OpenSim? Sure, the open source community might contribute patches and fixes to the server code -- as they do now for the viewer -- but it takes a great deal of effort to re-integrate these things back into the commercial version.
Meanwhile... image if Microsoft, when they released their Windows Server, didn't support HTML and the World Wide Web, but a totally different standard, the MS Web, say. Since the Web was already out there, and with a massive head start, it would have been really hard for Microsoft to gain adherents to its own, incompatible system. Not impossible, but a really really big struggle.
For an example, look at what Microsoft is trying to do in the mobile space, trying to get people onto its own, proprietary platform -- and barely eking out a 3% market share as a result. Of course, it's not really practical for Microsoft to try to adopt the Android operating system, since it's an operating system company! Although I guess it could get into the hardware business and start manufacturing Android-powered handsets... it does make the Xbox, after all...
Linden Lab, however, isn't in the server software business. It's in the hosting business and the community infrastructure business, and can use OpenSim as well as its own server software -- the same way the Microsoft Windows server uses the World Wide Web protocols, and in the places were Microsoft is in the hosting business -- its Azure cloud service -- it supports Linux deployments.