Should Reverse Engineering of Computer Software Through Intermediate Copying Be Prohibited? Lorrie Ackerman Engineering and Policy Graduate Student Washington University LFA1@CEC1.WUSTL.EDU 24 July 1992 Prepared as part of the 1992 Washington Internships for Students of Engineering (WISE) program under the supervision of Dr. Herman Hill, Jr., Professor of Electrical & Computer Engineering at Ohio University and the 1992 WISE faculty advisor. This project was sponsored by the Institute of Electrical and Electronics Engineers (IEEE). Unless otherwise noted, all opinions presented are those of the author and do not in any way represent those of IEEE or other individuals or institutions referred to in the text. ACKNOWLEDGEMENTS I would like to thank Scott Grayson at IEEE-USA for his advice and assistance. I also appreciate the comments of all those who read previous drafts of this paper: Jim Burger, Stuart Bjornsson, Chris Byrne, James Chandler, Butch Hill, Victor Siber, Pamela Samuelson, and Richard Stallman. INTRODUCTION The U.S. software industry has grown rapidly over the past 25 years. In the mid 1960s estimated annual revenues for computer programs sold in the United States totalled $250 million; in 1991 revenues exceeded $20 billion (National Research Council 1991, vii; U.S. Department of Commerce 1992, 27-23). Companies that deal in both hardware and software report that their software revenues are generally outpacing their hardware revenues. This high rate of growth and innovation is expected to continue to keep up with changing hardware platforms and information processing demands. The National Research Council suggested, "Those who master the challenge of creating good software can expect to be the leaders in the world of information machines and services," (1991, vii-viii). Despite the success of the U.S. software industry, some experts say that the results of the current controversy over reverse engineering of software -- examining software developed by other people to determine its underlying ideas -- could severely hamper innovation. Several recent law suits have questioned the legality of some techniques for reverse engineering computer software. The reverse engineering question may eventually be resolved by the courts. However, some lawyers and computer professionals maintain that legislative action must be taken to clarify copyright law as it applies to computer software. Reverse engineering refers to the examination of a system to gain insight into how it was built. This is a common practice in all engineering fields. Reverse engineering of computer software includes non-controversial activities such as studying a program's manuals and observing a program's behavior, and controversial activities which require making a copy of a program under study. This type of copying is known as intermediate copying because the copy itself is never sold. Intermediate copying is required when translating a program from a language understood by computers to a language more easily understood by humans. A central question of this paper is: should it be an infringement of copyright to make unauthorized intermediate copies of a copyrighted program in order to analyze it? Software developers often use reverse engineering techniques to better understand systems with which their software will interact. In addition, developers use reverse engineering to learn the ideas behind other developers' successful techniques. "`[R]everse engineering' in any field of engineering simply means analyzing a machine that you did not build, to figure out how it was made. This is a lawful and accepted practice in all fields of engineering," explained Richard Stallman, founder of the League for Programming Freedom (1992). Computer programs can be written in a variety of languages including machine language, assembly language, or any of the many high-level languages such as Pascal, Fortran, C, or BASIC. The source code (original, untranslated code) for most programs is written in a high-level language or assembly language because these languages are most easily understood by humans. Machine language (also called object code), which is composed of sequences of 0s and 1s, is difficult for humans to understand. However, machine language is the format that all computer commands must be translated into before a computer can execute them. Assembly language differs from high-level languages in that it is machine dependent and generally results in significantly longer source code programs but shorter object code programs. Most commercial software packages are distributed to customers in machine language. Therefore, humans must disassemble programs, translating them from machine language to assembly language, in order to better understand the ideas on which they are based. It is not possible to reconstruct the original sequence of a program or the programmers' comments and annotations which are usually a part of high-level language programs, but through disassembly, programmers can gain insight into the way a program functions and how it can interface with other software and hardware (U.S. Congress 1992, 7). Programmers must take information gleaned through disassembly and add to it their own ideas for disassembly to be useful. Sometimes programmers translate disassembled code into a high-level language. This process is often referred to as "decompilation." However, some people maintain