7.3 AT&T Notices the Damage

Free For All. Go to the Table of Contents. Vist the Gifcom.


Soon after Network Release 2 hit the world, the real problems began for BSD. While AT&T didn't really notice 386BSD, NetBSD, or FreeBSD, they did notice a company called Berkeley Software Design Incorporated. This corporation created their own OS by taking Network Release 2 and adding their own versions of the missing six files, but they didn't release this for free on the Net. They started putting advertisements in the trade press offering the source code for $995, a price they claimed was a huge discount over AT&T's charge.


The modern, post-Internet reader should find this hilarious. Two to three groups and countless splinter factions were distributing the BSD software over the Internet for free and this didn't seem to catch AT&T's attention, but the emergence of BSDI selling the same product for almost $1,000 rang alarm bells. That was the time, though, before the Internet infrastructure became ubiquitous. In the early 1990s, people only halfjoked that FedEx was the most efficient Internet Service Provider around. It was much faster to copy hundreds of megabytes of data onto a magnetic tape and drop it in FedEx than to actually try to copy it over the Internet. Back then only real nerds were on the Internet. Managers and lawyers wore suits and got their news from the trade press and advertisements.


BSDI's cost-cutting was a major headache for AT&T. This small company was selling a product that AT&T felt it had shepherded, organized, and coordinated over time.


AT&T started off by claiming UNIX as a trademark and threatening BSDI with infringing upon it. BSDI countered by changing the ads to emphasize that BSDI was a separate company that wasn't related to AT&T or the subsidiary AT&T created to market UNIX known as UNIX System Laboratories, or USL.
That didn't work. USL saw its cash cow melting away and assumed folks would jump at the chance to buy a complete OS with all the source code for $995. The price seems outrageously high today, but that's only after the stiff price competition of the 1990s. It was still a good deal at the time. So USL sued BSDI for actually stealing proprietary source code from AT&T.


This argument didn't work, either. BSDI turned around and waved the Network Release 2 license they got from Berkeley. They bought all but six of the files from Berkeley, and Berkeley claimed that all of the source code was theirs to sell. BSDI wrote the missing six files themselves and they were quite sure that they got no help from AT&T or USL. Therefore, BSDI didn't steal anything. If AT&T thought it was stolen, they should take it up with Berkeley. The judge bought BSDI's argument and narrowed the case to focus on the six files.


This was a crucial moment in the development of the free software movement and its various kernels. AT&T found itself cornered. Backing down meant giving up its claim to UNIX and the wonderful stream of license fees that kept pouring in. Pressing ahead meant suing the University of California, its old friend, partner, and author of lots of UNIX code. Eventually, the forces of greed and omnipotent corporate power won out and AT&T's USL filed a lawsuit naming both BSDI and the University of California.


Taking sides in this case was pretty easy for most folks in the academic and free software world. The CSRG at Berkeley did research. They published things. University research was supposed to be open and freely distributed. AT&T was trying to steal the work of hundreds if not thousands of students, researchers, professors, and others. That wasn't fair.


In reality, AT&T did pay something for what they got. They sent their employees to Berkeley to get master's degrees, they shared the original Versions 5, 6, and 7 and 32/V source code, and they even sent some hardware to the computer science department. The original creators of UNIX lived and worked at Bell Labs drawing AT&T paychecks. Berkeley students got summer jobs at AT&T. There wasn't an official quid-pro-quo. It wasn't very well spelled out, but AT&T was paying something.


Some folks on AT&T's side might even want to paint the CSRG at Berkeley as filled with academic freeloaders who worked hard to weasel money out of the big corporations without considering the implications. The folks at Berkeley should have known that AT&T was going to want something for its contributions. There's no such thing as a free lunch.


