MS research chief holds court in London
Nathan Myhrvold boasts about bloatware, among other things...
"The second most famous person in Microsoft" was introduced to an audience in London yesterday afternoon by the Editor of the Economist. But it wasn’t Steve Ballmer (nor Bob Herbold, who fancies himself as number three). The number two famous person is Nathan Myhrvold, Microsoft's chief technology officer. And Myhrvold certainly has some claim to this status. But despite his being the guy who runs Microsoft’s research operations, Microsoft’s reputation for the introduction of (genuinely) innovative products remains dismal. Myhrvold was in London to talk about the future of software, and to this end he romped through his version of the history of computing, attributing modern computing to Von Neumann, although he was forced to retract this in response to a questioner who mentioned Alan Turing. Myhrvold introduced a number of laws modestly named after himself, all concerned with the theme that software expands to equal or exceed the size of the container. This explains a lot. Microsoft likes code to be big: it forces users to buy new computers, and hence more Windows, more quickly. Myhrvold was proud that NT code was increasing in size by 34 per cent/year, about the same rate of increase for the number of bytes per dollar of DRAM, it turned out. In his potted history of software milestones, he mentioned Java, although it seemed to have dropped off his slide. So far as what Microsoft was doing in the research area, Myhrvold had very little to say. There is no flagship product that was conceived in Microsoft's research department that has become a Microsoft product (although judging by the intellectual level of his accompanying video, perhaps the now defunct Bob or the talking paper clip came from his stable). Myhrvold claimed that "every major Microsoft product from Windows 3.0 on has technology in it from [Microsoft] research" but he was not giving many examples. He mentioned speech synthesis and speech recognition, but these do not seem to have appeared yet in a Microsoft product, although they are ten-year-old technologies. There are 350 people in Microsoft research in Redmond, San Francisco, Cambridge and Peking. It seems that the research group is still slaving away trying to produce a wallet PC, but nobody seems to have thought of buying Bill a Psion to satisfy his craving. Myhrvold fell back on a shaggy dog story, which seemed to be true: "The sheepdog can take commands from a very large set; it can do physics, applications, intelligent problem solving, it can learn and be trained – all aspects of intelligence. In each case the sheepdog does a vastly better job than today's computers. From a commercial perspective, if Microsoft were as smart as a sheepdog, we'd sell a hell of a lot more." Woof, woof. ®
Sponsored: DevOps and continuous delivery