Gary Kildall and what might have been

Last Updated on February 21, 2022 by Dave Farquhar

I didn’t have time to write everything I wanted to write yesterday, so I’m going to revisit Bill Gates and Gary Kildall today. Bill Gates’ side of the DOS story is relatively well documented in his biographies: Gates referred IBM to Gary Kildall, who for whatever reason was less comfortable working with IBM than Gates was. And there was an airplane involved, though what Kildall was doing in the airplane and why varies. By some accounts he was meeting another client, and by other accounts it was a joyride. IBM in turn came back to Gates, who had a friend of a friend who was cloning CP/M for the 8086, so Microsoft bought the clone for $50,000, cleaned it up a little, and delivered it to IBM while turning a huge profit. Bill Gates became Bill Gates, and Kildall and his company, Digital Research, slowly faded away.

The victors usually get to write the history. I’ve tried several times over the years to find Kildall’s side of the story. I first went looking sometime in 1996 or so, for a feature story about Internet misinformation I wrote for the Columbia Missourian‘s Sunday magazine. For some reason, every five years or so I end up chasing the story down again.

Late in his life, an embittered Kildall started to write a memoir. It was never published, but author Harold Evans was able to gain access to it and used it as source material for a chapter about Kildall in the book They Made America, a book that tries to present the also-rans alongside the rich and famous.

Microsoft called Evans’ account one-sided, but that’s the idea. Entire books have been written about Gates and Microsoft; Kildall gets a book chapter every decade or so. The accounts in Fire in the Valley, Accidental Empires, Hard Drive, and Gates collectively tilt in Microsoft’s direction, though they’re much more complete than the stories I found on the Internet in 1996. Gates’ self-serving “Gary went flying,” which is about all he’s ever said publicly about the DOS deal, is one-sided too. Kildall elaborated at least somewhat, with his famous question why print string (function 9, or by some accounts function 6–but function 9 appears to be correct) ends in a dollar sign, and his assertion that he was the only one in the world who knew why.

Perhaps that dollar sign has something to do with Kildall’s Easter egg, long rumored to exist but never widely disclosed.

Gary Kildall never widely disclosed the meaning of the dollar sign, or the key sequence that triggered the Easter egg, causing Gates partisans to argue that they don’t exist, and Kildall partisans to make apologies. Easter eggs were a common practice in 8-bit computing, and disguising them in the code so that other developers and managers wouldn’t see them was part of the art. At the time Kildall died in 1994, CP/M itself was a dead product but its direct descendant, Novell DOS 7 (a renamed version of DR-DOS), was attempting to compete with MS-DOS 6.22 in the market.

In July 1996, Novell sold its DOS product to Caldera, then a Linux distributor. Caldera, now known as SCO, was interested in Novell DOS for two reasons: It wanted a DOS clone to ship with its Linux distribution, and it was more comfortable suing Microsoft than Novell or Digital Research was. (This litigious streak eventually led to SCO’s demise, when SCO decided to sue IBM and several other large companies for copyright infringement because Linux used some of the same header files as SCO Unix, which turned out to be in the public domain anyway, if they were ever copyrightable in the first place. But that’s another story entirely.)

Had Gary Kildall still been living from 1996-2000, it’s possible that Caldera would have asked him to testify. The lawsuit, which was settled out of court in 2000, centered on Microsoft tying DOS to Windows with the Windows 95 product even though there was no technical reason why Windows 95 couldn’t sit on top of DOS like Windows 3.1 and previous versions had. Caldera demonstrated Windows 95 running on top of the rechristened DR-DOS 7.01 starting in 1998.

Perhaps, had Kildall still been alive and willing to testify, Caldera’s lawyers would have broadened the scope of the suit. That’s just speculation. By my understanding, the statute of limitations on any copyright infringement would have run out in 1985 or 86, so the question is what else Caldera, as Digital Research’s successor, might have stood to gain with this information.

It’s also speculation that Kildall, while still living, would be unwilling to talk about the deep internals of a commercial product that was still on the market, but it seems reasonable to me. Novell DOS was discontinued in September 1994, about two months after Kildall’s death.

If I sound overly sympathetic to Gary Kildall, I hope you can forgive me. Gary Kildall was very much ahead of his time. In the early to mid 1980s, he was messing around with things like optical drives and multitasking. There probably wasn’t much that Kildall could have done to make optical drives become affordable in the 1980s, but the idea of him introducing multitasking to the masses sometime before 1985 is tantalizing. By the time IBM released its PC/AT in 1984, Kildall could have had a multitasking operating system (MP/M) and a graphical interface (GEM) ready to take advantage of it–the functionality that Windows 3.0 delivered in 1990, if not better. The industry could have gained six years or more.

And I really believe that with a relentless tinkerer like Kildall alive and thriving in the computer industry, the first decade of the 21st century wouldn’t have been the doldrums that it was. And I think I regret that even more than I regret those lost years of 1984-1990, which introduced us to multitasking and the GUI, but with flaws.