that decompilation is technically a mythical process because it is not possible for a programmer to obtain an exact copy of a program's source code by decompiling object code (Johnson-Laird 1992b, 26). Intermediate copying has recently become a controversial topic and is being argued in courtrooms throughout the United States. However, much of the debate is being done by lawyers and intellectual property analysts -- not the computer professionals who are affected by intermediate copying rules (Samuelson, Denber, and Glushko 1992). Participants in the intermediate copying debate have been closely watching the case brought by Sega Enterprises Ltd. against Accolade Inc. Accolade disassembled some of Sega's game cartridges to learn how to make Accolade games compatible with Sega's Genesis game machine. Sega alleges that this method of reverse engineering is an infringement of copyright (BNA 1992a, 545). On April 3, 1992 the U.S. Court for the Northern District of California issued a preliminary injunction against Accolade and ruled that "reverse engineering software by disassembly can be infringing if it entails the duplication of the copyrighted work and the recasting or transformation of the object code into a form more intelligible to humans....the copyright owner need not prove substantial similarity to the defendant's final product, and that intermediate copies can provide a basis for copyright infringement," (BNA 1992a, 545). The U.S. Court of Appeals Ninth Circuit will consider Accolade's appeal. A decision is expected in October 1992. Critics of the court's ruling, including patent attorney and legal commentator Richard H. Stern, argue that "there is ample precedent for the opposing rule that an accused work infringes a copyrighted work only if the accused work contains a substantial amount of protected expression," (1992, 3). Further Stern criticized the court for suggesting that Accolade could have "peeled" the microchips -- a process by which the inside of microchips are inspected -- to legally reverse engineer Sega's code. Stern (1992, 5) explained, "The court simply has no idea of how you reverse-engineer software, or what an engineer does when peeling a chip." The U.S. District Court for the District of New Mexico took a different approach to deciding a software intellectual property rights case. On April 21, 1992 the court adopted the "filtering" approach when it tested a reading skills program for substantial similarity to a copyrighted program in the case of Autoskill Inc. v. National Educational Support Systems Inc. "Rather than looking at a program as a whole...this approach requires that all non-protectable elements of the work, such as ideas and any expression that has merged with an idea, be filtered out of the substantial similarity analysis," (BNA 1992b, 122). Although intermediate copying was not a factor, this case is still noteworthy because it places restrictions on how copyright laws can be applied to software. On June 22, 1992 the Second Circuit Court, in upholding a lower court ruling in a case brought by Computer Associates International Inc. against Altai Inc., restricted the use of copyrights to limit competition in the software industry. The court concluded that copyright does not protect the "behavior" of a program or subsections fundamental to producing that behavior. Lawyers on both sides of the issue agreed that the ruling would have far reaching affects, but disagreed on whether it would promote or hinder innovation in the industry (Markoff 1992, D1, D5). While the American legal system continues to wrestle with the reverse engineering issue, the European Community (EC) adopted a Council Directive on May 14, 1991 that explicitly outlines the legal framework for protecting the intellectual property rights of software developers. Article 6 of the EC directive allows for unauthorized "decompilation" of computer code, provided it is necessary to achieve interoperability with another program and a specified set of conditions are met (Official Journal of the European Community 1991, 45). There is a great deal of debate over this directive in both Europe and the United States. The interoperability exception is seen by some as an essential compromise and by others as an inherent flaw. This paper identifies actions policy makers should take to ensure that laws governing intermediate copying are in the best interest of the American computer industry and the public, regardless of the courts' interpretations of current copyright law. This paper does not assess the legality of reverse engineering or predict actions of the court. Legal scholars have devoted much time to analyzing and arguing the legal issues; however, these analyses are interpretations of current policy rather than proactive statements on what policy should be. Stallman (1992) explained, "[T]he decision is not ultimately in their [the court's] hands, but in ours. Citizens of a democracy should always consider what they think the government should do; if they start instead to consider what it `will' do, they are effectively abdicating their role in the decision." KEY CONFLICTS AND CONCERNS The intermediate copying debate centers around the question of how best to promote innovation and maintain the U.S. software industry's competitive edge internationally. Proponents of intermediate copying -- including many small software companies, manufacturers of