For many PC programmers, the topic of software patents has long been a MEGO topic--as in "my eyes glaze over." But like it or not, this subject may consume more programmer time over the next few years than dealing with the DOS 640K limit ever did. Furthermore, patent policies may have more lasting impact on our industry than any current OS war or API skirmish. For many, the wakeup call was the recently awarded Compton's multimedia patent which aroused concern in many who might otherwise be thinking about a new subroutine for the next killer app.
As with any controversy, there is a spectrum of opinion, as demonstrated by the recently concluded public hearings held by the U.S. Patent and Trademark Office (PTO) in San Jose, California, and Arlington, Virginia.
Speaking in San Jose were high-level representatives of many major software and computer companies, including IBM, Apple, Microsoft, Borland, Adobe, Oracle, Intel, Autodesk, AT&T, and Sun. There were also some multimedia-oriented firms, a lot of lawyers from intellectual-property law firms, and a small handful of programmers.
One thing that this diverse group seems to agree upon is that the present process of granting software patents needs to be fixed. Opinions differ on whether the fix should be a minor tweak or major surgery. One surprising force for change--surprising to those who think nothing good can come from government--comes from the PTO Commissioner himself, Bruce Lehman.
The soft-spoken Lehman, who previously worked as an intellectual-property lawyer for the high-tech industry, is conversant with all areas of the controversy, remarkably candid about PTO actions, and actively seeking advice from all corners of the debate. Some weeks before the Compton's brouhaha erupted in November, Lehman admitted at the White House-sponsored "high-tech summit" that:
The patent office has not been very well prepared to deal with the issue of software patents. We did not have patent examiners skilled in this area. There is a unique aspect to the software industry, compared to other high-tech industries such as biotech. In other sectors, there is a strong database of prior art, which is absolutely essential to patent examination. We still don't have a database of prior art for software, and the industry has historically relied on trade secrecy. The combination of all these factors has resulted in a mess. It is pretty clear that we issued patents that will not be held valid in court. We want to get to the bottom of it and fix this. We have increased the training and sophistication of the examining staff. We have tried to expand the prior-art database. We are also seeing what can be done in the area of policy issues. But what we do is determined and constrained by the courts. Regarding a moratorium on software patents, we have no choice in that. We have to carry out the law, we'd be under injunction if we tried a moratorium. We need to fix what is in the pipeline. We will have hearings in January. I want to find out what the industry is thinking. If the courts tell us we can't undertake a particular course of action, we need to get the courts to clarify the law. We can file amicus briefs. If that doesn't work, we'll go to Capitol Hill [to get the laws changed]. We need to get this policy problem under control.
By contrast, the statements at the January hearings were more formal and less candid than Lehman's remarks quoted here. Each person got 11 minutes to deliver prepared remarks. Among those who spoke were the general counsels for IBM, Borland, AT&T, Exxon Research, Intel, Sun, Taligent, Microsoft, and Silicon Graphics. This preponderance of lawyers proved a disappointment to those hoping to hear remarks by people who actually know how software is created. Among the few programmers, engineers, or nonlawyers who spoke were Doug Brotz (principal scientist at Adobe Systems), Richard Stallman (well-known antipatent activist), Jerry Fidler (CEO of WindRiver Systems) and Jim Warren (eminence grise of the PC industry and former DDJ editor). With the possible exception of Warren, there were no speakers representing the general public, the consumer who has to deal with the impact of patent decisions--such as artificially incompatible user interfaces, higher prices, or longer times for technology delivery.
There were 14 law firms represented, all favoring software patents. For example, Robert Sterne, an attorney for an intellectual-property law firm, spoke of how he is currently involved in a $100M financing of a high-tech startup by a venture capital firm, and how important it was to the venture capitalists that the startup have a portfolio of patents: "Without strong intellectual property rights [in the software industry], the business community will invest in less-risky ventures."
