The Caldera v. Microsoft Dossierby Andrew Schulman
On January 9, 2000, Microsoft and Caldera announced that they had reached a settlement after a four-year antitrust legal battle. As a consulting expert to Caldera during its lawsuit against Microsoft, Andrew Schulman worked on some of the technical details of the case, which concerned Microsoft's use of Windows to protect MS-DOS while undermining DR DOS. However, because the case was recently settled out of court, valuable court documents could be sealed. Fortunately, there's still a valuable dossier online that offers insight into Caldera v. Microsoft, and the judge has ruled to unseal additional documents. The public can still inspect this case, and learn from this example how technical decisions are influenced by business and political interests. Schulman talks about what you'll find in this dossier, encouraging you to look deeper into matters that are still in dispute today.
It's understandable if you didn't hear about it, what with everything else that was happening the second week in January. Besides the AOL/Time Warner merger announcement, Steve Ballmer's transfer to CEO of Microsoft, and the news that the US Department of Justice (DOJ) would seek to break up Microsoft, another piece of news didn't receive much notice: the eleventh-hour settlement of the Caldera v. Microsoft antitrust suit, for an amount which remains confidential, but which the Wall St. Journal (Jan. 11) estimated at around $275 million. At the same time, Caldera Systems filed for an IPO (see The Register, Jan. 12).
The Background of Caldera v. Microsoft
In 1996, Caldera, whose focus is now Linux, had purchased the DR DOS operating system (whose principal value was, in retrospect, this lawsuit) from Novell for $400,000. According to Microsoft, in buying the lawsuit Caldera was playing "litigation lottery." But in settling the Caldera case out of court, even after having won the (very different) Microsoft v. Bristol contract/antitrust jury trial in July 1999, Microsoft was perhaps finally acknowledging that Caldera had a solid case. There was some speculation that Microsoft would settle with Caldera as a way to buy its way into the Linux market, but this does not seem to have happened.
Much of the case concerned ways in which Microsoft allegedly leveraged its Windows monopoly to crush competition to its MS-DOS "cash cow." DR DOS tried to compete with Microsoft's own MS-DOS; at one point, it had nearly 10% market share, and, in some markets outside the US, it had captured major accounts (such as Vobis in Germany).
Caldera filed in July 1996, alleging that Microsoft had engaged in anticompetitive acts against DR DOS, including intentionally misleading product pre-announcements, vaporware and FUD ("fear, uncertainty, and doubt") announcements, exclusionary licensing, beta-test blacklists, building deliberate incompatibilities into Windows to hinder it from running with DR DOS, and trying to create the misperception that DR DOS couldn't work properly with Windows.
At first, Caldera's case appeared to be limited to events that were so old that they pushed up against the statute of limitations. In the software industry, of course, anything that happened more than three months ago gets the "history is bunk" treatment. However, MS-DOS was the foundation of Microsoft's fortune, and you can't understand the role of Microsoft in the software industry without closely reading DOS's history.
|In early 1998 Caldera amended its complaint to include a fresh allegation that Windows 95 was part of Microsoft's ongoing employment of Windows to eliminate competition in the DOS market.|
Besides, Caldera v. Microsoft turned out to involve even current versions of Windows. In its initial complaint, Caldera had strangely treated Windows 95 as irrelevant to the case, stating that Windows 95, like Windows NT, does not "run on top of" DOS. However, in early 1998 Caldera amended its complaint to include a fresh allegation that Windows 95 was part of Microsoft's ongoing employment of Windows to eliminate competition in the DOS market. According to Caldera's then-president, the ever-quotable Bryan Sparks, "Windows 95 is a walking antitrust violation because it is nothing more than Windows 4.0 ... and MS-DOS 7" (NC World, February 1998).
