Open Source History: What if GNU and Linux Had Cloned MS-DOS, Not Unix?

Here’s food for interesting thought: What if the Free Software Movement had built a replacement for what became Microsoft (MSFT) Windows, rather than an operating system that eventually became the one open source fans know today as GNU/Linux? Would Bill Gates still have ended up a billionaire, and Microsoft have imposed a monopoly on the PC world? Read on for some reflection on what might have been had a few things gone differently.

First, let’s run through what actually happened. When Richard Stallman started the GNU project in 1984, he intended from the beginning to write a clone of the Unix operating system. He explicitly rejected the notion that GNU might instead aim to copy an operating system like MS-DOS. As he wrote in the February 1986 GNU newsletter, platforms like DOS, although “more widely used” than Unix, were “very weak systems, designed for tiny machines.”

That remained GNU’s attitude through 1992, when the project cautioned users curious about running its software on PCs, “We do not provide support for GNU software on microcomputers because it is peripheral to the GNU Project.”

It was not until 1993 that GNU did an about-face and started distributing its software for MS-DOS environments. That change meant that GNU had started taking the PC market more seriously, and saw the need to offer software designed for the inexpensive microcomputers that, by that time, were taking over the market. But waiting until 1993 to do so made GNU late to the PC party, and arguably helps to explain why Linux—which Linus Torvalds designed from the start for PCs—ended up as GNU’s kernel. But that’s a longer story …

Why Unix?

To note that GNU ignored the PC market and the influence of software like Microsoft’s until 1993 is not to fault it. Although it may seem obvious in retrospect that PCs were destined to become so important, their significance was much less clear in the early 1980s, when Stallman was planning GNU. By that time, the microcomputer market was rapidly growing, with Apple (IBM), Commodore and IBM (IBM) leading the way. But PCs were for hobbyists who lacked the time or interest to use computers to their full potential. Heavy-duty computing still took place on large, expensive computers, like Digital Equipment Corporation’s VAX line.

And, anyway, Stallman’s ultimate goal in founding GNU and the Free Software Movement was not to take on commercial software companies. (On the contrary, one of the less well-understood realities of GNU’s history is how eagerly the project promoted the use of its software in conjunction with commercial ventures by companies such as Cygnus Solutions, which showed starting in 1989—well before the likes of Red Hat rose to prominence—that it was possible to make money using code that most people today would call open source. But that’s fodder for another blog post.)

The only thing Stallman really cared about when he founded GNU was saving hacker culture from what appeared, in the early 1980s, to be its impending doom, as a result of the breakdown of the community that had existed around AT&T Unix when it was a free operating system, along with other changes taking place in locations including MIT’s AI Lab, where Stallman worked.

Hackers didn’t care much about software from companies like Apple and Microsoft. They wanted to be able to continue using Unix with the same liberty they had enjoyed before AT&T began selling it as a commercial product in 1984. So, to save hacker culture, Stallman and GNU needed to build a free Unix, not a clone of MS-DOS.

What If … ?

But what if GNU had aimed to be just like MS-DOS, and the developers had released an operating system for PCs that anyone could obtain for free, and whose source code could be freely shared?

If PC manufacturers—and, importantly, the companies that dominated the very commercial PC software market—had adopted GNU’s system and business model, the history of personal computers very well may have turned out quite differently. Microsoft would have faced enormously tougher odds cornering the market, and GNU’s code might have ended up running the world of microcomputers.

Linux also may never have been necessary. It became the kernel that GNU ended up using to tie together the rest of its software because GNU’s own kernel project, the Unix-like Hurd, became hopelessly overcomplicated. GNU may have had a much easier time writing an operating-system kernel if it had only PC hardware to contend with.

Of course, GNU code, along with Linux, did end up playing a huge role on servers, where Microsoft never established such a dominant position. But that’s a very different market from PCs, where the near-monopoly that Windows enjoyed for more than a decade—and may continue to hold today, depending on how you define “monopoly”—arguably had quite negative impacts on the pace of innovation, as well as users’ experiences.

On the other hand, maybe Bill Gates’s marketing and business acumen would have assured DOS’s and Windows’s success even if they had had to contend with a free, “open source” alternative from GNU. (Stallman and the GNU developers adamantly call their products “free software,” not “open source,” but the former is a confusing term in this context.) Few people would argue that Windows proved as popular as it did because it was the best operating system from a technical or usability standpoint; instead, Microsoft management simply did an exceptionally good job hurting the competition in ways that did not require clever coding.

All the same, it’s a question worth asking: Would GNU(/Linux?) be the only operating system in the PC world today if the project’s founders had made different decisions 30 years ago about which platforms to emulate? There’s good reason to think so.