Click for next page ( 22


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 21
2 Background to Basic Legal Issues Like other industries, the software industry operates within a large legal framework of which intellectual property laws are only a part, albeit a crucial one. Tax laws, the Uniform Commercial Code, and antitrust laws, among others, also regulate the behavior of software firms, and they too can influence the scope of inventive activity and the dissemination of innovations to the public. Most legal debate in the software industry, however, is confined to copyright, patent, and trade secret laws. That the debate encompasses both the copyright and patent domains of intellectual property law seems to provide some support for the contention of software devel- opers that their technology is unique. "Software is perhaps the first patentable subject matter which is also copyrightable," writes Duncan Davidson (Davidson, 1986, p. 1055~. "No one yet knows what this means, if anything." But Vanderbilt University law professor Jerome Reichman questions the historical accuracy of this assertion. "Industri- al designs," Reichman said, "are, to varying degrees, covered by both copyright and patent laws in all industrialized countries. Indeed, efforts to broaden copyright protection of industrial designs in the period 1900 to 1950 and the corresponding tensions this generated seem to anticipate the present tensions concerning software in almost every respect." Under the three major domains of intellectual property law, pro- tection is awarded to software not as a class, explained Ronald Laurie, head of Irell & Manella's computer law group, but on the basis of whether a program, language, interface, or other software element 21

OCR for page 21
22 INTELLECTUAL PROPERTY ISSUES IN SOFTWARE "possesses attributes or characteristics that fall into one or more of the statutorily prescribed categories of protectable subject matter." For copyright law these hallmarks are expressions of ideas embodied in original works of authorship. Patent protection is awarded for products and processes that are novel, nonobvious, and useful. Trade secret law can be invoked to protect secret information that is used by a business and contributes to its commercial advantage. The ex- tent to which software products satisfy these criteria is, of course, a matter of interpretation, which is where much of the uncertainty arises. Failure to satisfy the criteria means a work is available for all to use. Within copyright law and, to some extent, patent law, these broad defining characteristics of protectable subject matter may have finer gradations. Particularly relevant to software are subcategories of copyrightable literary works, such as works of fiction, works of his- tory, manuals, and telephone directories. The scope of protection var- ies according to the range of creative expression found in each type of work. (See Box 2.1.) COPYRIGHT Copyright has been advanced as the preferred intellectual prop- erty protection for software in the United States and, with this nation's strong encouragement, internationally. In the 1980 amendments to the Copyright Act of 1976 (P.L. 94-553), Congress confirmed its inten- tion that copyright protection applies to computer programs, but the 1976 revision of the federal copyright law did not stipulate the man- ner in which the protection applied. Through the 1980 Software Amendments, and acting on the recommendations of the National Commission on New Technological Uses of Copyrighted Works (CONTU), Congress placed software more squarely within the embrace of copyright by providing a definition of computer programs. In so doing, Congress continued a long tradition of extending copyright protection to new media of expression. Only books, maps, and charts were singled out in the first copyright law, passed in May 1790. Subsequent additions to the law included protection for musical compositions, photographs, drawings, statuary, paintings, motion pic- tures, and sound recordings. Reichman pointed out, however, that statutory authority to protect industrial designs-a category that, he believes, corresponds closely to software-was never implemented until 1954, and from 1958 to the present, the separability doctrine put forward by the U.S. Copyright Office "has effectively denied copy- right protection to functional 'high-tech' designs of useful articles of high technology that have three-dimensional embodiments."

OCR for page 21
BACKGROUND TO BASIC LEGAL ISSUES 23 BOX 2.1-LEGAL PROTECTIONS AS ISLANDS IN THE SEA OF FREE COMPETITION To give forum participants the lay of the intellectual property law landscape, attorney Ronald S. Laurie of Irell & Manella used a meta- phor-based model to describe the conceptual underpinnings of copy- right and patent laws and their application. Intellectual property protections are like islands in a sea of free competition. . . [Ilf one is not able to place the fruits of one's investment, ingenuity, or creativity on one or more of the islands, then one is in the sea, and we all know what lives in the sea. The model is, of course, simplistic and imperfect because it assumes that one can always tell when one is on dry land and one is in the water of unrestricted competition. In fact, the copyright island, which is low and sandy, has a gradual sloping beach called the merger of idea and expression. Thus, opinion may differ on how far out one has to wade before the boundary is crossed; that is, when the water is up to your neck, are you still on the island' This is especially true where the tide (read: inconsistent judicial decisions) varies with the time of day. The patent island is a volcanic island with sheer cliffs rising to a commanding view of the surroundings. But the patent island also has a sandy beach, tucked away in a corner. It is called the doctrine-of-equivalents beach. I think we are going to hear a lot more about this area with the rising tide of software patents that are being issued and, of course, the consequent flood of software patent litigation. One of the often-cited differences between patents and copy- rights is that patents offer a much more predictable right that patents have claims and, within certain limits, adverse parties can look at claims and decide whether they are on the right side or wrong side of the infringement rule. Whereas, as we all know, no one knows the scope of a copyright until a judge tells us what it is, and different judges will tell us different things about the same copyright. I suggest that this reassuring element of predictability in the case of patent law is not as sure as you might think. Indeed, Laurie predicted that interpretation of the doctrine of equiva- lents is likely to spawn controversies and appeal to levels of abstraction akin to those encountered in arguments of substantial similarity invoked in copyright infringement cases involving the structure, sequence, and organization of programs. "I suggest," Laurie continued, "that as we discuss some of the difficult legal and technical issues, such as the protection of interfaces, commu- nication protocols, and languages, the relevant question is not whether these areas are generally protectable as a class, but whether they may possess attributes or characteristics that fall into one or more of the statutorily prescribed categories of protectable subject matter."