Microsoft attempted to have this and other claims thrown out of court by filing a series of nine "summary judgment" motions in February 1999. Caldera responded to these motions in May 1999, and a comparison of Caldera's detailed responses on the one hand, with its somewhat sketchy initial complaints on the other, shows that Microsoft did Caldera (and the public) an enormous service by filing the summary judgment motions: Caldera had to respond with richly-detailed motions of its own.
After reviewing Microsoft's summary judgment motions and Caldera's responses, Judge Dee Benson denied Microsoft's motions, ruling that Caldera had presented sufficient evidence to present its claims before a jury -- including the claims involving disputes over technology such as the use of deliberate incompatibilities in Windows 3.1, and "technological tying" or "integration" in Windows 95. The trial was scheduled to begin on February 1, 2000, in Salt Lake City (Caldera is based in Utah, near Brigham Young University; Microsoft had unsuccessfully sought a change of venue).
|Microsoft did Caldera (and the public) an enormous service by filing the summary judgment motions: Caldera had to respond with richly-detailed motions of its own.|
The Case Stays Open, Sort Of
To the extent that the Caldera v. Microsoft confidential settlement was noticed, it tended to be viewed as at best a washout, and at worst a betrayal by Caldera. While Caldera explained that it had decided that the DOJ was "in a better position to pursue a public remedy" (Sm@rt Reseller, Jan. 10), the fact remains that there were high hopes for another nice, long trial in which Caldera would air Microsoft's dirty laundry in public.
For example, Dan Gillmor complained in the San Jose Mercury News (Jan. 16) that "Part of the settlement is an agreement to keep the terms and evidence secret. So the public will never get the chance to see just how badly Microsoft did -- or didn't -- behave a decade ago when it was in the middle of its rise to power."
In general, confidential settlements do have just this sort of result. A book titled No Contest, by Ralph Nader and Wesley J. Smith (1996), has a lengthy chapter describing the problems caused by confidential settlements (and also by "vacated" trial results, such as the 1994 patent and trade-secret infringement case of Stac Electronics v. Microsoft, in which, following a $120 million jury award to Stac, the parties instead decided to have Microsoft purchase 15% of Stac). Nader and Smith make a convincing case that such confidential settlements typically allow "evidence of corporate wrongdoing, evidence relevant to public policy debates and consumer interests, to remain hidden" (p. 60).
However, the settlement of Caldera v. Microsoft isn't quite like this. Months before the settlement, Caldera had already publicized a large collection of Microsoft "smoking gun" internal documents. Between them, Caldera and Microsoft had already placed about six megabytes of court documents on the web, from Caldera's side at http://www.drdos.com, and from Microsoft's side at http://www.microsoft.com/presspass/caldera (although some of the most important Microsoft summary judgment motions are heavily "redacted").
|Statement of Facts||Caldera statement (PDF)||Microsoft response|
|Technological Tying (Windows 95)||Microsoft motion||Caldera response (PDF)|
|Incompatibilities and Predisclosure||Microsoft on intentional incompatibilities,
Microsoft on perceived incompatibilities,
Microsoft on predisclosure
|Caldera response (PDF)|
|Product Disparagement||Microsoft motion||Caldera response (PDF)|
|Product Preannouncement (Vaporware)||Microsoft motion||Caldera response (PDF)|
|Licensing Practices||Microsoft motion||Caldera response (PDF)|
|European and Japanese Claims||Microsoft motion||Caldera response (PDF)|
|Benson ruling (June 28, 1999) on preannouncement, disparagement, and licensing|
|Benson ruling (November 2, 1999) on intentional and perceived incompatibilities, predisclosure, and technological tying|
|Confidential settlements typically allow evidence of corporate wrongdoing to remain hidden. However, the settlement of Caldera v. Microsoft isn't quite like this.|
Nor have we perhaps seen the last of the documents in the case. The Salt Lake Tribune and the San Jose Mercury News filed a motion last year, asking the court to unseal documents (Salt Lake Tribune, March 20, 1999). The newspapers argued that the First Amendment to the US Constitution "applies to the public's right of access to judicial documents," and cited recent precedents such as the Timothy McVeigh trial and the ongoing tobacco litigation.