interoperable computer systems, and the United States Activities' arm of the Institute of Electrical and Electronics Engineers (IEEE-USA) -- maintain that a restrictive interpretation of copyright law will lead to less competition and, in turn, less innovation. Opponents -- including many large computer manufacturers and software companies -- say a restrictive interpretation is necessary to give companies the financial incentive to innovate. Part of the intermediate copying controversy is caused by differing perceptions of the status quo. The degree of impact of any law depends on the frequency of the activities which the law protects or prohibits. Opinions on how much intermediate copying is currently done by U.S. software manufacturers vary. Some companies deny ever disassembling other companies' code, while others claim to do it occasionally or even frequently. "There are programmers who never disassemble anything. There are programmers who spend all their lives disassembling," said Stuart Bjornsson (1992) an independent software developer, hardware manufacturer, and a member of the IEEE-USA Intellectual Property Committee. Further, some people maintain that the current legal climate -- in which the software industry has grown successfully -- prohibits making intermediate copies of software; other people maintain that making intermediate copies has always been legal. "The computer software industry is thriving in the absence of a 'right' to decompile," stated the Computer and Business Equipment Manufacturers Association (CBEMA) and the International Anticounterfeiting Coalition, Inc. (IACC) in an amicus curiae (friend of the court) brief submitted to the Ninth Circuit Court of Appeals (CBEMA and IACC, et al. 1992, 7). However, law professor Pamela Samuelson (1990) wrote, "Many might say that reverse engineering of software has contributed to rapid growth in the software industry and to the increased competition in product development and refinement than would have occurred in an environment in which software reverse engineering was illegal." PROTECTION UNDER COPYRIGHT LAW Copyright law was designed to protect expressions of ideas, not the ideas themselves. It was meant as a contract between copyright holders and the public: copyright holders are guaranteed "a 75-year limited monopoly in exchange for [giving the public] the opportunity to read and be inspired by the work and a free license to copy all of the ideas and processes embodied in it," (Copyright Law Professors 1992, 13). The intent was to give people the incentive to both innovate and share their innovations for the benefit of the American public. The distinction between expressions and ideas is relatively straight forward and noncontroversial for literary works. However, in functional works such as computer programs, the distinction is more controversial. Supporters of intermediate copying say that when a company withholds human-readable code from its customers it is using technology to prevent them from examining the ideas behind the copyrighted material. Opponents counter that code obtained using intermediate copying techniques does not contain any more ideas than machine-language code does. In addition, opponents argue, due to the nature of computer code, it is not possible to protect the code once the examination of human-readable code is permitted. An amicus curiae brief, submitted to the Ninth Circuit Court of Appeals by the Copyright Law Professors (1992, 14), explained why reverse engineering is necessary to obtain a program's ideas. All other copyright-protected works carry their ideas and other unprotected elements on their face. Anyone is free to use these unprotected elements without the permission of the author, even in works that directly compete....Computer programs in object-code form are alone, among all types of publicly distributed copyright-protected works, in being unreadable by human beings." CBEMA and IACC countered, "It is possible to read object code....Indeed prior to the advent of assemblers and compilers in the early 50's, all programming was done in machine language," (1992, 18). In addition, they stated, "The argument that it is necessary to copy and adapt the object code version of a copyrighted computer program to understand its 'ideas' is without merit." A software developer can analyze a program using a variety of other means including reading manuals, observing a program's operation and screen displays, and physically examining "internal parts of the computer while the program is running," (1992, 18-19). However, Andrew Johnson-Laird (1992b, 11), president of a consulting company specializing in the forensic analysis of software and plagiarism assessment, explained that reading and analyzing object code on a computer screen is so time consuming that it is "completely impractical" given the size of most programs on the market today. Under U.S. copyright law, displaying code on a computer screen is not considered copying, while printing out code is. Therefore, in order to analyze the object code of a typical program without making an intermediate copy, one would have to examine over 1000 pages of 1s and 0s without printing them out or taking notes. He added that saying intermediate copying is unnecessary is comparable to "suggesting that as nail scissors can cut a blade of grass, that all lawn mowers should be abolished and scissors used instead." In