Patently Absurd
click here to go back to the home page 
Part 3: The Universal Machine
Back to Part 1 / Part 2


With the advent of computing, human invention crossed a threshold into a world different from everything that came before. The computer is the universal machine, almost by definition, machine-of-all-trades, capable of accomplishing or simulating just about any task that can be logically defined. Even more so with cyberspace — everybody's computers, connected — the universal machine to the nth degree. Take any real-world procedure or activity; transfer it to cyberspace; and now we have an activity that seems both new and technological. Patents have recently been awarded for selling airplane tickets on line; selling software on line; "network sales systems" (many); "distributing advertising over a computer network," and countless more.
  But patents played no role in the early history of personal computer software. Microsoft, having been founded in 1975, received its first patent in 1986 — for a sort of hinged box. Another two years passed before it got its first computer-related patent. IBM and some other companies were patenting software, but they had to engage in a careful sophistry to do it: making sure to say that a program was an apparatus or a system and that it was embodied on a computer-readable medium. Applicants and examiners employed a mutually understood pretense that machinery was still involved.
   The biggest battles over intellectual property in the software business focused on copyright. Apple sued Microsoft in 1988 for copying the ''desktop metaphor'' and other key elements of Windows, and Lotus sued Borland in 1990 for copying menus and command sequences in its spreadsheet software. Both defendants had clearly copied something, but it was not the actual code; it was something vague, in the nature of "look and feel," and ultimately the courts ruled in favor of the defendants. For books, copyright law protects actual text, not ideas or plot; the law had to figure out what that meant for this new form of expression, software. Some software authors in the 1980's tried to extend their copyright protection beyond the literal code, to elements at higher levels of abstraction: a common phrase was "structure, sequence and organization." Ultimately the courts held that copyright law did not cover these broad elements. The computer industry itself was divided; so was the Supreme Court, which split 4-4 in refusing to overturn the Lotus v. Borland ruling.
   Patent officials like to suggest now that those cases might have ended differently if Apple and Lotus had patented their innovations. The decisions "tilted the system away from copyright," Dickinson says. "People shifted and started to use the patent system."
   As the software industry began flooding the office with applications, the system came around to a view of software programs as machines. They are machines, in a way. "Anything that can be done in hardware can be done in software," says one examiner. "We're not using gears; we're using some other mechanism, some other means." The patent office used to require an exact model of every invention it considered. Edison's original light bulb is sitting in Commissioner Dickinson's office, near a window with a fine view of Reagan National Airport. The models accumulated until the government ran out of storage space and abandoned the requirement. Lucky thing, because these new machines are machines without substance — incorporeal machines, machines made of imagination and logic. No one will be building models of these airy phantasms of bits.

One more crucial court decision came in July 1998: the State Street Bank appeal. State Street had been sued for infringing a 1993 patent for an intricate computerized strategy of managing a multi-tiered portfolio of mutual funds. A federal court in Massachusetts found that the real subject of the patent was not a "machine" but a business method in data-processing garb, and declared the patent invalid. Mathematicians and physicists sometimes argue about whether they invent or discover the laws of nature. Did Einstein invent his formula E=mc2 or was it there all along, waiting to be discovered? He couldn't have patented it, the Supreme Court has stated: such laws are "manifestations of nature, free to all men and reserved exclusively to none." Judge Patti B. Saris cited that principle in the lower-court decision, noting some other key precedents: "Mental processes, and abstract intellectual concepts are not patentable, as they are the basic tools of scientific and technological work"; and "an improved method of calculation, even when tied to a specific end use, is unpatentable subject matter."

Another real patent:
Method of Exercising a Cat

"A method for inducing cats to exercise consists of directing a beam of invisible light produced by a hand-held laser apparatus onto the floor or wall or other opaque surface in the vicinity of the cat, then moving the laser so as to cause the bright pattern of light to move in an irregular way fascinating to cats, and to any other animal with a chase instinct."

   Saris decided that the so-called invention was no more than a way of calculating: "The same functions could be performed, albeit less efficiently, by an accountant armed with pencil, paper, calculator, and a filing system." The patent gives its owner a monopoly, she pointed out: "patenting an accounting system necessary to carry on a certain type of business is tantamount to a patent on the business itself."
   But the Court of Appeals reversed that ruling in 1998 and reinstated the patent. Even "abstract ideas constituting disembodied concepts or truths" could be patented, it declared, as long as they performed a useful function. The useful function in this case was to produce "a final share price." In other words, the court ruled, software that merely manipulates numbers, juggling them and exchanging them and transforming them into other numbers, is producing something tangible. A powerful conclusion, in a digital age. It opened the floodgates.
   There is a sense of infinite regress to the argument, a feeling of abstraction upon abstraction, words like algorithm and formula and process and method defined and redefined circuitously, if not circularly. Judges and lawyers have devoted millions of words to the nuances. Maybe the trajectories of culture, economics and technology have reached a point where a distinction between idea and machine can no longer be sustained; where no bulwark of logic, but only the mist of undecidability, separates E=mc2 from the light bulb.

