I can't argue with your observations about Sakai, except to add that it's really too early to pass judgment. As an employee of a SEPP institution, I can tell you there's a lot more activity apparent from the inside. But, given how hard it is to develop a mature set of applications and frameworks --not to mention a mature community-- I think Sakai has a couple of years to go before they have a shot at competing with dotLRN in terms of both feature-richness and quality of implementation.
As you point out, OSPI, though now related to Sakai, is its own project and is somewhat further along in terms of both the software and the community. It sounds like you know more of the details of the system's guts than I do; I can only tell you that one of the more innovative SUNY campuses are experimenting with it and have been giving positive reports so far.
