Orem firm says bug derailed Caldera softwareIn 1991, when a competitor threatened to break Microsoft Corp.'s lock on desktop software, Microsoft engineers discussed an unusual counterattack: a software bug to be hidden inside an early version of Microsoft Windows.
In a Sept. 30, 1991, message about the plan that referred to members of his team in the shorthand of electronic mail, David Cole, head of Windows development, told another executive that "aaronr had some pretty wild ideas after three or so beers - earleh has some too." If the bug detected a rival's program, he further wrote, it would "put competitors on a treadmill" and "should surely crash at some point shortly later."Cole also warned that the existence of the bug had to be kept secret. "We need to make sure this doesn't distract the team for a couple of reasons," he wrote. "One, the pure distraction factor, and two, the less people know about exactly what gets done, the better."
Cole's e-mail came in response to a challenge from what Microsoft saw as a clone of its DOS operating-system software. Now it is a key piece of evidence in a private antitrust suit brought two years ago in federal court here by tiny Caldera Inc., of Orem, with the backing of Ray Noorda, the 72-year-old former chairman of Novell Inc., of Provo. The suit charges that Microsoft intended "to destroy competition in the software industry."
The e-mail is among previously secret internal documents subpoenaed by the U.S. government in a 1995 antitrust suit against Microsoft that was settled; these documents are at the center of the Caldera suit and could become part of the government's current antitrust suit against Microsoft, which is scheduled to go to trial next month. (In pursuing its suit against Microsoft, the U.S. government has subpoenaed documents from Intel Corp.)
Microsoft concedes the authenticity of Cole's e-mail, but its lawyers deny Caldera's antitrust allegation and have asked that the lawsuit be dismissed. Thomas Burt, a senior Microsoft attorney, said in an interview that something mentioned in an e-mail between managers doesn't make it company policy. "An e-mail message in a company like Microsoft is a conversation - it's where ideas are explored, and sometimes shot down," he said. "Because it is both informal and not conclusive, you can't really look at any single e-mail or excerpt out of context."
Burt says that the Salt Lake lawsuit is unrelated to the current U.S. case and, in any event, is a fight over products long since obsolete. But a federal judge in Utah disagreed earlier this year and instead broadened Caldera's case by expanding Microsoft's potential financial liability to include the potential damages caused by Windows 95. A trial is scheduled before a Salt Lake jury next June.
The hundreds of e-mail messages and documents collected by the government in the 1995 case, but never released to the public, could have a strong emotional impact on a jury. Some also could be introduced as evidence in the current federal case, which goes to trial next month in Washington, as investigators try to show a longstanding pattern of predatory actions by Microsoft against potential rivals.
Microsoft executives say they weren't concerned about DR-DOS, the rival program then owned by Novell, which had less than 5 percent of the market. But the subpoenaed Justice Department doc-u-ments suggest that top Microsoft executives were in fact worried and devised a campaign to win long-term, exclusive contracts with personal-computer makers "to freeze DR-DOS out," and "derail the train before it starts," according to an Aug. 7, 1991, memo by senior Microsoft marketing executive Brad Chase.
Among the memo's prescriptions: Build "unique synergies" between MS-DOS and Windows, and "lock in PC makers to long-term contracts." The memo shows concern that DR-DOS could take off, especially if International Business Machines Corp. adopts the program. A Microsoft spokesman said that restrictions on its contract lengths weren't imposed by the government until the settlement of the 1995 case.
DR-DOS was one of the few products presenting an alternative operating system to Microsoft. That posed a threat to Microsoft's most basic strategy: controlling the software plumbing of desktop computing, and leveraging that control to sell more PC software.
Microsoft dismissed DR-DOS as a knockoff, but the lineage of the two programs is more complex. The original code for Microsoft's DOS was itself a technical cousin of C/PM, an early operating system on which DR-DOS was also based. DR-DOS was a product of Digital Research Inc., a software company famous in high-tech lore for missing an early chance to supply IBM with the operating system for its first personal computers. That was the very contract that vaulted Microsoft to prominence. Novell later bought Digital Research and then sold it to Caldera.
In its suit, Caldera alleges that Microsoft tied Windows to its version of DOS and then forced PC makers to take both programs, shutting out DR-DOS.
This was accomplished through a tactic known as per-processor licensing, in which a computer maker must pay a set fee for every PC it makes, regardless of what operating system it installs, Caldera alleges.
The tactic was labeled illegal in a 1994 Justice Department complaint against Microsoft, as were the long-term sales contracts and bundling of one Microsoft product with another. After Microsoft settled that case in 1995, the e-mail and other documents collected by federal prosecutors were sealed.
Microsoft is likely to use the same argument in the Utah suit that it embraced in Washington - that it has an absolute right to determine its products' design. Governmental interference, the company says, stifles innovation in an industry vital to the U.S. economy. Microsoft is also likely to argue that consumers are better served by a single integrated Windows product; and that selling the products separately would have been inefficient for consumers.
Caldera, too, will argue it was defending consumers and innovation. The DR-DOS product was compatible, cost less and offered more advanced features. It goaded a complacent Microsoft, after years without a major upgrade to its flagship operating system, to add new features that matched those in DR-DOS. Those features included support of larger disk drives, improved memory management and other advancements, an edge that helped DR-DOS sales rise in 1990 and 1991 - before Microsoft struck back, Caldera says in its suit.
The lawsuit alleges that part of Microsoft's campaign against DR-DOS was to hide a bug inside a test version of Windows, which was sent in 1991 to thousands of developers working on software that would run on Windows and to nearly every PC maker in the world. Many of these companies were then considering whether to use DR-DOS, and the bug - triggered whenever it encountered DR-DOS - sowed doubt that it was compatible, the suit charges.
At the time, the bug was a mystery in the software industry. It was encrypted - the only piece of encrypted code in Windows - so that it was almost impossible to identify. But the author couldn't resist signing his work, leaving one string of letters unencrypted, and the bug became known among puzzled software engineers outside Microsoft as the AARD code-letters that Windows developer Aaron Reynolds used to tag his code.
Microsoft said Reynolds and others involved in the case wouldn't comment individually.
When a "foreign" operating system was detected, the bug froze the computer and displayed an ominous error message suggesting that the user contact Microsoft. When developers questioned the message at the time, Microsoft denied it was intended to derail DR-DOS. Documents to be produced at the Caldera trial suggest otherwise. "What the guy is supposed to do is feel uncomfortable when he has bugs, suspect the problem is DR-DOS and go out and buy MS-DOS and not take the risk," a Feb. 10, 1992, Microsoft memo said.
The bug was decoded by a self-appointed software sleuth, Andrew Schulman, who published his suspicions in a technical monthly with a cult following, Dr. Dobbs Journal, in 1993. He thought the code may have been placed there deliberately but wasn't sure, and unraveling it took some work. The code "turned out to be XOR-encrypted, self-modifying and deliberately obfuscated - all in an apparent attempt to thwart disassembly," he wrote.