OCR for page 21
24 INTELLECTUAL PROPERTY ISSUES IN SOFTWARE Owners of copyrighted software are awarded exclusive rights to their works for varying periods, but typically with a 75-year maximum. Procedurally, copyright is an automatic protection, conferred as soon as an expression is fixed in a tangible medium, even if the work is never published. Registration with the Copyright Office is not re- quired, nor is full disclosure necessary. To secure a registered copy- right, the creator of a program need only submit descriptive materials on the first and last 25 written pages of the work, which usually do not constitute the entire program. Copyright gives the creator control over several activities, includ- ing reproduction, distribution, and adaptation or translation into deriv- ative works. Several recent and ongoing lawsuits are testing the limits of these rights by raising the issue of what constitutes "sub- stantial similarity," the basis for determining copyright infringement. Expressions, Ideas, and Functions Copyright protection is extended only to expressions of ideas, not to the underlying ideas themselves. This distinction is especially critical for the software industry, where independent invention is common. Two designers may set out to tackle the same problem, each one creating different programs that accomplish essentially the same task. There are no grounds for copyright infringement if neither competitor copied the other's work. The situation is akin to one that might exist for two writers of romance novels, both of whom have the same general idea for a plot but whose finished works, although comparable at an abstract level, differ greatly in style, character de- velopment, and other literary aspects. Allowance for independent invention distinguishes copyright law from patent law, which, as is described below, provides protection for ideas at some level. Thus copyright protection is said to be narrow, or "thin," because it permits duplication of function, a feature that becomes apparent when one shops for a word-processing program. Not all copyrightable works are accorded the same extent of copyright protection. Artistic or fanciful works are considered to have a broad scope of protection by copyright because they are predominantly ex- pressive in character. Factual works have a somewhat narrower scope of protection, because the facts and theories they contain are not protected by copyright. Works of a more functional character, such as architectural plans, recipe books, rule books, and the like, have traditionally been considered to contain little expression. Copyright law has regarded the functional content of such works to be among their unprotectable features, and hence the scope of protection for

OCR for page 21
BACKGROUND TO BASIC LEGAL ISSUES 25 such works has been quite narrow, requiring exact or near-exact copying of the expressive aspects of the work for infringement to be found. Some commentators on software copyright law would compare the complex structure of a computer program to the complex struc- ture of novels, and would argue that just as the complex structure of a novel might be protectable expression under copyright law, so too should the complex structure of a program be protectable. Others would argue that the functional character of computer programs would suggest that a narrower scope of protection is appropriate under copy- right law on the theory that the design of software is more akin to the engineering design for a bridge (which a copyright on a drawing would not protect) than to the design of a novel. Software's peculiarities have spawned uneasiness in some quar- ters about the ability of copyright law to provide effective protection, a term with many interpretations. Indeed, Paul Goldstein, StellaW. and Ira S. Lillick Professor of Law at Stanford University, reflecting on the many changes in the industry, speculated that if CONTU had had the benefit of today's understanding of software, that body's recommendations would have been markedly different from those actually made a decade ago. One source of vexation, at least for some, is the dissimilarity be- tween software and its copyright analogue, literary works. Only rarely, for example, are computer programs sold or licensed in a written form even remotely understandable to people other than expert computer programmers. Most software is distributed as machine-readable ob- ject code, written as sequences of numbers. Except for the user inter- face, software-unlike literary works-does not reveal its expression to the consumer. Not all find this objection compelling. "The fact that it requires a certain competence to read a computer program with appre- ciation," Anthony L. Clapes, IBM's senior corporate counsel, has written, "merely puts computer programs in company with foreign language texts and other specialized literature: the ideas in them are not unintel- ligible; they simply do not yield themselves up to those untrained in the language in which the ideas are expressed" (Clapes, 1989, p. 143~. Also grounds for debate is the functionality of software, a prop- erty that further blurs copyright law's already fuzzy line between idea and expression. For practical purposes, idea and function are virtually synonymous, falling within the realm of patent law. Thus, in theory, copyright law does not protect the functionality of programs, but rather the expressions that result in the accomplishment of a given task. But the question of the extent to which copyright pro- tects the functional elements software possesses, such as the command sequence for footnoting text, the arrangement of graphical symbols

OCR for page 21
26 INTELLECTUAL PROPERTY ISSUES IN SOFTWARE on a display, or other behavioral components of a program, will turn on the courts' judgments as to whether nonfunctional elements have been copied. Goldstein attributed the legal uncertainty besetting the software industry to the "tension between the essential functionality of computer programs and copyright law's historic refusal to protect functional elements of otherwise qualifying works." Morton David Goldberg, of the New York law firm Schwab, Goldberg, Price, & Dannay, disagreed with this contention. He noted that copyright applies to several types of utilitarian works, including maps and charts. Pamela Samuelson of the University of Pittsburgh contended, however, that copyright law has traditionally not considered maps and charts to be "utilitar- ian" in character, for they merely convey information or display an appearance. "Works that have functions-in addition to conveying information or displaying an appearance have been utilitarian in a copyright sense, and hence unprotectable by copyright," she said. "Computer programs are the first truly utilitarian work to be protectable by copyright." Copyright law recognizes the "merger of expression and idea," an argument raised in many copyright-infringement cases but one that seems to have particular relevance for software. The merger doctrine holds that if an idea is inseparable from its expression, then only one or a few options exist for accomplishing a specific function. Therefore the expression is not protected, and copying is permissible. To do otherwise would grant monopoly control over an idea. Here, the machine side of software's dual nature becomes the focal point. The ability to do work, it is often argued, entangles idea and expression. If two programs are designed to perform the same function, will not the range of expression be limited, dictated by the task? After all, other machines cars, for example vary in style and detail, but their form is constrained by their function, necessitating many shared fea- tures. The validity of the argument that expression is idea in software hinges on circumstances and, thus, judicial interpretation. In one case, for example, the court ruled that duplication of the sequence of data-input formats for a structural-analysis program did not consti- tute infringement, finding that the sequence was not expression, but an idea. The merger of idea and expression also was found to underlie the similarities in the "sequence and organization" of competing marketing information programs for cotton farmers. In this case, Synercom Technology, Inc. v. Unni Computing Co. (462 F. Supp. 1003 [N.D. Tex. 19781), the need to conform with the conventions of the cotton mar- ket was thought to constrain the range of expression. The "merger"