There's something to this argument because running a high-rent research project at a top-notch school requires a fair amount of guile and marketing sophistication. By the 1990s, the top universities had become very good at making vague, unofficial promises with their pleas for corporate gifts. This sort of coquetry and teasing was bound to land someone in a fight. McKusick, for instance, says that the CSRG designed the BSD license to be very liberal to please the corporate donors. "Hewlett-Packard put in hundreds of thousands of dollars and they were doing so under the understanding that they were going to use the code," he said. If the BSD hadn't kept releasing code like Network Release 2 in a clear, easy-to-reuse legal form, he says, some of the funding for the group would have dried up.


But there's also a bit of irony here. McKusick points out that AT&T was far from the most generous company to support the CSRG. "In fact, we even had to pay for our license to UNIX," he says before adding, "although it was only ninety-nine dollars at the time."


AT&T's support of the department was hardly bountiful. The big checks weren't grants outright. They paid for the out-of-state tuition for AT&T employees who came to Berkeley to receive their master's degrees. While AT&T could have sent their employees elsewhere, there's no doubt that there are more generous ways to send money to researchers.


McKusick also notes that AT&T didn't even send along much hardware. The only hardware he remembers receiving from them were some 5620 terminals and a Datakit circuit-based switch that he says "was a huge headache that really did us very little good." Berkeley was on the forefront of developing the packet-based standards that would dominate the Internet. If anything, the older circuit-based switch convinced the Berkeley team that basing the Internet on the old phone system would be a major mistake.


To make matters worse, AT&T often wanted the BSD team to include features that would force all the BSD users to buy a newer, more expensive license from AT&T. In addition, license verification was never a quick or easy task. McKusick says, "We had a person whose fulltime job was to keep the AT&T licensing person happy."


In the end, he concludes, "They paid us next to nothing and got a huge windfall."
Choosing sides in this battle probably isn't worth the trouble at this point because Berkeley eventually won. The hard work of Bostic's hundreds of volunteers and the careful combing of the kernel by the CSRG paid off. AT&T's case slowly withered away as the University of California was able to show how much of the distribution came from innocent, non-AT&T sources.


Berkeley even landed a few good blows of its own. They found that AT&T had stripped copyrights from Berkeley code that they had imported to System V and had failed to provide due credit to Berkeley. The BSD license is probably one of the least restrictive ones in the world. Companies like Apple use BSD source code all the time. The license has few requirements beyond keeping the copyright notice intact and including some credit for the University of California. AT&T didn't pay attention to this and failed to cite Berkeley's contributions in their releases. Oops. The CSRG countersued claiming that AT&T had violated a license that may be one of the least restrictive in the world.


The battle raged in the courts for more than a year. It moved from federal to California state court. Judges held hearings, lawyers took depositions, clerks read briefs, judges heard arguments presented by briefs written by lawyers who had just held depositions. The burn rate of legal fees was probably larger than most Internet start-ups.


Any grown-up should take one look at this battle and understand just how the free software movement got so far. While the Berkeley folks were meeting with lawyers and worrying about whether the judges were going to choose the right side, Linus Torvalds was creating his own kernel. He started Linux on his own, and that made him a free man.


In the end, the University of California settled the lawsuit after the USL was sold to Novell, a company run by Ray Noorda. McKusick believes that Noorda's embrace of free competition made a big difference, and by January 1994 the legal fight was over. Berkeley celebrated by releasing a completely free and unencumbered 4.4BSD-Lite in June 1994.


The terms of the settlement were pretty minor. Net Release 2 came with about 18,000 files. 4.4BSD-Lite contained all but three of them. Seventy of them included a new, expanded copyright that gave some credit to AT&T and USL, but didn't constrain anyone's right to freely distribute them. McKusick, Bostic, and the hundreds of volunteers did a great job making sure that Net Release 2 was clean. In fact, two people familiar with the settlement talks say that Berkeley just deleted a few files to allow USL's lawyers to save face. We'll never know for sure because the details of the settlement are sealed. McKusick and the others can't talk about the details. That's another great example of how the legal system fails the American people and inadvertently gives the free software world another leg up. There's no information in the record to help historians or give future generations some hints on how to solve similar disputes.