I think it's highly unlikely the patent office would ever have the sophistication necessary to distinguish a truly new approach to an algorithm or architecture from a derivative one.
Additionally, there is so much depth and breadth in CS that innovation happens everywhere and is often hidden in the bowels of universities, garages, and offices. Even an apparently novel invention is unlikely to be invented only in one place, or to be invented first by the entity that files for a patent.
Additionally, there is so much depth and breadth in CS that innovation happens everywhere and is often hidden in the bowels of universities, garages, and offices. Even an apparently novel invention is unlikely to be invented only in one place, or to be invented first by the entity that files for a patent.