Patent (Reform) Pending

By Larry Downes  |  Posted 10-05-2005

"Everybody talks about the weather," Mark Twain said, "but nobody does anything about it."

Judging from recent headlines, blogs and lawsuits, Twain's observation could apply equally to the patent system, especially for information technology. Over the past 20 years, there has been an explosion in the number of applications to the patent office, creating multiyear backlogs and the awarding of wildly inappropriate patents, which must then be overturned through long and expensive litigation.

Meanwhile, leading companies in a variety of industries increasingly pool or cross-license their collections of dubious patents, often secretly, effectively shutting out new competitors and inflating prices. And almost no one seems to have a kind word to say about patents for software, which have been recognized since the mid-1980s, or for "business method" patents, which the U.S. Patent and Trademark Office has been awarding since 1998.

Those complaining include not only legal scholars and free-culture junkies, but also some of the principal beneficiaries of the current system, including senior executives at Amazon, Microsoft, IBM, Google and Cisco Systems. Even the Federal Trade Commission and the U.S. patent office itself have called for a major overhaul of the system.

Good luck. This summer, the European Parliament abandoned an excruciating four-year-long debate on whether and how to grant computer-related patents. At almost the same time, Congress began deliberations on the Patent Reform Act of 2005, which, if passed, would make significant changes both to the process by which patents are awarded and to how courts subsequently deal with challenges to those awards. But even if the bill does eventually become law, no one imagines it will do anything but slow the system's accelerating race to disaster.

To paraphrase a famous Chicago politician: The patent system ain't ready for reform yet.

To understand why, we need to take a step back and answer some fundamental questions about the system. At its core, a patent is a bundle of property rights awarded to an inventor for a device (the cotton gin) or a process (vulcanizing rubber). To receive a patent, the inventor must apply to the patent office, where specialists examine the application to ensure that, among other things, the invention is actually useful, is not obvious, and has not, in fact, already been invented by someone else.

If these requirements are met, the inventor is given 20 years of exclusive control over the invention, which she may sell, license or keep to herself. Once the patent expires, however, the invention becomes part of the public domain. Indeed, at the time of application, the inventor must provide sufficient documentation to enable any competent practitioner in the field to reproduce the invention, royalty-free, at the end of the patent period.

Why are patents given at all? As the U.S. Constitution puts it, the purpose of a patent is

"To promote the Progress of Science and useful Arts." The goal, in other words, is to get as many useful inventions into free use as quickly as possible—where they become, to use the modern phrase, "open source"—and thus enrich society as a whole. The limited monopoly is presumed to provide an important incentive for inventors, since invention can be a long, expensive process, and recovering those costs and making a profit might be difficult if others could reverse-engineer your work and duplicate it without penalty. So the irony (and the beauty) of the system is that in order to make it free, we start by doing the opposite—creating a monopoly.

But the pace of technological improvement and the nature of the global economy have strained this simple system to the breaking point. For information technology, the problems are especially acute, and in some sense stem from the very thing that makes computers so useful. In essence, a computer is a general-purpose machine (an "analytical engine," as the mathematician Charles Babbage envisioned it in the early 19th century) that can do anything it is programmed to do. The combination of (general) hardware and (specific) software turns the computer into a virtual machine. Since their invention, computers have been programmed to perform thousands of tasks, many of them useful, nonobvious and novel.

But which virtual machines deserve patent protection and which do not? As a starting point, courts rejected the idea that software could ever be patented, on the theory that software merely encoded a set of "mental steps" taken by the programmer. As IT has become more complex and more central to the global economy, however, the federal courts and the U.S. patent office began carving out exceptions that eventually—and haphazardly—swallowed the rule. Today the patent office is flooded with applications for patents for software and for business methods that may be implemented in software (e.g., Amazon's patent for "1-click shopping").

The patent office, understaffed and largely unskilled in evaluating information technology, can't seem to distinguish the innovative combinations of IT from the obvious or nonnovel applications, and so has left to the market (that is, to lawyers and litigation) the problem of sorting out the "d'uhs" from the breakthroughs. One 2001 study found that almost half of all granted patents that were litigated to a decisive conclusion were held to be invalid—that is, should never have been awarded—and one suspects the numbers for software and business-method patents would be even higher.

For CIOs, the patent crisis in IT leads to increased uncertainty and cost, and the need to rely on legal experts for even the most basic vendor agreements, software licenses and consulting contracts. Microsoft spends approximately $100 million a year defending itself from patent suits, and last year Dell was sued by a company holding a patent that, it claims, covers all international business handled by computers. Even open-source software licensees may be subjected to patent litigation by a programmer who claims to have previously invented patentable features of the "free" code. Does the software you write infringe on an existing patent? How about your suppliers' software?

I am not one of those who believe there should be no patent protection for software and business methods, in large part because I believe that the alternate sources of protection—trade secrets and copyright—are even less appropriate given the unique structure of IT. (More on that in a later column.) And despite the economic success of many IT companies, IT innovation does and will continue to need the kind of incentives the patent process provides. Not to make a few individuals or corporations exceedingly rich, but to serve the actual goal of intellectual property protection—that is, to get as many useful new inventions into the public domain as efficiently as possible.

Of course no one believes that's what is happening now. But until legislatures in the U.S. and abroad take over the job of patent reform from the courts, and face head-on the special nature of IT, the complaints will just get louder. And the costs—for inventors, for IT giants, and most of all for innocent users—will soar ever higher.

Fortunately, IT generates so much excess social value that the waste associated with the current patent mess is more than outweighed by the benefits. So far, anyway.

LARRY DOWNES is Associate Dean of the School of Information Management and Systems at the University of California-Berkeley. He is the author of Unleashing the Killer App and The Strategy Machine. His next column will appear in January.