OCR for page 21
BACKGROUND TO BASIC LEGAL ISSUES 27 argument has also been rejected in many software cases, including those in which the issue was not literal copying but nonliteral simi- larity. What Constitutes Copyright Infringement? Until only recently, intellectual property concerns in software were limited largely to literal copying. In the 1980s, new issues arose and, with them, a diversity of views on the scope of copyright protection. One boundary on this wide spectrum of opinion, the minimalist ar- gument, posits that extending protection beyond the computer code and perhaps some audiovisual and tutorial features of the user interface will encourage monopoly and stifle competition. Maximalists, who reside at the opposite boundary, counter that such limited protection will remove the incentive to innovate. Although some may align themselves with one view or the other, most software firms are not likely to endorse either one, preferring a practical view consistent with the realities and behavior of the industry. They recognize the need for programs to work with other programs, as demanded by users. Meeting this need will require using parts of others' software, as has been the industry's practice for much of its history. Overly broad protection would present a major obstacle, with each effort to create compatibility carrying the risk of a lawsuit. Underprotection, however, could create a paradise for free riders, producers of "knock-off," or cloned, programs. Some lawyers see the law as steadily evolving to workable compro- mise that will accommodate software. In a review of decisions on the protectability of the "structure, sequence, and organization" of pro- grams, Goldberg contends that the "developing body of case law in this area does provide helpful guidance, and will do so increasingly as more cases are decided" (Goldberg and Burleigh, 1989, p. 296~. In her review of the situation, Samuelson draws the opposite conclusion, finding consensus extending only to copyrightability of computer programs. "Almost all of the important questions about what copy- right protection means for software have yet to be answered defini- tively," she has written (Samuelson, 1988, p. 611. Not surprisingly, these two legal experts also disagree on the ap- propriateness of, arguably, the most influential decision handed down thus far on the scope of copyright protection for software, that of the Third Circuit Court of Appeals in Whelan Associates, Inc. v. fasiow Dental Laboratory, Inc., et al. (U.S. District Court for the Eastern District of Pennsylvania, 797 F. Supp. 1222 [19851~. Goldberg's and Samuelson's differences are explored below, with the aim of elucidating some of

OCR for page 21
28 INTELLECTUAL PROPERTY ISSUES IN SOFTWARE the disagreements in assessments of the applicability of copyright law to software. At issue in Whelan was whether a business-management system developed for a dental laboratory and written for an IBM PC computer was a copy of a predecessor system that operated on an IBM Series 1 computer. The court held that "copyright protection of computer pro- grams may extend beyond the programs' literal code to their structure, sequence, and organization." By the court's parsing of the dichotomy between idea and expression, an idea in a program is its "purpose or function"; the expression of the idea is "everything that is not neces- sary to that purpose or function." File and data structures, data flow, and the structure and sequence of screen displays that manifested program routines were construed as protectable expression. In endorsing the appellate court's ruling that "substantial similar- ity" of nonliteral elements of programs can be proof of infringement, Goldberg suggests that limiting protection only to code "would be equivalent to permitting one freely to publish a copyrighted English- language novel in an unauthorized French translation, or to dramatize it or make it into a motion picture without authorization" (Goldberg and Burleigh, 1989, p. 301~. He also dismisses arguments that the Whel- an decision confers protection on all elements of a program except its underlying idea. Goldberg notes that the court based its finding of comprehensive nonliteral similarity on its review of data structures, screen outputs, and five subroutines or modules. The court considered the possibility of "merger of idea and expression" but dismissed it, pointing to evidence cited by the district court that competing, com- mercially available programs had different structures and designs and yet had the same purpose. As evidence that the Whelan decision does not imply blanket protection to programs, Goldberg points out that a number of plaintiffs have lost "structure, sequence, and orga- nization" cases decided after Whelan. Goldberg also concurs with the court's rejection of the view that the incremental nature of progress in software development necessi- tates some copying. The court did not see any qualitative difference between progress in the development of software and progress in other areas of science and the arts where copyright law applies. Samuelson, in contrast, contends that the Whelan decision not only takes the "radical step" of regarding the overall structure of a program, which she regards as a functional design, to be protectable expression under copyright law, but more radically than that, adopts a test for software copyright infringement that would recognize everything about a program except its general function or purpose as copyrightable "expression." Samuelson points out that the appellate court quoted

OCR for page 21
BACKGROUND TO BASIC LEGAL ISSUES 29 with approval a passage from the trial court decision that described as protectable expression "the manner in which the program oper- ates, controls, and regulates the computer in receiving, assembling, calculating, retaining, correlating, and producing useful information." Such wording, she adds, would make the functionality of a program into protectable expression, ignoring copyright law's exclusion from protection of processes, procedures, methods of operation, and systems that might be described in a copyrighted work. Functional processes and designs have traditionally been within the domain of patent law, not copyright. Finally, Samuelson suggests that applying the Whelan decision's test for differentiating between idea and expression seems to have "swept" algorithms "into the fold of 'expression."' Within recent years, the U.S. Patent and Trademark Office has begun awarding patents for "method algorithms" embodied in software-related inventions, providing protection for 17 years. However, if the Whelan decision is to be taken seriously, copyright protection and its 75 years of coverage could be extended to algorithms, Samuelson maintains, "particularly if a software copyright plaintiff's lawyer is astute enough not to call the algorithm an algorithm, but rather the structural backbone of the software." To Samuelson, copyright law's aversion to technology is manifested in the way the courts have chosen to address software-related issues. Virtually all courts have treated software entirely by analogy to literary works, ignoring its status as technology. While acknowledging that certain aspects of software fall clearly within the province of copyright law, Samuelson advises that, for the law to embrace the technological aspects of software, fundamental changes in the copyright system would be required. "Judges have been blind to the fact that software is a technology and that progress in the field of technological arts may more easily be impeded by strong copyright protection than might be the case in the field of the literary arts," she maintains. TRADE SECRET A commonly used protection, trade secrets are often used in con- junction with copyright, which serves as a first line of defense in the event that a program is "reverse engineered" by a competitor. Indeed, the Copyright Office accommodates the software industry's heavy reliance on trade secrets by allowing registration with descriptive, identifying materials that do not reveal confidential information. Under state trade secret laws, firms use licensing contracts that stipulate conditions under which software can be used, copied, modi

