When I look through MSR's papers, I see a pretty massive level of output, across many of the domains you say get most of the money. Of course there's going to b

When I look through MSR's papers, I see a pretty massive level of output, across many of the domains you say get most of the money. Of course there's going to be information that isn't published. There's lots of information not worth publishing. As a matter of defense against patents, though, MSFT has a strong interest in insuring that its basic research is thoroughly documented in public, establishing prior art.