Software patents need shelter from the storm

COMMENTARY--Software patents are a controversial issue, all themore so as the European Parliament considers changesthat would bring European patent law closer to themore permissive regime found in the United States.

This has resulted in a flurry of criticism, so much sothat the European Parliament has deferredits decisions until Nov. 4 in order toconsider objections raised by economistsas well as spokespersons for the open sourcecommunity.

This is a somewhat unusual debate, as patent-relatedissues have less to do with technical matters normallythe subject of computer-oriented discussions so muchas economic questions of efficiency, levels of R&D and incentives. Many would argue that since patents are useful in other technical disciplines, they will be useful in software development. I argue that the nature of software development makes it sufficiently different from other technical disciplines as to make unnecessary the economic expedient of patents.

The case for patents, in a nutshell
As areport commissioned by the European Parliamentnoted, A patent is not a "natural' right, but aninstrument of economic policy. As such, it must bejudged on its economic merits.

Patents are an economic tool designed to remedystructural aspects of certain markets which hindertechnical research. Certain markets, such aspharmaceuticals, require large R&D expenditures todiscover a new medicinal compound. Medicinalcompounds, however, are hard (and expensive) todiscover but easy to copy, which can deprive aresearch company of any return on the billions itspends on research.

Patents grant their owner a limited monopoly on the“idea” identified by the patent. This ownership islimited to 20 years, a length which is enshrined inthe TRIPs agreement to which all members of the WorldTrade Organization (WTO) are signatories. Suchmonopoly power gives the owner the right to license atprices he or she sees fit, or conversely, the rightnot to license at all.

Such monopoly rights restrict competition for thelength of the patent, which is an obvious societalcost. On the other hand, it also ensures the owner areturn on research expenditures which went into thediscovery of the patented idea. This makes largeexpenditures for easily copied product, such as pharmaceuticals, easier to justify, thus boosting research and development throughout an industry.

Such an assured source of revenue can have benefitsfor Small to Medium-sized Enterprises (SMEs). Patentsare valuable assets, and they can be used to secureloans and/or venture funding, as well as serve as asource of stable revenue. Nicholas P. Godici,Director of the U.S. Patent Office, mentioned Herman Hollerith’s 1889 patent on a method for tabulating and compiling statistical information in testimonybefore an American congressional subcommittee. This patent provided a stable source of revenue forMr. Hollerith’s fledgling business, which enabled itto grow into a certain company better known under itslater name, International Business Machines (IBM).

Patents must be fully described as part of the patent application process. This facilitates the dissemination of the patented idea into the public domain, which becomes most useful once the patent expires. The monopoly aspect of patents also provides incentives to properly market the idea, as a patent-owning company will generate all revenue from the results of that marketing. This serves to popularize the technology so that it is universally understood by the time the patent lapses.

A final justification is what I’ll call the “exemplar”theory. The United States has the most vibrant ITindustry in the world, an industry that has grown todominate IT markets around the world. America alsohas a liberal patent regime, and this is causallylinked, according to the theory, to America’s ITsuccess. Therefore, Europe needs software patents inorder to make an IT environment wherein Europeancompanies can thrive as American companies presumablydo under America’s patent system.

Regarding the R&D effect
Most ideas are built, in some fashion, on older ones. This is particularly true of software. The kernels ofmost operating systems are reported to be verysimilar, which doesn't apply to the higher-leveltechnology that has grown around this core like coralaround a sunken ship. In addition, simple RemoteProcedure Calls (RPC) are an old concept, but SOAP andXML-RPC calls are ehancements of that model, buildingon basic principles laid out in the "ideas" associatedwith RPC.

In addition, the market for software creation is huge. It is estimated that there are 10 million activeprogrammers in the world. This is a truly massivenumber, dwarfing the number of innovators in othertechnical disciplines. This large number is drivenboth by the demand for software product as well asbarriers to entry which are unusually low. Anyonewith a cheap computer can acquire the tools requiredto build software. This means that R&D in thesoftware industry is noticeably cheaper than R&D inother industries, and this, combined with largenumbers of programmers, leads to a lot morecontributors to global software R&D.