OCR for page 21
30 INTELLECTUAL PROPERTY ISSUES IN SOFTWARE fled, translated, or transferred. Such contractual prohibitions can be circumvented, however, and enforcement problems are magnified by the growing prevalence of distributed computing, which makes licensed programs easily accessible to nonlicensed users. The reverse-engineer- ing proscriptions are not necessarily binding on nonlicensed users, a problem with potentially disastrous consequences for innovators who rely on this form of protection alone. Unlike machines and other technologies that have a material manifestation, software, as already noted, can be disassembled and reassembled into a form that bears little resemblance to the copied work, thus defeating the trade secret protection. In addition, a competitor could independently develop and then patent a software-related process that a company had protected through trade secret. Since patent law does not recognize trade se- crets as prior art, the original innovator would be forced either to stop selling the software product or to license the process from the competitor that holds the patent. Sellers of off-the-shelf software often attempt to impose agreements- "shrink-wrap licenses" upon purchasers opening the plastic wrapper that encases the disk on which the program is stored.2 Buyers are allowed to make archival copies, but self-executing shrink-wrap licens- ing agreements may prevent them from copying the program and distributing it to others. To treat programs sold in volumes of hundreds of thousands and even millions as trade secrets, according to one commentator, "offends common sense the fact that so much copying takes place indicates how few users take the agreement seriously" (Branscomb, 1988, p. 43~. Software firms are becoming increasingly concerned about the viabil- ity of contracts applied to large numbers of users. On the one hand, wide-scale distribution may render the claim of trade secret meaningless. On the other, there are fears that the "first sale doctrine" will under- mine licensing agreements. Under the doctrine, the purchaser of a copyrighted work is free to disseminate that work, although copying is limited to certain prescribed circumstances. Such circumstances include those cited in the copyright law's "fair use doctrine," which allows copying for the purposes of research and criticism. Both doctrines, some observers speculate, could serve as the means to erode the scope of trade-secret protection, as well as copyright protection. Another concern is that federal copyright law will preempt state trade secret laws. Indeed, in a closely related area, a federal appellate court in Louisiana ruled in 1988 that portions of that state's shrink-wrap en- forcement law are preempted by copyright law (Samuelson, 1988~. Uncertainty over the legitimacy of shrink-wrap licenses hangs over the entire industry, said Charles M. Geschke, president and chief

OCR for page 21
BACKGROUND To BASIC LEGAL ISSUES 31 operating officer of Adobe Systems, Inc. He warned of the potential danger that competitors could "take what I have licensed to them and use it as a weapon to come back and compete with me by [auto- mated] reverse engineering. That is very hard, from a developer's point of view, to accept as a really fair form of competition." PATENT In the 1972 case Gottschalk v. Benson, the U.S. Supreme Court ruled that mathematical algorithms were not patentable, spawning the view that software-related inventions, the essence of which is often an al- gorithm, were not eligible for patent protection. In fact, CONTU, in its analysis of potential intellectual property protections for software, shared this perception and dismissed patents as an option (Samuelson, 1988~. A year after Congress enacted CONTU's copyright recom- mendations, however, the Supreme Court opened the patent door to software, ruling in Diamond, Commissioner of Patents and Trademarks v. Diehr et al. (Supreme Court of the United States, 450 U.S. 175 [19811) that the use of an algorithm does not render unpatentable an in- vention that would otherwise be eligible for the protection (Kahin, 1989). The decision in Diamond v. Diehr, which concerned the patentabil- ity of a rubber-curing process directed with the aid of a computer program, was hardly an unequivocal endorsement of the patenting of software. Some commentators suggest that the wording in the 5 to 4 decision allows the court sufficient room to reassess the issue in the future. But in the nearly 10 years since Diamond v. Diehr, the high court has chosen not to revisit the matter. Meanwhile, the number of software- related patents awarded by the Patent and Trademark Office rose from none in 1980 to about 200 annually in recent years, according to estimates prepared for the Computer Law Committee of the State Bar of Texas (cited in Brian Kahin, "The Case Against Software Patents," unpublished paper, 1989, p. 1~. As is true for the copyrightability of software, the patentability of software-related inventions may be viewed as a boon or a bane, depend- ing on one's perspective. A simplified pro-patent argument holds that the protection is most appropriate for addressing the technolog- ical aspects of software, while providing a powerful incentive for innovation. The contrary argument views patents as anticompetitive because, unlike copyrights, they do not allow for independent inven- tion, and as increasing the risk of litigation because of the secrecy of the patent-approval process.

