Based on the briefs filed in the Microsoft v. AT&T case, the Supreme Court faces a substantial risk of establishing logically indefensible precedent relating to computer software. The status of software in patent law is a fair subject of debate, but such policy decisions should not be based on folk science.
In the briefs now before the Court, software-and information generally-is characterized as an ethereal construct, barely capable of being controlled or understood. In its amicus brief, Autodesk calls information "meta-physical." The Software & Information Industry Association describes information as "purely intangible." According to Microsoft, information is "abstract," and is "floating in the ether." Information has been so denigrated that AT&T, who holds the patent on the software method relevant to the case, claims that "software is not 'information.'"
Colloquially speaking, information is hardly a new concept. Perhaps this is why the briefs assume it is reasonable to use a dichotomy pulled from Aristotelian philosophy: optical disk as "substance" versus computer program as "accident." But like physics, chemistry, and other topics of interest to the ancients, information has become a science, and it should be discussed as a science.
One prerequisite of science is the ability to isolate an object of study from the complicating influence of the environment. The first quantitative understanding of physics came only after natural philosophers turned their attention to the unimpeded motions of planets and pendulums, and the modern era of chemistry, it is argued, began when a French lawyer isolated the process of combustion. In biology, evolution was discovered only after the study of an isolated island ecosystem. Just as physics, chemistry, and biology were once "proto-sciences," so too was it difficult until recent decades to speak of a science of information. Information had always been bound up with books and thoughts, where its quantitative study would have been unimaginable. Digital computers, though, have given us both the ability and the motivation to study information in a nearly isolated form.
AT&T's denial that software is information is strange, and even disappointing, because it was at AT&T's own Bell Labs that the scientific study of information originated. See Wiki. In the intervening half-century, the physical properties of information have been studied thoroughly. Lawyers and the courts cannot be expected to understand this young science in detail, but they should not assume that there is nothing to understand.
One of the most profound results of the scientific study of information is that information, like other physical phenomena, is governed by the laws of thermodynamics. In particular, information has the properties of the thermodynamic quantity known as "entropy." See Wiki. These physical laws impose real limitations on the storage and manipulation of information, without regard to whether the information is embodied in magnetic dots, microscopic pits, punched cards, or even radio waves. Many of these laws and limitations are summarized in Michael P. Frank, "The Physical Limits of Computation," Computing in Science & Engineering (May/June 2002) at 16-26, available here.
The science of information need not be studied in depth to appreciate how uninformed many of the arguments to the Supreme Court have been. A comparison between computer software and "design information" (such as CAD/CAM codes, blueprints, instructions, or recipes) is repeated among the briefs. The desired inference, spelled out by amicus Amazon, is that there is "no scientific basis" for distinguishing between them. This inference, though, is incorrect. In the study of information, as in thermodynamics, there is a fundamental distinction between actions that are "reversible" and actions that are "irreversible." An irreversible process is one in which information is lost. For example, overwriting data in memory, as is done continually in a computer system, is not reversible. Copying a computer program from a master disk onto a blank portion of a hard drive, though, is reversible: even if the master disk is lost, its information can be restored using the information from the hard drive, and the hard drive can be restored to its original blank state.
Contrary to Amazon's arguments, then, computer software can be scientifically distinguished from design information, because using design information to produce a product is irreversible. A manufactured product cannot be changed back into CAD/CAM codes and raw materials. A computer with Windows installed, though, can be changed back into a computer with a blank disk and an uninstalled copy of Windows.
The amicus law professors have argued that compressed or encrypted information is "useless gibberish." Again, the science of information provides a helpful insight: compressing and encrypting data are both reversible processes. No relevant information is lost, since the data can be decompressed and decrypted to its original form.
Microsoft provides a nearly Biblical description of its Windows software as "abstract information" that "lacks physical existence" but to which Microsoft has "given a physical manifestation" as a "golden master." See John 1:14 ("And the Word was made flesh... and we beheld his glory."). How the first manifestation of Windows was pulled from the ether, though, is irrelevant. The issue before the Court relates only to what happened after Windows was originally compiled, and it is clear that any copies made thereafter were the mundane result of a reversible copying process.
Thus, the distinction in Federal Circuit case law between software, which can be a "component," and design information, which cannot, is a valid distinction. It can be seen as a question of whether supplying the alleged component is reversible: information is a component if and only if it can be added and removed in a substantially reversible way. This parallels our understanding of mechanical components: most parts can be both added to and removed from a mechanical device. Of course, gray areas would remain (compiling source code into object code, for example, is not entirely reversible), but they remain in other technologies as well.
Information has other properties that allow it to be easily understood as a "component." Like mechanical components, information can be traced to its source (to determine who is liable), and it can be counted (to measure damages). Indeed, information can be far easier to trace to its source than anything made of molecules. A U.S. manufacturer could deny that it exported any physical part of an accused product: a connector made overseas may look exactly like one made domestically. Microsoft, though, could hardly deny that it was the original source of the Windows program: the probability that someone would arrange so many megabytes of information the same way as Microsoft, without copying, are astronomically small.
Normally, a single copy of a program is provided on each disk or on each computer. This makes it easy to count the instances and the location of a software "component" to determine damages: the number that are made in the United States, the number that are exported, and the number that are made abroad. Even in other, more complicated scenarios, information theory provides ways of characterizing the number of copies that exist (using, for example, measures of "redundancy" or of "mutual information").
Without proper guidance, the Supreme Court is likely to perpetuate proto-scientific characterizations of software. Information, such as the information used to store and transmit software, has well-understood physical properties. Because of those properties, software can be logically and consistently treated as a "component" under the patent laws. Whether those properties imply that software merits patent protection is open to debate, but neither side in that debate should disregard the scientific understanding of information in the hope of a victory based on ignorance.
Note: Jeff Steck is a patent litigator at McDonnell Boehnen Hulbert & Berghoff in Chicago. He studied physics at the University of Chicago.