This leads to a fast-paced software market composed oflarge numbers of competing products. These productscompete by creating new “features” to include in newreleases. In other words, software is uniquelydependent on the creation of new ideas, and as such,will be naturally inclined to high levels of R&D. The question, therefore, is if there is any R&D gain to be derived from the economic expedient of software patents.

Evidence suggests that there isn’t. James Bessen andRobert Hunt, ina "Research on Innovation" study titled "An EmpiricalLook at Software Patents," found that R&D wasactually REDUCED in the presence of a vibrant softwarepatent system. From their introduction…

…software patents substitute for firm R&D; theyare associated with substantially lower R&D intensity. Overall, the predominant use of software patentsappears related to strategic “patent thicket”behavior.

This makes logical sense if you keep in mind thatsoftware is extremely reliant on the creation of newideas. In the absence of patents, old research whichslowly worked its way into the public domain (evenproprietary software gets reverse engineered) wouldforce companies who wished to remain competitive tomaintain high-levels of R&D. In contrast, a strongpatent system allows companies to generate longer-termrevenue benefits from patentable IP, giving them lessincentive to conduct the same levels of research.

In addition, remember that new ideas are reliant uponthe presence of older ones. If those older ideas arepatented, then someone has a 20 year lock on that coreidea. This can be dangerous for innovation in thesoftware market, a market characterized by rapidchanges and a proliferation of new ideas. Licensingtollbooths on the foundations of new ideas can greatlyhinder the appearance of new ones.

Regarding SME revenue from patents
If Small to Medium-sized businesses were creatingpatents in large numbers, then such businesses mightbenefit from the value inherent in the grant of apatent. Unfortunately, the primary actors on thepatent stage are large companies. A number of thestudies I read for this article recommended doing moreto educate small to medium-sized businesses of thebenefits of patents, in hopes that this would spurthem to patent more.

This won’t have much effect, in my opinion, as thereare a number of reasons why smaller companies willstill choose to avoid the patent process.

…, SMEs relative lack of interest in the patentsystem might have a more rational explanation. Possibly, SMEs’ make conscious decisions to stay awayfrom patents, for a variety of reasons – costs andeffort being important factors. Typically, patenting(or defensive action against patent claims) will taketime from the same few people that are the creativebrains behind the inventions. In practice, many SMEstend to rely on copyright protection for software,which is readily available and may provide all theprotection SMEs really need. (Thepatentability of computer programs)

Likewise, consider that SMEs will never acquirepatents in numbers comparable to large companies. This gives large companies a stronger hand innegotiations, and reduces the revenue-generating powerof the handful of patents SMEs have in theirpossession. More often than not, SMEs will be forcedto effect patenttrades with large companies rather than managelicensing arrangements which would serve to bolstertheir bottom line.

Another reason small companies avoid the patent raceis cost. The application process is expensive enough,but it is the legal costs associated with defending aclaim in court which is the real burden. Of course,defending against an external patent claim can be justas expensive, but in future, there may be no way toavoid such patent suits, even with large, defensivepatent libraries.

I base this on the growth of American companiescomposed of nothing more than a handful of patents anda team of lawyers to defend them. Such “litigation enterprises,” such as Eolas, owner of a patent on generic plug-in technology withwhich they have extracted large fees fromMicrosoft, and PanIP, a group that is suing small e-commerce sites over its claim to own the “technology” of online sales, are free of the need to trade for patents, as they have no product to speak of which would require access to technology patented by large companies.

They are free, therefore, to pursue the full revenuepotential that patent-derived monopoly power grantsthem. These companies produce no new ideas, and serveas a tax on software companies oriented aroundcreation of tangible product and the new ideas that goalong with them.

Regarding information dissemination throughpatents
Few in the software industry cull the patent libraryto find algorithms or business methods. For the mostpart, patents are a legal instrument to be used insoftware competition, not a source of information.

It is, however, doubtful whether the informationfunction actually plays an important role in thesoftware industry. German research shows that patentarchives are primarly valued as a source ofinformation for legal purposes, and only to a lesserextent as a source of technical information. (Thepatentability of computer programs)

This is not to say that patents can’t serve as adecent source of information. Unfortunately, patentsare barely readable unless a programmer happens to beversed in the verbal acrobatics of legal documents. Patents are written to enhance the appearance ofinnovation in order to enhance the application’schances of being approved. Simple concepts that canbe described in a sentence somehow get expanded intothree or four paragraphs. This makes them poorsources of information, and partly explains whydevelopers don’t tend to explore patent libraries.

