Peter Merholz observes that R&D departments within large corporations, be they in the pharma or software industries, have a hard time getting their research papers and prototypes translated into actual products. I don't necessarily agree with Peter's admittedly cursory opinion of Microsoft Research, whose output actually happens mostly under the hood in the database space, with patents to boot. (Whether Microsoft eventually reaps all the benefits from that research remains to be seen -- look how Oracle was built upon IBM work or how frustrated TI people left to create Compaq.) I'm also curious about what will come out of the new, smallish Greenhouse created by Jeff Raikes "to identify and cultivate new software applications."

Whatever happens in the specific Microsoft case, the larger point seems to hold. In many industries big companies should probably focus on being platforms or carriers focused on logistics, marketing or distribution, from which smaller companies can take off and innovate. Let the small fish worry about granular customer requests and the concrete products to match them, because the big whales are not going to be responsive enough anyway.

Instead of gobbling those innovative start-ups and turning their output into the same old boring morass, the behemoths might act as service providers but let product development remain in separate, nimbler structures. And tiny but committed software developers such as Ranchero Software or NewsGator Technologies are probably more in tune with the official UI and development guidelines provided by Apple and Microsoft respectively, than people writing apps for these vendors themselves, so that semi-independence would not necessarily mean a loss of alignment.