You can't keep tacking on new features to an old OS indefinitely

Microsoft Chairman and Chief Software Architect Bill Gates' appearance last week at his company's remedy hearings had me on the edge of my seat: Gates was a liability during the original Microsoft antitrust trial when the US Department of Justice (DOJ) played embarrassing excerpts from his videotaped testimony; I was curious about how he would present himself this time. Apparently, Gates and company didn't expect the earlier testimony to ever see the light of day, and the person who appeared on tape bore little resemblance to the public perception most people have of the man. Instead of seeming intelligent and technical, the Gates in the videotape appeared sullen and uncooperative, and he appeared to have little idea what was going on with his company.

Well, the Gates who appeared in court last week was a different person. This Gates vigorously defended his company's right to innovate and, perhaps most controversially, fought against the nonsettling states' request to force Microsoft to produce a modular Windows version that would let end users, PC makers, and IT administrators add and remove middleware products such as Internet Explorer (IE), Windows Media Player (WMP), and Windows Messenger. Gates said such a requirement was impossible and would force Microsoft to take Windows off the market.

But perhaps that outcome isn't such a bad idea. For the past decade, the industry has watched Microsoft meld its legacy Windows products with Windows NT technologies, and the latest Windows OS—Windows XP—is the combination of these two product families. NT provides the sophisticated low-level services enterprise IT departments need in a modern OS, but most of the fluff (e.g., the UI, IE, and the digital media functionality) came from outside the NT team. In giving us the best of both worlds, Microsoft seems to have stripped the soul from NT by layering the core services under mountains of other garbage.

I've written about NT's origins and the ways that Microsoft has compromised the OS over the years, such as when the company made IE (then-buggy and unreliable) a required component for installing key server products such as Microsoft SQL Server or IIS. And in XP, the needs of consumers now seem to outweigh the needs of the enterprise. Microsoft has relegated NT—once the domain of businesses, developers, and other technical users—to the barely mentioned underpinnings of a system designed to not crash while Johnny is blasting space aliens or mom is ordering groceries online: It's a sad state of affairs.

So given Microsoft's recent security strategy, perhaps the time has come for the company to walk away from Windows in the enterprise and design a replacement that offers binary compatibility but none of the foundational problems. Remember, NT was a brand new world when Microsoft developed it in the early 1990s, but back then, the big connectivity concern was LAN Manager-based networking in small businesses, and security wasn't high on the priority list. Perhaps Microsoft needs to start thinking about another grassroots development project—one rooted in security—that could replace NT. Almost 15 years have passed since Dave Cutler wrote the requirements for NT, and that product was supposed to offer MS-DOS, OS/2, and POSIX compatibility as well as support for RISC processors and other technologies so far-out-of-date today as to be almost ridiculous. You can tack features onto an existing product for only so long before it's time to start over from scratch.

Interestingly, the Linux world might create that replacement OS first. I'm not sure I believe the Linux security promise, but Linux has a decent reputation in certain areas, and it's a viable alternative in various situations. NT interoperability has been a Linux goal for years, and various options are available that let you integrate Linux servers into NT-based domains and workgroups and even use a Linux server as a domain controller (DC). On the software front, various conversion technologies are also available that let you move ASP-based Web sites to Apache, for example, or interoperate with SQL Server databases. And earlier this year, a small Linux company released the software behind the Windows-compatible Lindows OS, which lets users run Microsoft Office, IE, and other Windows applications on a Linux desktop system. As Linux' ease of use improves, cost becomes more of a concern, and Linux can certainly be cheaper to deploy than Windows—a crucial deciding point in these economic times.

I don't think Windows will go away any time soon, but finding viable alternatives is possible now, more than ever. If Microsoft is serious about embracing security, perhaps the company should let go of its Windows cash cow and start anew. XP might be secure enough for the home, but it seems increasingly insufficient for the needs of the enterprise. And if the company doesn't start working on a solution now, it might find Windows collapsing under a mountain of security exploits and vulnerabilities far more damaging than any nonsettling states' plan.