Be aware - some algorithms described in publications on research.microsoft.com can be patented by microsoft (and this fact not referred in publication itself, google patent search can be useful for this).
Isn't there a requirement to mark every publication of a patented idea? Did Microsoft just invalidated those patents? (Do they even know what patents apply?)
I copied down this quote from somewhere earlier this year but I can't find the source now: "One of the things they tell us developers here at Microsoft is that we should never read any patents. That's because you're liable for triple damages if you knowingly infringe on a patent."