In addition to attorneys at law firms, there were presentations by lawyers who serve as general counsels for high-tech corporations or who are in top corporate-management positions. There were 12 such testimonies, and almost all favored software patents, along with suggestions for minor improvements to the system. For example, Victor Silber, general counsel for IBM, was one of several company lawyers who stated that any real change to the existing system "will hurt the computer industry in the short term, and in the long term, it will hurt all industry." In Silber's view, the key issue is whether PTO can assign patents to inventions that are truly new and unobvious. This, he says, is "a job that is doable." Silber said that IBM "completely opposes" any proposal to set a different standard for software patents, because it would shift investment away from software to other industries.
In general, those who predicted economic doom did so without any economic or statistical data to back up these assertions. By contrast, the League for Programming Freedom (LPF) provided hard data to back up its statement that "most software patents are held by companies that history has proven_to be totally incapable of delivering innovative software products to market." For software patents granted from 1990 to 1992, the top five patentholders are IBM (with 500 patents), Hitachi, AT&T, DEC, and Toshiba, for a combined total of 1350 patents. By contrast, the companies that form the "backbone of the software industry," such as Microsoft, Adobe, Lotus, Novell, Borland, and Oracle, weigh in with a total of 29 patents. According to LPF's printed handout,
IBM has a very strong software patent portfolio. It is oversized even in proportion to the size of IBM itself. This is a result of IBM's patenting every single trivial idea every employee comes up with, rather than any great propensity to be truly innovative.... IBM even has a patent on a software application to permit employees to automatically document ideas for later patenting.
By contrast, "the companies that collectively constituted the microcomputer revolution" have relatively weak patent portfolios, resulting in a "negative correlation between the number of patents granted to a company and its ability to bring innovative software products to market."
A few representatives of these self-same innovative companies did speak at the hearings. The position held by Adobe, Oracle, and WindRiver Systems, among others, is that they are changing their business practices to emphasize patents, not because they support the notion of software patentability, but because patent portfolios have become a new weapon to use in today's competitive arena. According to a written statement from Oracle,
Oracle Corporation opposes the patentability of software. [We] believe that existing copyright law and available trade secret protections are better suited to protecting computer software developments... . [Patents] are not appropriate for industries such as software development in which innovations occur rapidly, can be made without a substantial capital investment, and tend to be creative combinations of previously known techniques.... Unfortunately, as a defensive strategy, Oracle has been forced to protect itself by selectively applying for patents which will present the best opportunities for cross-licensing between Oracle and other companies who may allege patent infringement.
The purest antipatent argument was expressed by Richard Stallman, who, as primary author of GNU Emacs and a world-class programmer, has arguably come up with more patentable ideas than the rest of the people in the room combined. GNU Emacs consists of 500,000 lines of code and, along with other GNU software is in daily use by more than one million people. Stallman used a mathematician's reasoning: Assume, for the moment, that the patent office is perfect and only issues patents for ideas that are new. The problem, then, is that software systems, unlike any other human artifact, are very complex and rely on combinations of numerous components. Many systems are now over one million lines of code. Programs such as these are not composed of one basic idea, but of a multiplicity of small ideas, perhaps even several thousand. If a programmer has to consider--for each idea, subroutine, or data structure--whether it infringes on an existing patent, progress in software will come to a standstill. According to Stallman, the present system of copyright laws and trade secrets provides ample protection for those companies making commercial software, as evidenced by the tremendous pace of growth and innovation in the early '80s, before the issue of software patents emerged
It is telling that of the 12 participants who were identifiably programmers or technical managers, ten were opposed to software patents. In addition to this group, there were perhaps a dozen or so participants that don't fall into any of the categories mentioned earlier. Opinions in this group were split. Note that DDJ did not attend all sessions on both coasts, but did rely upon a detailed summary of the San Jose testimony posted to the Internet by Greg Aharonian. Note also that it is hard to characterize certain positions as pro or con, because the issues are more complex than a binary decision.