addition, supporters of intermediate copying have drawn other analogies to explain why the copyright bargain, when applied to computer software, should allow for intermediate copying. The Copyright Law Professors (1992, 27) explained, "One who uses unprotected elements extracted from disassembled code to build a competing, noninfringing program is in principle doing nothing different from one who creates a competing book by extracting creative ideas from another's book and expressing them in noninfringing language." Norton Cutler, Assistant Chief Counsel for NCR, a manufacturer of personal and business computers added, "Imagine how the car business would work if you had to sign a contract every time you bought a car saying you wouldn't look under the hood," (Cutler 1992). In an appellant's reply brief, Accolade, Inc. (1992, 15-16) stated, "Even James Joyce could not have precluded purchasers of Ulysses from reading it, studying its structure, and writing notes in the margin to aid their analysis....as long as the works subsequently created by the other author were not substantially similar to Joyce's." Opponents argue that intermediate copying supporters are using the wrong analogies. Jim Burger (1992), director, government law for Apple Computer Inc., explained that "decompilation doesn't get you any closer to the ideas -- it merely gets you the expression in a different form. The analogies to physical forms of true reverse engineering are simply inappropriate and fail to advance the debate." CBEMA and IACC (1992, 5) suggested that "decompiling" a computer program is "little different from converting an English novel to one in French or Russian." Intermediate copying supporters point out, however, that it is legal to translate a novel if the translation is not published. CBEMA and IACC also reminded the court that there is no requirement "that an author make available more information than is apparent on the face of the published work." The groups explained, "An author can be obscure, oblique or deeply allegorical. Others have no entitlement to the author's notes or rough drafts. Nor can the author be compelled to elaborate on the ideas the author has chosen to express in a cryptic fashion," (CBEMA 1992, 41). Burger added, "When you read a book you only interpret the expression that the author decided you should read. You don't have the right to go to the author and...say 'I don't understand this book, therefore I am free to make unlimited copies so that I can understand the ideas in your original work.'" Intermediate copying supporters point out that the right to make intermediate copies should not be confused with the right to make unlimited copies. In addition, they say, allowing intermediate copying does not force the authors of programs to disclose anything; however, prohibiting intermediate copying forbids other people from trying to figure out what authors do not disclose. Those opposed to intermediate copying maintain that when intermediate copying is permitted, expression, protected under copyright law, is not suitably protected. Howard Figueroa, vice president of commercial and industry relations for IBM Corporation, explained that reverse engineering "facilitates the low-cost adaptation of the protected expression in the original program, resulting in a quick and cheap generation of a competing program," (National Research Council 1991, 78). Victor Siber, senior corporate counsel for IBM Corporation (1992) agreed, "When a competitor can reverse assemble someone else's product it is the first step towards essentially ripping off the product....I think it would be a terrible disincentive for people to create new programs." Burger (1992) added that a software engineer can completely "decompile" a program in one day, using an automated decompilation program. The decompiled program, which is 80 to 90 percent as optimal as the original program, can be automatically modified to disguise its similarity to the original program, he said. Charles M. Geschke, president and chief operating officer of Adobe Systems, Inc. said he is afraid that competitors could "take what I have licensed to them and use it as a weapon to come back and compete with me by [automated] reverse engineering. That is very hard, from a developer's point of view, to accept as a really fair form of competition," (National Research Council 1992, 31). However, Johnson-Laird (1992, 344) maintained that disassembly is not a quick and inexpensive process. Explaining that automated disassemblers cannot completely recreate a program, he likened the disassembly process to attempting to "make eggs from omelettes." He added (1992, 348), "[D]ecompilation simply cannot decrease the cost of getting a competitive program to market -- it takes more time to reverse engineer an entire program than to design a clone from scratch." Stallman (1992) concurred, adding that permission to disassemble will not "have a significant effect on how easy it is for someone to write programs that can overall compete with existing programs. A competent programmer can do this more easily without disassembly." However, disassembly is sometimes the easiest or only method for obtaining information not available elsewhere. The problem with automated disassembly, explained Bjornsson (1992), is that it does not reveal the programmers' comments and annotations. Without these, it is quite difficult to learn enough about the