Microsoft filed a response. Just before the settlement, the court denied Microsoft's motion and ruled that parts of the court record must be unsealed. Even with the case's dismissal, Magistrate Ronald Boyce retains jurisdiction for purposes of enforcing a protective order, and for seeing to it that Microsoft continues to unseal documents, as appropriate. (For more on the unsealed documents, some of which allegedly point to document shredding by Microsoft officials, see a somewhat odd article by Graham Lea, "Unsealed Caldera files detail MS evidence shredding claims", in The Register ["Biting the hand that feeds IT"], a useful online publication from the UK.)
Tying DOS in Windows 95
Apart from the documents released in Caldera v. Microsoft, there's another public result: Judge Benson, in denying Microsoft's summary judgment motions, issued an important decision -- and this decision is not "vacated" by the settlement. (In fact, Lawrence Lessig refers to Caldera, Inc. v. Microsoft Corp., 72 F. Supp.2d 1295 (D. Utah 1999) in his recent brief on software tying in US v. Microsoft.)
Much of the court's decision concerned Caldera's claim of "technological tying" in Windows 95. According to Caldera, Microsoft chose to produce Windows 95, rather than separate Windows and MS-DOS products as before, so that every Windows 4.0 user would in effect be required to buy MS-DOS 7.0, rather than some competing DOS such as DR DOS.
This Windows 95 claim was one reason for my involvement as a consulting expert to Caldera. My book Unauthorized Windows 95 devoted several hundred pages to an overkill demonstration that Windows 95 was actually a bundle of Windows 4.0 and MS-DOS 7.0, even though Microsoft in its lengthy preannouncement campaign had proclaimed, for example, that Windows 95 "no longer can be likened to a fancy paint job on an old MS-DOS Yugo" and would be "freshly designed from the ground up" (Microsoft Systems Journal, February 1994).
My book did not, however, really speculate why Microsoft would go to such lengths to hide the presence of MS-DOS, all the while continuing to rely on it. Even close inspection of code usually can't tell you much about its intent; understanding intent usually will require access to internal documents such as emails -- and of course I didn't have access to Microsoft's internal emails when I was writing my book.
|In digging through emails produced by Microsoft, Caldera seems to have found the reason for the elaborate Windows 95 "DOS is dead" charade: disguising the presence of MS-DOS inside Windows 95 would allow Microsoft to continue selling MS-DOS, while killing off any competition.|
During its "discovery" phase, a legal case can open up such documents. In digging through emails produced by Microsoft, Caldera seems to have found the reason for the elaborate Windows 95 "DOS is dead" charade: disguising the presence of MS-DOS inside Windows 95 would allow Microsoft to continue selling MS-DOS, while killing off any competition. DOS in fact is not dead: every time someone runs Windows 95, they are also running version 7 of MS-DOS.
Discussing the project codenamed "Chicago," which would years later become Windows 95, a June 1992 internal "Chicago Strategy Document" stated:
"Novell is after the desktop.... This is perhaps our biggest threat. We must respond in a strong way by making Chicago a complete Windows operating system, from boot-up to shut-down. There will be no place or need on a Chicago machine for DR-DOS (or any DOS)."
A December 1993 email by then Microsoft vice president Brad Silverberg confirms that the presence of MS-DOS within Windows 95 would be reflected in its price:
"[A reseller] asked about chicago pricing. i said that since it's the combination of windows and msdos and more, you should expect a price in the same ballpark as windows + ms-dos"
This was an internal communication, though, and Microsoft continues to deny that every copy of Windows 95 includes a copy of MS-DOS, for which the consumer has paid (usually in the form of the so-called "Microsoft tax" included in the price of every PC), without having had the opportunity to purchase an alternate DOS with which to run Windows 4.0. In its summary judgment motion on technological tying, Microsoft asked Judge Benson to toss out this part of Caldera's case, arguing that even a single plausible explanation for the "integrated" design of Windows 95 would automatically entitle Microsoft to win on this point.
