On Wednesday, the Supreme Court heard oral arguments in Microsoft v. AT&T concerning liability for patent infringement for the distribution of software outside of the United States. Although the case is officially about arcane details of a 1984 revision to patent law, the decision could have broader significance if it helps to shed light on the patentability of software, a question the Supreme Court hasn't addressed in a quarter century.
AT&T sued Microsoft, claiming that its NetMeeting software violated a patent on audio codecs. Microsoft ultimately agreed to pay royalties for its domestic distribution of NetMeeting but refused to pay royalties for copies distributed overseas, arguing that software distributed overseas is not subject to American patent law. AT&T disagreed, citing an 1984 law prohibiting American manufacturers from supplying the components of a patented invention for assembly overeas. Microsoft countered that as an intangible string of ones and zeros, software as such could not be a "component" of a patented device. Rather, they argued, each individual copy of the software was a "component"; but these copies had been made overseas, they had not been "supplied" by Microsoft.
The United States Court of Appeals for the Federal Circuit sided with AT&T in 2005, and Microsoft appealed the case to the Supreme Court.
Is software a blueprint or a widget?
Although the case is ostensibly about the meaning of the words "supply" and "component," the fundamental issue is the status of software under patent law. The courts have held that software per se is not eligible for patent protection. Instead, patents are permitted on a machine that executes a particular kind of software. For the most part, this is a distinction without a difference, because software isn't good for anything if you're not allowed to run it on a computer.
But the distinction matters in this case, because all of the hardware in the machines sold overseas originated overseas. The only "component" that came from the United States was a golden master of Windows, which overseas OEMs duplicated and installed on each machine they sold. Microsoft urged the court to consider software to be like a blueprint: an abstract description of how to carry out a particular task. Clearly, an abstract description of how a machine works cannot be a "component" of that machine.
In his oral argument, AT&T's lead counsel, Seth Waxman, countered that software is nothing like a blueprint, because a copy of the software ends up on the hard drive of the final product and is essential to its operation. He noted that software instructions "move back and forth from the CPU and RAM—millions of times a second they are replicating themselves." Apparently oblivious to the existence of interpreted languages, he insisted that computers do not understand source code, and that a lot of testing and debugging is required to transform source code into object code. Hence, he reasoned, a blueprint is analogous to source code, whereas the object code is a functional "component" of the final device.
Several justices showed skepticism about this argument. Justice Breyer noted that American patent law specifically allows someone in America to transmit detailed instructions on how to construct a patented device overseas. Since software is fundamentally just information, he asked, why doesn't the same principle apply?
Nixing software patents?
The Software Freedom Law Center filed an amicus brief urging the court to take this line of reasoning even further. The courts have long held that laws of nature, abstract ideas, and mathematical algorithms are not eligible for patent protection. Software is nothing more than a description of a mathematical algorithm. Therefore, the center asks the courts to declare that software is not eligible for patent protection on that basis.
The brief, written by SFLC counsel Eben Moglen, argues that the Court of Appeals for the Federal Circuit, which has had jurisdiction over all patent cases since it was created in the 1980s, has gone off the rails in recent years. Moglen claims that the federal circuit has misread the Supreme Court's precedents on the patentability of software from the 1970s and early 1980s in a way that removes almost any limit on the scope of patentable subject matter.
During oral argument, several justices seemed sympathetic to this argument. Although none of them mentioned the SFLC brief specifically, Justice Breyer and Justice Stevens both asked Daryl Joseffer, who was arguing for Microsoft on behalf of the Bush administration, whether software is patentable. And Justice Scalia asked Theodore Olson, Microsoft's lawyer, "You can't patent on-off, on-off code in the abstract, can you?" Both Mr. Olson and Mr. Joseffer agreed that you could not do so.
But even if some of the justices are uncomfortable with the idea of patents on software, it seems unlikely that the court will vote to strike down software patents entirely. The legal status of software patents was not the focus of the arguments in the case, and the justices will be reluctant to issue a sweeping ruling on the patentability of software without giving the issue a thorough hearing. Moreover, companies now earn billions of dollars from their software patent portfolios, and the court will be reluctant to reverse legal principles on which companies have relied for a decade or more, even if those principles have never been endorsed by the Supreme Court.
This isn't the only patent case the Supreme Court is considering this term. The court heard oral arguments in the case of KSR v. Teleflex last November. That case centers on the requirement that patents cannot be granted on inventions that are "obvious." Although the case concerns an automobile throttle, not software, EFF filed a brief in that case urging the court to beef up the obviousness requirement as a way to rein in the excesses of software patenting. In November's oral arguments, the justices showed considerable skepticism about the obviousness test the Federal Circuit had crafted; Justice Scalia went so far as calling it "irrational" and "gobbledygook."
It seems likely that the Supreme Court will use at least one of these two cases as an opportunity to rein in the Federal Circuit, which has dramatically expanded the scope of patents in the last two decades. The court may decide that beefing up the obviousness standard with Teleflex may be an indirect and less drastic method of dealing with the problems of software patents than striking them down outright.
In any event, AT&T v. Microsoft is likely to be a landmark. The last time the Supreme Court considered a case related to software patents, IBM was just unveiling the first PC. So even if the Supreme Court does not squarely address whether software is patentable, it will provide important guidance to lower courts struggling with the difficult issues raised by software patents.
reader comments