OCR for page 21
32 INTELLECTUAL PROPERTY ISSUES IN SOFTWARE Characteristics of Patents A patent provides 17 years of protection to owners of inventions- works that are nonobvious, novel, and useful in exchange for full disclosure of the inventions at the time of application. Because it pre- cludes independent invention, patent law is said to provide "thicker" protection than does copyright law, forcing competitors to "invent their way around" a patented innovation. Donald S. Chisum, professor of law at the University of Washing- ton, maintained that, compared to copyrights, patents are a "nobler" form of protection. By "dangling out a reward, the right to exclude others from using useful processes and products," he said, the first- to-file requirement and its winner-take-all result hasten the pace of in- novation. Even companies with large market shares cannot be content to rest on past accomplishments because, according to Chisum, they run the risk of being "completely knocked out of the market by a techno- logical innovation that is developed and patented" by a competitor. Moreover, Me disclosure requirement provides competitors with a clear un- derstanding of the inventive hurdle they must leap to compete with the patent holder, thereby setting the stage for the next round of innovation. Unlike the protection provided by copyright law, patent protection extends to functionality. Different implementations of an idea or incre- mental improvements to a patented invention are not eligible for protec- tion because they do not satisfy the patent law's requirement for nonobviousness. Indeed, such implementations or improvements, if undertaken by a competitor, constitute infringement of the patented innovation. In theory, the monopoly grant awarded to the patent hold- er fosters more efficient development within industry by discouraging duplication of effort. The scope of patent protection is subject to uncertainty, however, in the guise of the doctrine of equivalents, patent law's counterpart to copyright's doctrine of substantial similarity. Determining whether two inventions are substantially similar is a matter for judicial inter- pretation, and the reasoning applied in the few software-related cases that have reached the courts thus far has not always been consistent. "Courts are essentially schizophrenic about patents," Chisum ex- plained. "On the one hand, we say that it is the function of the patent claim to delineate what is covered. If you don't claim it, you don't have any exclusive rights over it, and yet on the other hand, they [the courts] apply . . . the doctrine of equivalents. In some cir- cumstances, something that literally does not conform to the verbal statement in the patent is nevertheless found to be an infringement, and as you can imagine, that is a great source of ambiguity." Added Ronald Laurie, of Irell & Manella, "I think we are going to

OCR for page 21
BACKGROUND TO BASIC LEGAL ISSUES 33 be hearing a lot more about this area [doctrine of equivalents] with the rising tide of software patents that are being issued and, of course, the consequent flood of software-patent litigation." Proponents of patent protection for software suggest that some of the issues now being contested in copyright cases stem from the pre- sumed unavailability of the protection following the Supreme Court decision in GottschaZk v. Benson. People have tried to "stretch the boundaries of copyright protection to fill in for the fact that no one [had] patent protection," said John Shoch, general partner at the As- set Management Co. He cited the current spate of "look and feel" cases as symptomatic of the problems that arise when copyright law is used to address issues that belong in the domain of patent law. "EWle find ourselves in this tremendous fight about look and feel, and function, and interfaces in the copyright domain, which is abso- lutely the wrong place to have much of this fight," Shoch contended. The proper legal arena for these issue, he added, is patent law. In his view, the software industry will benefit if patent protection is made fully available, alongside copyright and trade secret protection. Not everybody sees the emergence or reemergence of patent protection as a positive development. Brian Kahin, adjunct research fellow at Harvard University, argued that the late arrival of patent protection is potentially disruptive, suggesting that the virtual ab- sence of patenting until recent years may undermine the highly decentral- ized structure of the software industry. "If software had clearly been protectable from the outset, there would be no surprise, no defeated expectations," Kahin has written (Brian Kahin, "The Case Against Software Patents," unpublished pa- per, 1989, pp. Sol. "But now it appears that the industry may have to be reshaped to fit the patent system and that the rapid develop- ment of software products may have to be slowed to fit the review, processing, and publication cycle of the Patent and Trademark Office." Yet, the nature of software is such that many of its components satisfy patent law's eligibility criteria, Laurie maintained, noting the interchangeability of function between software and hardware. "If you say that hardware is patentable and software isn't," he said, "then you are saying to developers and to engineers that depending . . . on how they resolve the design trade-off, they may get protection or they won't get protection" even though they are solving identical problems in virtually identical ways. "That is injecting, in my opinion, too many legal considerations." Adds IBM vice president for sys- tems and programming Peter Schneider, "The most disturbing part about patent protection as it relates to software is that it is an indica- tor of how fast the legal community changes its mind."

OCR for page 21
34 INTELLECTUAL PROPER7'Y ISSUES IN SOFTWARE What Is Patentable? What is eligible for patent protection? According to the U.S. Su- preme Court decision in Diamond, Commissioner of Patents and Trade- marks v. Chakrabarty (447 U.S. 303, 100 S. Ct. 2204 [19801), Congress intended patent law to embrace "anything under the sun that is made by man" anything, that is, falling into any one of the four categories of statutory subject matter: process, machine, manufacture, or composi- tion of matter. Among the items not sheltered by the broad umbrella of patentable subject matter are laws of nature, physical phenomena) abstract ideas, "mental steps," and methods of doing business. Software clearly qualifies as a human-made artifact. But the technology's critical dependency on mathematical algorithms, which patent law views as akin to laws of nature, positions software right at the drip line of the umbrella of patentable subject matter. Thus the Supreme Court ruling in Gottschalk v. Benson, holding that "an algo- rithm, or mathematical formula, is like a law of nature, which cannot be the subject of a patent," was widely interpreted as excluding software from patent protection. In Diamond v. Diehr, the high court did not reverse its 1972 decision, but rather it said the mere presence of an algorithm did not automatically render a process or machine ineligible for a patent. It did not say categorically, however, that computer programs are patentable. Nevertheless, the highest specialized court in the domain of patent law, the Court of Customs and Patent Appeals (CCPA; the predecessor to the Court of Appeals for the Federal Cir- cuit), "has held that computer processes are statutory unless they fall within a judicially determined exception" (U.S. Patent and Trademark Office, 1989, p. 61. Using the somewhat equivocal guidance provided by the courts, the Patent and Trademark Office has developed an operational concept of what software elements are eligible for patents, giving rise to such terms as software-related inventions, computer processes, and computer algorithms a classification distinct from unpatentable mathematical algorithms. The upshot, according to Michael S. Keplinger, an attor- ney advisor in the Office of Legislation and International Affairs at the Patent and Trademark Office, is that patents are not awarded for computer software. '~We grant patents," he explained, "on computer pro- cesses, processes that may be implemented in a computer, just as they might be implemented in a production-line machine or in any other hardware embodiment." Abstract descriptions of computer processes do not qualify, Keplinger added. Only specific implemen- tations of processes in hardware are eligible. Nor does the Patent and Trademark Office's parsing render math

