A patent is a form of regulation. It is a government-granted monopoly - an exclusive right backed by the power of the state. This monopoly is granted by a bureaucrat - a well-meaning, hardworking bureaucrat no doubt, but a bureaucrat nonetheless. This government employee decides whether an idea is novel, useful and nonobvious. If it is, the government guarantees the inventor an exclusive right to the idea for 20 years. Last year, some 150,000 such exclusive rights were granted, up one-third from the year before.
No doubt we are better off with a patent system than without one. Lots of research and invention wouldn't occur without the government's protection. But just because some protection is good, more isn't necessarily better. Especially in cyberspace.
There is growing skepticism among academics about whether such state-imposed monopolies help a rapidly evolving market such as the Internet. What is "novel," "nonobvious" or "useful" is hard enough to know in a relatively stable field. In a transforming market, it's nearly impossible for anyone - let alone an underpaid worker in the U.S. Department of Commerce who spends on average of eight hours evaluating the prior art in a patent and gets paid based on how many he processes - to identify what's "novel." Costly mistakes get made. On average it takes $1.2 million to challenge the validity of a patent, which means it is often cheaper simply to pay the royalties than to establish that the patent isn't deserved.
"Bad patents" thus become the space debris of cyberspace. Nowhere is this clearer than in the context of business-method patents. At a recent conference in Israel, I watched as a lawyer terrified the assembled crowd of Internet startups with stories of the increasing number of business-method patents that now haunt Internet space. Patent No. 5,715,314, for example, gives the holder a monopoly over "network-based sales systems" - we call that e-commerce. Patent No. 5,797,127 forms the basis for Priceline.com and effectively blocks any competitor. Patent No. 4,949,257 covers the purchase of software over a network.
To West Coast coders, it seems bizarre that East Coast coders - the Patent Office - consider these ideas nonobvious. But the real problem is the incentives such a system creates. Awarding patents of that type siphons off resources from technologists to lawyers - from people making real products to people applying for regulatory privilege and protection. An increasingly significant cost of Net startups involves both defensive and offensive lawyering - making sure you don't "steal" someone else's "idea" and quickly claiming as yours every "idea" you can describe in a patent application.
But this is absurd. When the world was given TCP/IP and the collection of protocols it induced, a billion ideas became obvious to anyone who took the time to think. These were not ideas that were discovered because some lone inventor spent years toiling away in his basement, but because TCP/IP was a language with which practically anything could be done. And with very little promise of protection by government, lots was done. The Internet revolution was born long before lawyers arrived on the scene.
The question economists are now asking is whether expanded patent protection will do any good. Certainly it will make some people very rich, but that's different from improving a market. The questions are many, while good answers are few. Does it make any sense in the context of code to protect an idea for 20 years, let alone the 95 years that copyright law gives code? Berkeley economist Joseph Farrell has floated the idea, although he has not endorsed it, of a moratorium. Berkeley lawyer Robert Merges has proposed badly needed changes to the patent system, to force better disclosure to competitors so that the government can determine which ideas really merit protection.
These ideas are good, but they underline a more fundamental problem. Washington is obsessed with intellectual-property rights. It lives under the mistaken idea that stronger IP always means
