CNET también está disponible en español.

Ir a español

Don't show this again

Christmas Gift Guide
Tech Industry

The fear war against Linux

Open-source activist Bruce Perens uncovers the SCO-Microsoft connection behind a campaign to convince people that trade secrets of Unix have been copied into Linux.

Microsoft's connection to the anti-Linux campaign being waged by the SCO Group is becoming clear.

In the latest move, Microsoft has stepped up the battle with an announced agreement to license SCO's Unix patents and the source code, describing the deal as a reflection of its "ongoing commitment to respecting intellectual property and the IT community's healthy exchange of IP through licensing."

Nice rhetoric.

This comes after SCO just last week sent a letter to big IT customers, threatening legal action. And the reality is that Microsoft is tying SCO's allegations into its own anti-GPL campaign, a mostly unsuccessful effort to convince customers that the sharing and openness methods used in Linux development are unhealthy for the market. (SCO has gone so far as to publish "Quotations from Linux Leaders," a collection of inaccurate and out-of-context quotes of GNU Public License creator Richard Stallman and myself to paint Linux developers as nothing better than software pirates.)

Someone should tell SCO that IT customers don't like to be threatened by their vendors. In fact, the increasingly bellicose tone of SCO's communications and the refusal to show any evidence might well suggest that its claims are nothing more than grandiose ravings. To be sure, Microsoft will take advantage of those ravings while it can.

The real story here is the lack of substance to the SCO claims, and the increasingly remote chance that its lawyers will prevail. A similar case alleging plagiarism of Unix by an open-source operating system was litigated in the early '90s. AT&T sued the University of California, claiming that the BSD system infringed upon AT&T's copyrights. Eventually, the court narrowed AT&T's concerns down to only four source files, which the university simply replaced rather than argue about them. AT&T then settled the case by paying the university's court costs. SCO is not likely to do any better.

SCO's lawsuit against IBM is not a patent case. The fundamental patents on Unix would have expired long ago, while SCO's handful of patents aren't significant. The main allegation is that trade secrets of Unix have been copied into Linux. To win a trade secret case, you have to prove the information was secret. Detailed knowledge of Unix has been available in libraries for 30 years, and a full Unix specification was distributed by the U.S. government as part of its POSIX standards.

The Unix source code has been licensed to universities for the past 30 years, and most good college computer science departments made use of a copy until the publicly available Linux came along. While students were contractually restricted from committing plagiarism by directly copying code, they also were expected to use what they'd learned from the Unix internals during their entire careers. And computer science textbooks have documented every aspect of operating system internals for a long time. It is thus extremely unlikely that SCO can show that any of the claimed "trade secrets" were secret at all.

Someone should tell SCO that IT customers don't like to be threatened by their vendors.
Besides the older Unix secrets, SCO claims that IBM copied newly developed SCO trade secrets that got revealed through the two companies' partnership during the Monterey project. The use of secrets is now alleged to have enabled Linux to run efficiently on the Intel Itanium processor. In fact, the lead engineers on the Linux Itanium port weren't even at IBM; they are Hewlett-Packard's David Mosberger and Stephane Eranian. Their book, "IA-64 Linux Kernel," documents the work they performed. Itanium was derived from HP's PA-RISC architecture through an HP-Intel partnership, and thus it's much more likely that any Itanium rocket science in the Linux kernel has come from HP and Intel, rather than IBM and SCO.

There's also a problem with attributing too much interest in Itanium to IBM, which makes its own CPU chips in competition with Intel. And the very notion of IBM being the Robin Hood of operating systems--stealing from SCO to give to Linux--is difficult to believe. Of the companies that could be expected to have stringent policies and training about intellectual-property issues, IBM must head the list.

Similarly, the open-source community has been careful to create its own code rather than copy others. Richard Stallman's 1984 "GNU Manifesto" deals extensively with copyright issues and ownership of software, and later documents, like my "Open Source Definition," have continued that process. Our developers are smart enough to understand the consequences of plagiarism.

Where SCO stands
And yet SCO officers allege--without any substantiating evidence--that their copyrighted code has been appropriated into Linux. If there's been any copying, it's much more likely that the publicly available GNU/Linux code has been copied into the secret SCO source. To prove otherwise, SCO would have to present evidence regarding the date its code was written. The creation dates for Linux code aren't in question because CD-ROM archives exist of all stages of its development and have been sold to thousands of witnesses.

SCO's allegations can't be proven until its lawyers produce the evidence. But no publicly disclosed evidence is forthcoming until the trial, says SCO's Darl McBride, because it would give Linux developers a chance to launder their issues.

As if they could wipe the bits off of the hundreds of thousands of Linux source-code CDs already sold. SCO says it might show the evidence to "independent experts" under a nondisclosure agreement in a few weeks. But why is an NDA necessary? By SCO's own attestation, the code in question is already available to the public. The company would not be further damaged by a public display of evidence. The no-spin answer is that by delaying the public release of evidence, SCO can continue to make unsubstantiated assertions about Linux for as long as possible.

It's interesting to look at the size of the Linux development organization. Some 440 people are listed in the Linux kernel credits--that's only a partial list. They produce 50,000-plus lines of new or modified code per month. And that's just the kernel--other sizable teams produce libraries, utilities and applications.

SCO, a small and troubled company, can't hope to match those figures. Indeed, any Unix team in history would have been hard-pressed to rival them. Why, then, would Linux or IBM need SCO's contribution?

And what of SCO's own participation in Linux, and its effect upon its lawsuit and future income? SCO shut down distribution of its own Linux system, citing "intellectual property risk." So what? The company's behavior had already driven Linux customers elsewhere. And shutting down its Linux business does not change SCO's intellectual-property risk: The company had already distributed the Linux kernel and other critical components under the GPL license as part of SCO's own products.

Who really benefits from this mess? Microsoft.
SCO had more than adequate chance to notice if it owned any of the code in question. The GPL does not have a termination date and promises royalty-free use of the licensed code to everyone. The fact is, SCO's potential to collect royalties from the Linux kernel or anything else connected with the GPL is nil.

For SCO to have been distributing the very code it contends was appropriated--under a license that assures everyone of the right to use it for free--further hurts its chances of prevailing.

SCO management also fails to grasp its liability for the harm it is causing to countless customers, developers and software projects involved in Linux. The group's actions will lead to loss of sales and jobs, delayed projects, canceled financing, and the like. The damage to others will certainly invite retribution when the frivolity of SCO's claims is revealed.

I earlier thought the suit could be a bid to force IBM to acquire SCO at a cheaper price than the cool billion dollars being demanded to settle the case. Big Blue, which isn't taking the bait, must be confident of winning.

Who really benefits from this mess? Microsoft, whose involvement in getting a defeated Unix company to take on the missionary work of spreading FUD (fear, uncertainty and doubt) about Linux is finally coming to light.

Microsoft hardly needs an SCO source license. Its license payment to SCO is simply a good-looking way to pass along a bribe, coupled with an announcement designed to further intimidate Linux users. It's hard to imagine former Microsoft adversaries SCO and David Boies doing Bill Gates' bidding, but Microsoft's money is green. SCO stockholders should be asking questions.