On the other hand, if you replace the “pull” modelwherein developers go hunting for technology found inpatent databases for the “push” model, whereincompanies go out and educate developers about theirpatented technology, the claim appears more credible. For instance, RSA Security had a vested interest ineducating as many developers as possible about theirpatented technology (now lapsed), as more usersimplied more licensing revenue.

If software patents offer any benefits, it would bethrough the incentives created to market properly anidea. If the net effect of software patents is, asI’ve claimed, to reduce R&D and new idea creation ingeneral, then this incentive is of dubious benefit.

A realistic appraisal of the American patent experiment
It is often suggested that America’s success in Information Technology is causally linked to a liberal patent regime. A consideration of the history of American software patents would suggest otherwise.

The foundation of America’s current patent frameworkwas laid by court cases in the 1980s and 1990s. Patents on algorithms were granted in the 1970s,though these were usually tied to a hardware processwherein the software was intimately associated withthe operation of that hardware. Even so, thesepatents were legally dubious until 1981, when theSupreme Court ruled in Diamond v. Diehr thatalgorithms were patentable under certain limitedcircumstances.

This did not result in an explosion of algorithmpatents, as the ruling was sufficiently vague as torequire subsequent rulings to clarify matters further. The most important clarification came in 1994 withthe In re Alappat case, where the court ofappeal for the federal circuit ruled that the standardby which algorithm patents would be judged was thatthe result must be “useful, concrete and tangible.”

The doors to business method patents were opened in1998, in the case State Street Bank and Trust Co.v. Signature Financial Group Inc.. As NicholasGodici explained, (t)he Federal Circuit in StateStreet explicitly rejected the notion that a “businessmethod” exception exists in United States patent law,thereby ending any notion that inventions deemed to bebusiness methods, by whatever criteria, would beexcluded from patentability on that basis alone. Thus, the State Street decision clarifies that aninvention deemed to be a “business method” will betreated in the same manner as any other method orprocesss invention ( NicholasP. Godici, Acting Undersecretary of Commerce forIntellectual Property).

There are a number of reasons I might suggest forAmerica’s success in Information Technology, such asan entrepreneurial culture, a relatively hands-offapproach to business, a flexible labor market, apersonal responsibility approach to businessorganization (something I should write about in afuture article) and low-cost computer products(relative to other places in the world) which leads tomore consumers who can use software products. Software patents, however, simply played no role inthe success of American IT companies. They becamepopular too late to have much of an effect.

Some Parting Thoughts
America is the largest software market in the world. America also has the most permissive patent regime inthe world. Since most European companies will want tosell to that large market, they will be obliged tochase down patents in order to navigate America’spatent minefield.

One might argue, therefore, that Europe’s softwarechanges make little difference, given that Europeansoftware companies will still have to contend withsoftware patents. I disagree. It is worthwhile forEurope to chart a different course while Americaconducts its experiment in software patents. Thoughvarious parties disagree as to whether softwarepatents are good for the industry, the consensus seemsto be that there isn’t conclusive evidence as of yetwhether software patents are a help or a hindrance.

I believe that once a final accounting is made,America’s liberal software patent regime will provemore a hindrance. When that becomes clear, it wouldbe worthwhile to have a working example in Europe toserve as a guide when America decides to alter its ownlaws.

As a final point, one of the motivations for theproposal before the European Parliament was the needto standardize patent rules across Europe. Enforcement is currently left to national governments,and the rules governing the patentability of softwareare sufficiently vague as to allow differences ofopinion. A patent that was ruled invalid in the UKmight pass muster in Germany.

Standardization is a noble goal, and such aclarification is necessary if Europe is to create aunified market for software production which iscompetitive with America. I would suggest, however,that that standardization should move away from aconfirmation of software patents. Allow the parallel experiment in patent regimes to continue a bit longer. We already have one large market with liberal software patent laws (two if you include Japan). We don’t need another until the evidence points to the clear benefits to be derived from software patents.

biography
John Carroll is a software engineer now living in Geneva, Switzerland. He specializes in the design and development of distributed systems using Java and .Net. He is also the founder of Turtleneck Software.