program to make changes to it. Therefore, it will be difficult to improve the program or correct errors in it without spending a considerable amount of time studying it. The only easy, but illegal, way to make a cheap copy of a program is to copy it directly, Bjornsson said. Regardless, of the effort or expense involved in creating programs that are substantially similar to copyrighted programs, it is difficult to prove that these programs are infringing stated CBEMA and IACC (1992, 7). "Programming languages are not familiar to the lay fact-finder, who may have difficulties in discerning substantial similarity in expression in these arcane languages. As a result, software authors face serious proof problems akin to those faced by authors of novels or plays in unfamiliar foreign languages." CBEMA and IACC maintained that software plagiarism could lead to less investment in innovation. Permitting copying for decompilation would open the door to the "creation" of computer programs through successive adaptations of successful programs that essentially disguise the plagiarism. Decompilation provides a short cut -- by infringing someone else's copyright rights -- for a copier to circumvent much of the development time and expense that creating an innovative program requires. Allowing commercial decompilation would thus diminish incentives to create computer programs, especially sophisticated programs that require substantial investment," (CBEMA and IACC, et al. 1992, 31). Intermediate copying supporters say intermediate copying need not lead to the creation of a product that is substantially similar to a copyrighted work. Often intermediate copying leads to a product which builds on existing ideas. Stallman (1992) gave a hypothetical example in which he explained that it would be easier to design his own code than disassemble someone else's code in order to create a clone of a popular spreadsheet program. "However, if the file format were secret, and I wanted to read and write the same file format, then I would have to disassemble. This would be true whether my project were a complete clone, or a totally different and innovative spreadsheet. Even for a project so different you could hardly call it a spreadsheet at all, I might have occasion to read those files, and then I would need to disassemble. Bjornsson (1992) explained that most products are developed by copying the ideas behind other products. For example, he noted, most cars have four wheels and look basically the same. "A certain amount of copying is desirable....If you had three-, five-, or seven-wheeled cars, we would all be pretty confused," he explained. In the computer industry, too, most products are not completely original. In fact, as stated by the New Mexico court in Autoskill Inc. v. National Educational Support Systems Inc., an important goal of copyright is "encouraging others to build upon the ideas conveyed in a work," (BNA 1992b, 122). Therefore, the court filtered general ideas from specific expressions when testing for substantial similarity. George Washington University law professor James Chandler (1992) added that most computer programs are incremental improvements over previous programs, not entirely new program ideas. Therefore, he said, it is important that there be a sharing of ideas in the software industry. "If you're working in a limited area, everybody is stepping on the shoulders of everyone else as they get more and more sophisticated -- if one person hides what they produced they basically cheat everybody else." FOREIGN COMPETITION Companies on both sides of the intermediate copying debate are concerned about foreign competition. The United States currently has an estimated 80 to 85 percent of the world packaged-software market share and U.S.-based software vendors earned 52 percent of their revenues outside the United States in 1991. Software is one of the seven "critical" technologies selected by the Council on Competitiveness, indicating that it is an area in which the United States is strongly competitive (U.S. Department of Commerce 1992, 27-24, 27-25). How best to maintain the United States' competitive edge in software is a debatable question. Those who consider intermediate copying the first step towards plagiarism are particularly concerned that foreign companies may reverse engineer U.S. products to develop cheaper substitute products. They say that proving plagiarism through substantial similarity tests will be difficult or impossible. Therefore, foreign competitors will be able to use cheap copies to drive American manufacturers out of business. "It would open up our products to access at the source code level to large competitors over seas," Siber explained (1992). Indeed, stated economics professor Edwin Mansfield, Japan concentrates much of its industrial research and development effort on "improvement and adaptation of existing products and processes, many of which were invented in the West," (1988, 16- 17). Japanese companies tend to be able to adapt, improve, and market innovations developed by other companies more quickly and less expensively than U.S. companies can. Mansfield (1988, 19) suggested that this may be a result of the fact that "the Japanese, in carrying out such innovations, have been more likely than the Americans to make significant technical adaptations of the imitated product and/or to reduce its production costs substantially." He