Patents for business methods and software are under attack by a variety of critics. Some claim that many of the innovations patented fail to meet the statutory tests of novelty and nonobviousness. Others, including Stanford University law professor Lawrence Lessig, contend that extending intellectual property protection to these innovations might chill technological development. Even Amazon.com founder Jeff Bezos--whose company owns a patent for its "one-click" technology--says that business methods and software patents' lifespans should be shortened from 17 years to three or five.
But are these high-profile opponents on the right track? Certainly these patents have been plagued with problems. However, it is not, as the critics contend, because of the dubiousness of patenting these particular subject matters. Instead, the problem lies both in the current system's inability to evaluate applications and in the weaknesses of the current patent law.
The solution is to strengthen the approval process, not to ban particular types of patents. The U.S. Patent and Trademark Office has recognized the need for reform and has suggested several steps for improving the process. But only a concerted effort by that agency, Congress, and the courts can bolster the patents' credibility--and ultimate acceptance. And only by allowing such patents will innovation be safeguarded in the United States.
First, Congress should ensure that the patent office has the resources to hire and train a sufficient number of patent examiners. If threatened budget cuts become a reality, the agency's efforts at making business method and software patents viable will be further undermined. Additionally, examiner requirements should be liberalized to permit individuals who have extensive software experience, but not the current academic requirements, to become patent examiners. Similarly, consideration should be given to hiring examiners with business degrees.
The patent office also must address problems with its prior art database for software and business methods. And that may require still more funding. Software historically has been categorized by use, rather than as software. This classification system has made it very difficult for examiners to find prior art. Further, because software applications are relatively new, the patent office has not compiled a sufficiently large prior art library. Making matters worse, many software inventions are not described in industry publications. In fact, these inventions often are recorded in ways that do not appear in the more formal system of technical documentation followed in other scientific fields. Innovations in biotechnology and electrical engineering, for example, generally are well documented in professional journals that have been subject to peer review. In contrast, software innovations may be recorded only in the commentary to a source code.
As Mark Lemley, a law professor at Boalt Hall at Berkeley, and Julie Cohen, an assistant law professor at Georgetown University, point out in a working paper (www.lawschoolstanford .edu/library/lsn/cyber22800.htm) to be published next year in Boalt's law review, the patent system presumes the existence of comprehensively indexed technical literature that examiners can search. But the diffuse nature of the knowledge base and the lack of a comprehensive system for cataloging and indexing software-related developments make it hard for any one person to locate prior art. This remains the problem despite the efforts of the patent office to improve examiner access to nontraditional sources of software documentation.
That is why the patent office should broaden the requirements to qualify as a patent attorney to those who may have the greatest knowledge about software and business method prior art. Applicants and their patent attorneys have a duty to disclose known prior art. PTO regulations demand that registered patent attorneys and agents have a certain level of scientific and technical expertise. The agency only recently expanded its rules to include courses in software design, programming languages, algorithms, and networking. Still not accepted are courses in mathematics, business administration, and operations research, or computer courses directed to business applications. Worse, many skilled programmers have very limited formal computer training. As a result, too many of those with the greatest knowledge about prior art are getting stopped at the patent office door.
The patent office is beginning to move in the right direction. In March, Q. Todd Dickinson, director of the agency, announced the "Computer Implemented Business Method Patent Initiatives" that, among other things, called for enhanced technical training for examiners, revised examination guidelines for computer-related inventions, and expanded search activities for all patent applications in the business method class.
From now on, a new examiner, not the one who originally rejected a particular patent, will handle a request to reexamine the patent based on previously undisclosed prior art. The plan also will test the use of an expert panel to decide whether to reissue patents that have been examined. A third policy change will require that prior art submitted by nonapplicants be reviewed at a higher level than presently as a possible basis for director-ordered reexaminations. Currently, prior art submitted by the public is put in the file and not reviewed. Director-ordered reexaminations occur when patents have a sufficient level of notoriety and raise a concern across an entire industry.
Although the patent office's initiatives are welcome first steps, legislation might still be necessary. Lemley and Cohen suggest that Congress should consider passing a law that, for research purposes, would expressly create a limited right to reverse-engineer software with patented components. Patent law today, unlike copyright law, contains no fair use or reverse-engineering exemption. Thus, it is possible for patent holders to use the grant of rights on a single component of a complex program to prevent others from making use of the program as a whole, including those temporary uses needed in reverse-engineering. Lemley and Cohen suggest that the present system may be impeding the growth of this market--which would be contrary to the principles of patent law.
Courts, too, can help by not construing business method patents too broadly. The doctrine of equivalents, for example, allows a finding of infringement even when the accused product does not literally satisfy each of the patent's elements, but does accomplish substantially the same result as the patented software in substantially the same way. Lemley and Cohen suggest that courts should apply the doctrine of equivalents narrowly in software infringement cases to avoid stifling legitimate efforts to design around existing software patents. They argue that a number of factors unique to software and the software industry (a culture of reuse and incremental improvement, a lack of reliance on systems of formal documentation, the short effective life of software innovations, and the inherent plasticity of code) severely complicate post hoc assessments of the "known interchangeability" of software elements. To avoid stifling creativity, Lemley and Cohen suggest that courts not only construe software claims narrowly but also refuse a finding of equivalence if the accused element is "interchangeable" with prior art that should have narrowed the original patent, or if the accused improvement is too many generations removed from the original invention. Clearly the patent office has issued a number of exceedingly weak software and business method patents in the last few years. But the proposal put forth by critics such as Lessig, to impose a moratorium or ban on issuing further patents, is not the solution. Rather, the answer is to reform the system and in that way foster technological innovation in the United States.