OCR for page 21
BACKGROUND TO BASIC LEGAL ISSUES 35 ematical algorithms patentable. "We used to say that we didn't issue patents on algorithms," Keplinger explained, "but you can't say that, because any process is an algorithm. It is a difficult line to [draw]- for a patent examiner to determine where you draw the dividing line between a purely mathematical algorithm, which we won't issue a patent for, and a process that may be expressed in mathematical terms," which will receive a patent if it is sufficiently useful, novel, and nonobvious. To help inventors and their attorneys, the Patent and Trademark Office recently published a legal analysis that describes a two-part test to determine whether a product or process containing an algorithm is eligible for patent protection. The test is the evolutionary product of decisions made by the Supreme Court and the CCPA (U.S. Patent and Trademark Office, 1989~. Test for Patentability The first step, determining the presence of a "mathematical algo- rithm," is relatively straightforward, although, on occasion, ascertaining whether an algorithm is included in a claim may require some inter- pretation. The description of a mathematical algorithm, defined as a procedure for solving a given type of mathematical problem, is in- consequential. It may be in the form of a mathematical equation or it may be described in prose. More significant is the nature of the claim, whether it is for a process or a machine. By themselves, algo- rithms are considered processes that are ineligible for patents. To enhance their chances for securing a patent, some innovators have filed machine, or apparatus, claims that typically describe an algo- rithm as a means for accomplishing a task. The label is often not convincing. The applicant must demonstrate that functions described in the claim can be performed with a specific apparatus only. Most applicants fail to prove this specificity, and so most machine claims embodying algorithms are treated as process claims. The second step in the test, distinguishing "between patentable process and unpatentable principle," is not free of ambiguity, as the Supreme Court noted in the 1978 case Parker v. Flook (437 U.S. 584, 593; 198 U.S.P.Q. 193, 198-99 [19781~. Lacking definitive tests for making the distinction, the Patent and Trademark Office's analysis offers sever- al guidelines that it "synthesized" from court decisions. Apparently key to the final determination, however, is the specificity with which an algorithm is applied to steps in a physical process or, in the case of a machine claim, to physical elements. This implied penchant for specificity stems from the desire to prevent preempting the use of an

OCR for page 21
36 INTELLECTUAL PROPERTY ISSUES IN SOFTWARE algorithm in other applications. That is precisely the result the Su- preme Court sought to avoid in GottschaZk v. Benson and in Parker v. FZook, noting in the former case that "the practical effect" of excluding broad areas of use "would be a patent on the algorithm itself." Especially influential here is a CCPA decision in an appeal of a claim denied by the Patent and Trademark Office, Lever Brothers Company v. Thrift-D-Lux Cleaners, Inc. (U.S. Court of Customs and Patent Appeals, 46 CCPA 798, 263 F.2d 842 [19591~. In In re Abele (684 F.2d 902, 214 U.S.P.Q. (BNA) 682 [CCPA 19821), the CCPA suggested re- viewing a claim without the algorithm. If the remaining process or machine qualifies as patentable subject matter, according to the court, then inclusion of an algorithm should not alter that determination. This procedure does not suggest, the Patent and Trademark Office stresses, that the inventive merits of the claim be assessed in this manner, since the entire process or machine must be evaluated to determine novelty and nonobviousness. A computer process is not patentable if the claim merely lists poten- tial applications. The Patent and Trademark Office, referring to the decision in Parker v. FZook, classifies such a listing as "insignificant or non-essential post-solution activity." Similarly, a process is not eligible if the end product is simply the recording of the results of a calcula- tion, nor is a process that is confined to assembling data for assigning values to variables in an algorithm. If, however, data gathering is dictated by steps other than the algorithm, then the process may be eligible. Less ambiguous are processes that transform "something physical into a different form," as compared with those that, by means of computation, transform one set of numbers into a different set. Thus the court has ruled as patentable a process that transforms "spherical seismic signals" into a "form representing the earth's response to cylindrical or plane waves" (Mehmet Turhan Taner, FuZton KoehZer, NigeZ A. Anstey and Michael J. CasteZberg, U.S. Court of Customs and Patent Appeals, 681 F.2d 787 [19821~. Such processes have been deemed analogous to the conversion of sound into electrical signals or other transformations of signals accomplished with electrical circuitry. Once one leaves the fairly well defined domain of signal transfor- mation and other discretely defined processes, uncertainty begins to intrude. For example, the Patent and Trademark Office notes that claims can be rejected as nonstatutory if the process represents busi- ness methods and mental steps, and "not a true computer process." Amplifying this point, the analysis cites a 1977 appeal (Gaetan de Coye de CasteZet, U.S. Court of Customs and Patent Appeals, 562 F.2d 1236 [19771~: "Claims to nonstatutory processes do not automatically and invariably become patentable upon incorporation of reference to ap