After also hearing Caldera's response, Benson denied Microsoft's motion, ruling that there was a legitimate question here. The question was posed graphically in two figures included in the ruling:
"the Court must determine whether the present state of the record support[s] the proposition that Windows 95 consists of Microsoft's prior products simply upgraded and packaged together as illustrated below in Figure 1, or rather whether legitimate technological improvements were achieved, thereby technologically integrating the prior products into an entirely new product as illustrated below in Figure 2."
A vendor such as Microsoft cannot defend against a charge of tying simply by putting forward some "facially plausible benefit" of its ostensible "integration." The court in Caldera v. Microsoft agreed with dissenting opinion in US v. Microsoft of Judge Wald, who stated that the "facially plausible benefit" standard allows a defendant such as Microsoft "too safe a harbor with too easily navigable an entrance."
The Evidence Is In
What of the material available on the web? One document is Caldera's lengthy "statement of facts", which presents many "smoking gun" quotations such as the Microsoft internal emails regarding Windows 95. Caldera's statement of facts also has some priceless quotations from emails by Bill Gates:
"DOS being fairly cloned has had a dramatic impact on our pricing for DOS. I wonder if we would have it around 30-40% higher if it wasn't cloned. I bet we would!" (August 6, 1989)
"I doubt they [Digital Research] will be able to clone Windows. It is very difficult to do technically, we have made it a moving target and we have some visual copyright and patent protection. I believe people underestimate the impact DR-DOS has had on us in terms of pricing." (May 18, 1989)
Apart from the image created of Microsoft as a price gouger troubled by even the smallest amount of competition, we also see Microsoft making Windows a "moving target" to keep it from being cloned. (This seems a better explanation than mere "feature creep" for the increasing size and complexity of Windows over the years.)
Seeing the impact of even DR DOS's small market share on pricing for MS-DOS -- and probably also seeing how successful cloning of the IBM PC BIOS by companies such as Phoenix had turned the BIOS into a commodity -- Gates apparently wanted to make sure there was no competition at all for Windows.
It is amazing that, to this day, something as ubiquitous and profitable as the Windows API has only one supplier. This is leaving aside partial efforts such as WINE and WABI on the one hand, and vendors such as Mainsoft and Bristol on the other, who license code directly from Microsoft itself under the WISE program. Incidentally, following the failure of its lawsuit, Bristol has announced that it has signed a new WISE agreement with Microsoft for Windows 2000).
|The documents released in Caldera v. Microsoft help explain the lack of AMDs and Cyrixes for the Win32 API.|
Sometimes the lack of "open source" for Windows is put forward as the explanation for Microsoft's Win32 monopoly, but the "moving target" problem (or, the "treadmill" or "churn" as other Microsoft emails put it) strikes me as a more important way that Microsoft has managed to keep its ubiquitous platform from becoming the commodity BIOS it ought to be by now. The x86 instruction set does not have a single supplier: besides Intel, there's AMD and Cyrix; the documents released in Caldera v. Microsoft help explain the lack of AMDs and Cyrixes for the Win32 API.
The documents quoted in Caldera's statement of facts should not be viewed simply as "smoking guns" (though some of them certainly are that). More important, they provide a rare inside look at the politics behind seemingly technical decisions. This flavor has been missing from most supposedly "inside" views of Microsoft, such as the books Microsoft Secrets by Michael Cusamano and Richard Selby, and The Microsoft Way by Randall Stross. Rather than rely on press reports and interviews with officials, as these books seem to, a serious historian of Microsoft will have to mine the material found in company emails.
Caldera's statement of facts characterizes the Microsoft internal emails by quoting from a 1916 decision by Judge Learned Hand regarding typewritten memoranda by company executives, introduced as evidence in a monopolization case: "The documents were never intended to meet the eyes of anyone but the officers themselves, and were, as it were, cinematographic photographs of their purposes at the time they were written." The widespread use of email means that such a record will play an increasing role, not only in litigation, but in any genuine (as opposed to hagiographic) corporate history.