By the time of the State Street decision, Amazon's application had been working its way through the patent office for nearly a year. Barnes and Noble, late to the Internet business and lagging far behind Amazon, had redesigned its site yet again. Its new Express Lane looked to Bezos like a clone of his one-click system. Indeed, anyone examining the faces these sites present to the world might conclude that Barnes and Noble has copied a lot more than just one-click ordering. Both sites let customers post reviews of books; both display an average of the customer ratings in the form of one to five stars. They both list books that "customers who bought this book also bought." In October 1998 Amazon filed a "petition to make special," asserting that an "infringing product" was already on the market. Barnes and Noble didn't know that at the time; pending patent applications are kept secret.
   In Crystal City, the examiner, Demetra Smith, was studying a small selection of prior art — a few earlier patents and articles from the computer trade press — trying to make sure she understood the difference between Amazon's invention and earlier shopping-cart systems. The key to the one-click system is actually quite simple. If a customer has shopped at Amazon before, credit-card and other information is stored in the company's computers, and meanwhile the customer's Web browser has stored a special file called a cookie. That's how the browser knows, when the customer returns to Amazon's site, to display a personalized message: "Hello, Todd Dickinson! (If you're not Todd Dickinson, click here)." Cookies were already widespread in e-commerce, and the examiner had among her papers a 1996 description of how Netscape's software implemented them.
   As far as the examiner could see in the papers before her, no one had talked about using cookies to let customers skip past the whole shopping-cart checkout process, so in a sense Amazon's idea was, in fact, new. Still, she could see the basic elements in a patent that Lucent Technologies had filed just a few months before Amazon, describing "user identifiers" sent back and forth on a network. She noticed a comment that the identifiers help a company "to recognize a returning user and, possibly, provide personalized service." Just what Amazon does! The next step seemed obvious — in the patent office's technical sense of the word. "It would have been obvious to one having ordinary skill in the art at the time the invention was made," she ruled tentatively, "to include various command mechanisms for a single user action in order to execute the user's request."
   Obviousness is the key problem. "How do you tell, in a new field of the economy, what is obvious?" says Joseph Farrell, an economist at the University of California at Berkeley. "If you have a wide open field suddenly opening up out of nowhere, there may be a lot of things just waiting to be done which are pretty obvious, and there's no need to make them patentable to provide incentives for people to do them. And then it's kind of a pity if they get patented."
   Every patent examiner's work is about making fine distinctions. Is that "ergonomic topographic toothbrush" novel enough to merit a new patent of its own? These are delicate enough questions when the subject is real stuff. Patent examiners can sink their teeth into concrete details like a "detachable and replaceable bristle head." With software, where the nuts and bolts are vaporous and intangible, questions of what's obvious and what's novel begin to float in the wind. In any event, the burden of proof is on the examiner to show that an application must be rejected. Her subjective judgment is not allowed — only actual references in published literature. "People send in some really strange stuff for patents, and I have no choice but to issue it," says one examiner. "I can't say, gee, that's obvious to me." Evidence of obviousness "has to be out there and public and in the same detail."
   So Amazon's lawyers argued back, pointing out that the Lucent patent specified anonymous browsing and nowhere mentioned single-action ordering. They negotiated, agreeing to drop some claims in their application and amend others, and they followed the examiner's directions for reworking their drawings (margins incorrect, numbers too small, lines too irregular), and in the end, like most patent applications, this one was approved. After all, examiners are motivated to issue patents, not to hinder them. U.S. 5,960,411 grants Amazon exclusive rights to its method of placing an order "in response to only a single action." Also, ". . . wherein the single action is clicking a button." And "the single action is speaking of a sound." And wherein a user "does not need to explicitly identify themselves." And "the single action is selection using a television remote control" or "a pointing device" or "depressing a key on a key pad." Amazon has itself pretty well covered. Then again, hundreds of new e-commerce patents will be issued in the coming month, just about all to Amazon's largest and toughest potential competitors. Until they are issued, neither Amazon nor anyone else will know whether they are infringing — perhaps with an innocent and obvious bit of homegrown programming.
   Amazon claims that it spent thousands of programmer-hours on its one-click method, but the patent system doesn't care about that. In determining what may and may not be patented, the law does not distinguish between inventions that require expensive research and inventions that amount to a momentary flash of insight. A new drug that costs ten years and millions of dollars of research gets the same protection as a bit of programming that comes to a lone hacker in a dream. In purely economic terms, this is inequitable and perhaps even dangerous.
   "We're not talking about Thomas Edison inventing the light bulb," says Lessig. "We're not talking about Monsanto spending tons of money on some chemical whatever. We're talking about people taking ways of doing business and, because they put it into software, they say, 'This is now mine.'"
   Amazon won't say how many patents it has pending. The one-click patent isn't its first, as it happens; Jeff Bezos got one in February 1998 for "a method and system for securely indicating to a customer one or more credit card numbers that a merchant has on file for the customer when communicating with the customer over a non-secure network." The method is this: show the customer only the last few digits of each credit-card number.
   Many companies use this very method, of course. Even if they thought it up all by themselves, they may be infringing Bezos's patent. He and Amazon won't say whether they've begun trying to collect licensing revenue.

Part 4: 'We don't only patent rocket science'.


               Copyright 2000 James Gleick