When it comes to the software industry, few academics have the knowledge and expertise of Michael Cusumano. Specializing in strategy, product development and entrepreneurship in the software industry, Cusumano is Sloan Management Review Distinguished Professor at MIT's Sloan School of Management and has authored eight books, including "Software: What Every Manager, Programmer and Entrepreneur Must Know to Thrive and Survive in Good Times and Bad." He also has consulted for more than 50 major companies worldwide and has been an adviser to numerous startups.

In an interview with John Roberts, CRN director of editorial research, Cusumano gave his views on the delay of Windows Vista, the escalating battle between Microsoft and Google, and how the Redmond, Wash., software giant can regain its competitive edge in software.

CRN: You have said that the latest delay in the release of Vista shows that Microsoft has "lost its way." Please explain.

CUSUMANO: I was referring to the early days when Microsoft had small teams of excellent people who could build products in a timely fashion. In those days, Windows was still 4 million or 5 million lines of code, with maybe 200 developers. But things have wildly expanded, and they've become bogged down by teams of thousands of engineers and tens of millions of lines of code in their products. And these days, they seem to be doing things more for strategic and political reasons, not for technology. Microsoft clearly made some adjustments after [postponing] Longhorn, but the newest delay in Vista suggests that they have not really made any fundamental changes.

CRN: So you think it's more a case of rearranging deck chairs, so to speak, rather than Microsoft changing its business practices.

CUSUMANO: Right. They have made some changes, and that helped them get a beta version of Vista out. But within a year or a year and a half or so, they will probably be back in the same situation of too much to handle in terms of code, too many people and an inability to stabilize the product. By this time, Windows should already be a pretty mature product.

CRN: You have also said that Microsoft needs to build a new operating system from scratch because the Windows code is too bulky and has too many holes. Do you see any signs that Microsoft is debating this idea internally or that it might already be starting work on a from-the-ground-up rebuild of its OS?

CUSUMANO: I have no evidence that they are scrapping Windows, but one never knows. They threw out a couple of years of work on Longhorn and went back to the Windows 2000 server code base, which is now the base of Vista. There was a time in the 1990s when they did build a new version of Windows from scratch. That was Windows NT, which was launched in parallel to the existing desktop Windows product and grew out of Windows 3.0, 3.1 and Windows 95. But the system grew too large, and it was never really designed to be a network operating system or have rock-solid security. So they have had to make lots of ad hoc changes to the Windows NT code base over time to try to adjust threats from Linux or to accommodate the Internet. So it ends up being a patchwork of code.

At some point--and I think sooner rather than later--they need to build a new operating system to meet current requirements, to be a true network system that is secure and not prone to hacking, and [one that] could handle the different types of platform concepts. They are moving toward that, but they're doing it too incrementally. I'd like to see Microsoft go back to the old culture of putting smaller groups of their best people to work on a problem. This is the kind of culture you have at Google, and they are doing all sorts of wonderful things very quickly with small teams of really talented programmers.