Microsoft dismisses Caldera's quotations from its internal emails as "pulp fiction," arguing that the quotations have been taken out of context, and that the proposals floated in these emails were just that -- proposals that were never acted on. This, judging from Nader and Smith's book No Contest, is a rather typical reaction to a "smoking gun" document.
Thus, Caldera also had to show some concrete results. Given Microsoft emails talking about putting code into Windows that would detect DR DOS and display an error message, for example, Caldera had to show that such "if detect DR DOS, then display error message" code actually existed. For example, given the email from Brad Silverberg stating, "We need to create the reputation for problems and incompatibilities to undermine confidence to drdos6; so people will make judgments against it without knowing details or facts," Caldera would need to show that Microsoft had in fact gone ahead and done something to bring about this result. A mere internal email, no matter how inflammatory or anticompetitive in its intent, could not have harmed DR DOS without some action in the real world.
This was another reason for my involvement in Caldera v. Microsoft. A book I coauthored and edited, the second edition of Undocumented DOS (1994), had discussed some odd code -- somewhat-widely known as the "AARD code" for the programmer's initials embedded in the code -- that Microsoft placed in widespread beta copies of Windows 3.1. I had also written over the years about other incompatibilities between Windows and DR DOS, and Caldera was now alleging that these were deliberate incompatibilities.
I was part of a small team, including Geoff Chappell and Caldera's expert witness, Dr. Lee Hollaar, who worked on matching up some of Microsoft's emails against the actual code in its products. While Hollaar had access to source code produced by Microsoft in the case, Chappell and I worked from Microsoft's binary code, including Windows 3.1 and Windows 95. (I'd also like to thank Chappell for his valuable assistance with this article.)
Some of our findings are reflected in Caldera's response on intentional and perceived incompatibilities. Besides the AARD code, we researched:
- a fascinating (and serious) incompatibility between Windows 3.1 SETUP and DR DOS that had to do with a failure to clear a single bit: the processor's "Nested Task" flag;
- an incompatibility based on Windows 3.1 SETUP's gratuitous requirement that an XMS server report an internal revision number of 2.6 or higher;
- an explicit test for (and refusal to run on) DR DOS in early betas of SmartDrive version 4 ("Bambi");
- gratuitous tests for Korean MS-DOS, performed by Korean versions of Windows;
We also studied the ways in which the Windows 4.x components of Windows 95 and Windows 98 are tied to the MS-DOS 7.x components, while engineers at Caldera were working on a demo program called "WinBolt" (see Seattle Weekly, Sept. 17, 1998) that allowed the Windows 4.x components to run on DR DOS instead of MS-DOS.
As the computer industry grows in importance, its disputes (even seemingly narrow technical disputes) will increasingly be of public concern, and will inevitably be aired in court. Seemingly technical decisions -- often these are political or business decisions that merely happen to employ technology -- will be reviewed by judges and juries. While this may sometimes result in surprising verdicts, such oversight is generally a good thing -- especially when one of the parties has been found to be a monopoly. And the evidence in these cases will necessarily involve, not only poring through internal emails, but also inspecting code.
Microsoft has recently published a book by Steve McConnell titled After the Gold Rush: Creating a True Profession of Software Engineering that pushes for professionalization of software engineering. Well, litigation of disputes is one of the things that comes with professionalization. In exchange for using the courts to resolve its disputes, the software industry will end up giving the public something in exchange: greater "inside" access, meaning not only a closer inside view of the products themselves, but also a more intimate view of decision-making within the company.
Andrew Schulman (http://www.undoc.com) was a consulting expert to Caldera in its case against Microsoft. He was editor and coauthor of the books Undocumented DOS and Undocumented Windows, and wrote Unauthorized Windows 95.