acm-header
Sign In

Communications of the ACM

Forum

Forum


In "Software Patents: The Good, the Bad, and the Messy" (Oct. 2007), Matt E. Thatcher and David E. Pingry concluded that "...the time and costs associated with collecting, analyzing, and maintaining the information necessary to identify and implement a good policy likely outweigh the potential benefits." The simple model of a lone innovator and lone imitator is not a good fit with the decentralized nature of the software industry. So I offer my own thesis—that software patents have actually harmed innovation and that an appropriate replacement is available to protect intellectual property.

Consider the following examples of patent system abuse:

The sole purpose of a multitude of companies is to find algorithms to patent, identify other companies that have translated their patented algorithms into software, and sue for royalties.

"Submarine patents" are deliberately unpublicized patents of algorithms that achieve widespread adoption, often through the approval of standards committees. After every software user has adopted the method, the patent holder reveals the patent and seeks royalties. Examples include the Unisys patent of the Graphics Interchange Format compression scheme and the Fraunhofer Institute for Integrated Circuits patent, standardized as MP3. In each case, adopters of the technology were ignorant of the patents until their use was widespread.

In his book Math You Can't Use: Patents, Copyright, and Software, Ben Klemens (programmer, mathematician, author of mostly technical stuff, and thinker about software patents) spelled out the damaging mismatch between the patent system and computer software and what can be done about it. Here's part of his story, in four points:

• The Church-Turing thesis implies that mathematics and computer algorithms are equivalent. There is universal agreement that mathematics cannot be patented, hence algorithms are not patentable. Nevertheless, the U.S. Patent and Trademark Office and its counterparts in other countries have been aggressive about issuing algorithm patents. They have little of the expertise needed to evaluate the appropriateness of applications or be familiar with "prior art."

• In some domains (such as development of new drugs), relatively few organizations participate, and keeping track of patents seems to work. However, many companies have IT departments or just a few employees working on keeping their computers and their users happy. The extremely distributed nature of software development means there is no effective means for a programmer, who may have "invented" an algorithm and expressed it in a programming language, to search for already-patented algorithms.

• Obvious algorithms are being granted patents. Klemens' best example is this snippet of JavaScript for pop-up browser windows, granted patent 6,389,458 on May 14, 2002:

  • function onExit(){
  • popup = window.open("pop.html", "Don't go!");
  • popup.focus() ; }

There are only two steps, and any experienced programmer would write them the same way. However, the USPTO was apparently snowed because the patent application described in great detail what these two library calls do. We thus need to simplify the process used to overturn patents already granted to repair the damage USPTO has done.

• Copyright law could protect computer software. People who plagiarize have violated copyright, and mechanisms are in place for redress. But if they independently reinvent an algorithm and express it in other software, there is no violation. There are effective automatic and semiautomatic text-comparison systems for overcoming obvious and mechanical program changes, and copyright holders are protected against such changes.

Patents should therefore not be granted for algorithms or software. Copyright law is perfectly capable of protecting intellectual property within the boundaries of what is professionally defensible.

Rod Oldehoeft
Santa Fe, NM

The premise of "Software Patents: The Good, the Bad, and the Messy" (Oct. 2007) wasn't focused correctly. The goal of any patent system is to provide incentives to innovators to make their inventions public so other innovators are able to benefit from knowledge of advances in the state of the art, enabling further innovation that would not occur if the invention were kept secret. The innovator must therefore be given significant protections that preclude imitation or duplication. The only consumer benefit allowed must derive from innovation that benefits from leverage provided by the disclosure but which includes no imitation. To this end, the profit protection provided by patents must be significant—long in duration, with substantial legal strength. Moreover, patents must be granted only for substantial novelty.

The main problem with the current patent system is that it requires insufficient novelty for the granting of a patent. This comes in part from a dearth of patent examiners and a philosophy of "grant a patent if there is any chance that the application has true innovative content, and let the courts decide if the patent should have been granted." This leads to patent applications with limited, often trivial, incremental innovation—a practice used in some quarters to box-in a significant innovation—while blocking incremental improvement by the original inventor.

This compounds the flaw introduced by requiring the courts to decide on a patent's technical merit. Also, the breadth and depth of today's technologies and literature about them make it difficult to find patent examiners with competence in every specialized area of technology.

To protect inventors from the loss of competitive advantage that might be realized if their innovation were kept secret, the secrecy of a patent application must be maintained until the legal protection of the patent is in place. This precludes adopting rules, like those in the European Union, that would publish a patent application after some period of time, even if a patent had not been granted.

Given the global economy, the developed nations must insist that all other nations recognize their issued patents, as well as their equivalent copyrights, and enforce them in their own jurisdictions. To be effective, such enforcement must include the possibility of punitive action against those who fail to do so, and such action must be automatic, with limited opportunity for circumvention or waiver.

John Montague
Columbia, SC

Authors' Response
Oldehoeft and Montague both agree with our model that insufficient novelty ("low height") is bad while citing process problems. Montague concludes that patents are usually bad. We conclude that while patents theoretically create benefits (see our "Good" discussion), the likelihood is low, given the information and process requirements for determining good policy (see our "Messy" discussion). Oldehoeft believes that patents, though often bad, are necessary to provide the incentive needed to make ideas public. Reverse engineering makes this point mute for software, since the idea is made public with the first sale. Our model structures this debate by focusing on the trade-offs among height, patent imitation level, and length, no matter how they are determined.

Matt E. Thatcher
David E. Pingry

Errata: A printing error in the November 2007 issue caused the omission of some author information on page 69. Paul P. Tallon (paul.tallon@bc.edu) is an assistant professor of IS at Boston College. Richard Scannell (rscannell@glasshose.com) is senior VP Sales and Marketing at GlassHouse Technologies, Inc., Framingham, MA.


©2007 ACM  0001-0782/07/1200  $5.00

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.

The Digital Library is published by the Association for Computing Machinery. Copyright © 2007 ACM, Inc.


 

No entries found