OCR for page 21
BACKGROUND TO BASIC LEGAL ISSUES 37 paratus." But closer examination may determine such machine- implemented processes to be eligible. Indeed, the analysis goes on to say that several cases decided during the 1970s indicate that "ma- chine or computer implementation of mental steps is statutory sub- ject matter." Thus the CCPA deemed a computer process translating natural languages as patentable. Indeed, a substantial number of people in the software industry would probably concur with that . OplnlOn. Grounds for greater debate may be a federal district court's deci- sion regarding the patentability of a cash management system. If carried out manually, the system would not be eligible, but in its incarnation as a computer process, the system was found to qualify ror prc'~ec~on. ~ ne court concluded that the process was not a busi ~ ~ ~ L L ~1 . ~ ~ ~ . ~ . . ~ ness method but rather a "method of operation on a computer to effectuate a business activity" (Paine, Webber v. Merrill Lynch, 564 F. Supp. 1369, 218 U.S.P.Q. 220 [19831~. This distinction has mystified some in the industry. Coherent or Incoherent? Meanwhile, despite the great effort expended by the courts and the Patent and Trademark Office to elucidate the nature of computer processes, some legal experts believe that the system can be manipulat- ed to secure patent protection for mathematical algorithms. "As long as you don't call it an algorithm," said Chisum, "it is probably going to be patentable as a method, even though it really is an algorithm." The Supreme Court decision in Gottschalk v. Benson, he maintained, remains as a source of lingering confusion and incoherence, making such behavior necessary. To avoid invoking that decision, Chisum explained, patent attorneys "dress up" algorithms "with language that makes it into a method or process." Mitchell D. Kapor of ON Technology, Inc., questioned the "intel- lectual coherence" of the patent system as characterized by Chisum. "[OIne's business strategy may hinge on legal subtleties that are appar- ently disconnected from anything that makes sense to us," he said. Another technical expert who fails to see conceptual clarity in the law's and the Patent and Trademark Office's treatment of algorithms is Allen Newell, a professor of computer science at Carnegie Mellon University. In an oft-cited article, Newell maintains' "The models we have for understanding the entire arena of the patentability of algo- rithms are inadequate not just somewhat inadequate, but funda- mentally so. They are broken" (Newell' 1986~. Many of Newell's numerous criticisms center on the practical utility of most discover

OCR for page 21
38 INTELLECTUAL PROPERTY ISSUES IN SOFTWARE ies of computer science, an attribute that distinguishes the field from other research areas. "With rare exceptions," he writes, "scientific knowledge in computer science is in the form of means-end relation- ships- what to do to obtain something of value. Indeed this is just the essence of algorithms: what to do to perform a task. But algorithms, far from being an applied part of computer science, are at the center of its basic theoretical structure." Therein lies a problem. It is plausible, Newell maintains, that for any process a limited number of reasonably efficient algorithms may exist. To grant monopoly over these efficient processes would impose a "stranglehold" on additive inventive activity that would otherwise build on important algorithms. In some technological domains, New- ell speculates, all inventive activity may involve algorithms, and the entire realm of invention "would come to reside in a computer." "If methods and processes over large technological domains be- come an exercise in algorithms," Newell explains, "then it is extraor- dinarily dangerous not to patent algorithms." Therefore, patents for algorithms may indeed be necessary, but that determination cannot be made with existing legal models of algorithms, he argues. At the forum, Michael 0. Rabin, a professor of mathematics and computer science at Harvard University and Hebrew University, strongly endorsed patent protection for mathematical algorithms because of their increasingly influential role in technology development. Algo- rithms, he said, are human inventions, not discoveries, which are unpatentable. Incentives and, therefore, protection are required to encourage people to aspire to tackle important problems whose solu- tions require new algorithms. If algorithms are viewed from this perspective, Rabin said, patent criteria that necessitate linking an algorithm to a specific process or a specific device will be found to be too confining. "I have a feeling," he said, "that this is going to be too narrow," reflecting, perhaps, the time lag between technological advance and legal response. "We do understand devices, linkages, values, and differentials, and so on, but we are moving to the edge of information and to the edge where the tools . . . are mathematical," he continued. "Yes . . ., they are ephemeral, but they are all powerful, and I think these innovations . . .. these tools, deserve at least as much protection as . . . the protection afforded to the cap on a soft-drink bottle, and so, this is my proposi- tion and my plea." THE INTERNATIONAL SITUATION With the strong urging of the United States, many foreign nations- more than 40 as of late 1989 have adopted copyright or roughly

OCR for page 21
BACKGROUND TO BASIC LEGAL ISSUES 39 equivalent protection for software. Although a smaller number of countries, including Japan, also allow patenting of software, there is growing international consensus that copyright should be the pri- mary means of protection and that software should be treated like other copyrighted works, according to Keplinger, of the U.S. Patent and Trademark Office. Two treaties, the Paris Convention for the Protection of Industrial Property and the Berne Convention for the Protection of Literary and Artistic Property, serve as the basis for international treatment of copyrighted and patented works and are administered primarily by the UN's World Intellectual Property Organization (WIPO). Most relevant to software is the Berne Convention, which the United States signed in 1988, thereby supplanting most, but not all, of the bilateral copyright agreements it had negotiated with other nations. Signato- ries to the Berne Convention agree to meet a minimum set of legal requirements and, more important, to afford "national treatment" to works produced by authors in member nations. Thus French authors of computer programs imported into the United States, for example, are awarded the same exclusive rights of ownership as U.S. software publishers who market their programs domestically. Conversely, U.S. software exported to France would be accorded full copyright protec- tion under that nation's laws. Keplinger noted that the international conventions do not explicitly mention software. Rather, protection for the technology stems from treaty wording that applies comprehensively to works of authorship. Although a WIPO initiative to develop a model law specific to the protection of software has not advanced, Keplinger said, one can "make a credible argument that the Berne Convention already requires its members to protect computer programs under copyright law, because they are generally regarded as protectable subject matter by countries that have addressed the problem." (See Box 2.2.) Nevertheless, copyright laws do vary among nations, creating some uncertainty. Keplinger explained that a German Supreme Court ruling seems to require "a relatively high degree of originality for computer programs, with the result that some German lawyers feel that many programs will be ineligible for protection under this high standard. . . ." France has chosen to classify software as an applied art rather than a literary art-and provides protection for a term of 25 years, compared with 50 years for most other treaty nations. However, the European Economic Community (EEC), of which France is a member, is preparing a directive on intellectual property protection for software that may (based on the content of drafts) propose full copyright cover- age for software, including protection for 50 years.

OCR for page 21
40 INTELLECTUAL PROPERTY ISSUES IN SOFTWARE BOX 2.2 U N FAI R COMPETITION LAW In many countries, slavish imitation of intangible creations is still regarded~vertly or covertly as a wrongful act in its own right, ac- cording to Jerome H. Reichman of Vanderbilt University. This contro- versial business tort, known as misappropriation, is widely recognized in state unfair competition laws, but these laws often conflict with federal intellectual property protections. As a result, the U.S. Supreme Court has on several occasions reduced the scope of state laws. Recently, Switzerland enacted a new misappropriation law deliberately aimed at preventing theft of new technology. This use of unfair competition law in this capacity has greatly influenced foreign intellectual property law, Reichman said, but so far, it has attracted little attention in the United States. Reichman suggested that it might be easier and more effective to add an antipiracy clause to the Paris Convention (to which more than 90 countries adhere) than to "stuff" computer programs into the Berne Convention, which governs I iterary and artistic works. (See text.) Th is would be feasible, he said, if a GArr agreement on intellectual property ultimately strengthened international arrangements covered by the Paris Convention, wh ich covers industrial property. The proposed EEC directive, which specifies how copyright pro- tection applies to software, contains a controversial measure that, opponents contend, would "drastically limit rivals' ability to decipher software interfaces and build compatible products" (Verity, 1990, pp. 138 and 140~. Proponents argue that the measure would be an effec- tive deterrent to "commercial copying" of software. In December 1990, the Council of Ministers of the European Com- munity adopted a "Common Position" on the Directive on the Legal Protection of Computer Programs, which permits more extensive re- verse analysis than earlier drafts of the directive. The Common Posi- tion would permit Recompilation to the extent necessary to develop competitive but themselves noninfringing interoperable products. A final directive is expected to be adopted by late summer or early fall, 1991, after it is reviewed again by the European Parliament, the Council of Ministers, and possibly the Commission of the European Community. Meanwhile, Japan has excluded algorithms, rules, and program- ming languages from copyright protection, a revision that observers inside and outside of Japan speculate will abet widespread copying. A 1989 decision by the Japanese high court, the first ruling under the amended copyright law, found that copying a program's "processing

OCR for page 21
BACKGROUND TO BA`SIC LEGAL ISSUES 41 flow" the court's interpretation of an algorithm~id not constitute infringement, according to Dennis S. Karjala, director of Arizona State University's Center for the Study of Law, Science, and Technology. Compared with the ruling in the U.S. case of Whelan v. lasiow, how- ever, the Japanese court's decision "did not go nearly as far in pro- tecting nonliteral features of programs," Karjala said. To assure reasonable levels of protection for software and other forms of intellectual property worldwide, the United States and other nations have proposed including an intellectual property code in the General Agreement on Tariffs and Trade (GATT), which has about 100 nations as signatories. Incorporated into this proposal are clarifi- cations of the rights accorded to owners of software under copyright law, as well as mechanisms for enforcement, which are lacking in the Berne Convention. A benefit of addressing intellectual property issues through the GAIT, Keplinger said, is the treaty's dispute-resolution process, which entails convening a panel to determine whether a member state is living up to its obligations. In contrast, disputes that arise under the Paris and Berne Conventions are referred to the World Court, which does not have enforcement powers. While Keplinger said that inconsistencies and other deficiencies in international con- ventions for protecting intellectual property need to be improved, he strongly advised against measures that would deviate from interna- tional norms of copyright and patent protection. Abandoning these norms, rather than "fine-tuning" them, he warned, would jeopardize past progress in "getting meaningful levels of protection for our works and our technology abroad." NOTES 1. Lotus Development Corporation v. Paperback Software International and Stephenson Software, Limited (Civil Action No. 87-76-K, U.S. District Court for the District of Mas- sachusetts, 740 F. Supp. 37 Dune 28, 1990]), for example, addressed this issue in con- siderable detail. In October 1990, Paperback agreed not to appeal the decision (Keefe, 1991). 2. The legal status of shrink-wrap licenses is quite uncertain. One school of legal analysts believes they are clearly unenforceable without the consent of the buyer. Others believe this is still an open question.

OCR for page 21
What is it that we want to protect? First . . . is the brilliant inven- tion, the idea, the notion that makes a new product and the in- sight that makes a whole new industry.... [T]he second thing we want to protect is the investment and the hard work. This is the grunt work. This is the pick-and-shovel engineering that turns the idea, the prototype, into a reliable, distributable, maintain- able, documented, supportable product. -Robert Spinrad, Director, Corporate Technology, Xerox Corp. Since the Industrial Revolution we have had copyright to protect printed works and patent to protect machines made out of iron and steel. The problem is that software is really some of both. It is the first technology that has content. Mitchell D. Kapor, Chairman, ON Technology, Inc. We have managed to lay down the flooring, so we don't have to stand in the mud. But the problem is we are trying to build cathedrals and the ceiling is still a ways up there.... Every time the floor comes up a little bit, the ceiling shoots up higher be- cause aspirations keep getting higher. Randall Davis, Associate Director, Artificial Intelligence Laboratory, Massachusetts Institute of Technology We have some wonderful arrogance in thinking that the world of software is so much different from mechanical engineering or the new biological and pharmaceutical disciplines. I, in fact, don't believe that is the case. John F. Shoch, General Partner, Asset Management Co.