added that U.S. firms tend to place more emphasis "on marketing strategies than on technical performance and production cost." Intermediate copying supporters say that banning intermediate copying in the United States will not prevent foreign software developers from making intermediate copies. They maintain that U.S. software developers will essentially be forced to work under a handicap. Johnson-Laird (1992, 353) explained, "If the European Community (and perhaps then the United States) were to ban reverse engineering, they would be engaging in a form of software protectionism. This protectionism would backfire badly if, for example, the next generation of innovative software were to come from the Pacific Rim -- our own programmers could not then develop competitive products." Samuelson (1992) added that the United States cannot control intellectual property laws in other countries. "We can put trade pressure on countries to make them conform to our laws, but they're entitled to their sovereignty," she said. INTEROPERABLE SYSTEMS The American Committee for Interoperable Systems (ACIS), "an informal organization of companies...that develop innovative software and hardware products to interoperate with computer systems developed by other companies," (ACIS 1992, 1) argued in an amicus curiae brief that intermediate copying is essential to the development of interoperable systems. Interoperable systems are computer system components, such as a computer and its disk drives, designed to function together as a single system. ACIS stated that intermediate copying is necessary when vendors refuse to disclose interface specifications which are essential for establishing communication between different elements of computer systems. A vendor that has achieved success in the marketplace with one or more of its computers has a natural incentive to restrict access to its interface specifications for these computers, for by doing so it can hold consumers already committed to its products captive to its product line, and, at its discretion, deprive them of a greater -- and competitively priced -- variety of products from other vendors (ACIS 1992, 4). Chris Byrne, senior intellectual property counsel for Storage Technology Corporation, explained that hardware and software vendors "should be able to understand how software interfaces with other hardware and software....Legally precluding this activity would lead to a loss of innovation." A manufacturer of data storage devices which attach to computers manufactured by other companies, the company relies on information about other vendors' operating systems. According to Byrne, Storage Technology manufactures products which do not compete with the products they attach to. However, if intermediate copying were not legal, Storage Technology's products would only be able to attach to systems whose manufacturers grant permission. "The interoperability information should be available. If there is no other way to get it, reverse engineering should be legal....If the owners of these systems can decide who can attach to their systems and who can't, they really have a stranglehold on innovation," said Byrne. However, IBM's Victor Siber insisted that interoperability "information is generally made available by companies because it is in their best interest to do so." Apple's Jim Burger agreed: "Apple can't survive without interoperability....But Apple software engineers do not engage in wholesale decompilation to achieve interoperability." Bjornsson (1992) maintained that sometimes interoperability information is simply not available because it involves parts of computer systems that have not been documented. "At the present level of electronic and computer technology there are vast dimensions of detail that are not well documented. If one wants to know about that particular thing -- be it the designer or the user -- they'll just look at it," he explained. "The thing itself is often the only reliable source of information about a high-technology product." Johnson-Laird (1992a, 347) added that computer programs are usually too complex to be thoroughly tested and documented. "Even in those cases where companies deliberately publish detailed internal information -- Apple Computer is a good example with its Inside Macintosh technical reference manuals -- such documentation has many discrepancies; it simply fails to provide complete and accurate information about the software as it really exists." A representative of one large computer manufacturer, who asked not to be named, said that interoperability information is usually available. However, he said his company uses intermediate copying techniques occasionally "on a research type level for better understanding certain techniques that may be employed to ensure interoperability when information is not available through any other vehicle." Burger argued that rarely is there information that cannot be obtained without reverse engineering through intermediate copying. "There may be some esoteric piece of information that may allow you to run your program one nanosecond faster, but that's an extreme case." Maintaining that wholesale disassembly to create a competing product is illegal, he added that sometimes it costs much more to find information legally. "Their basic message is it's expensive for us to do things legally," Burger said. However, supporters of intermediate copying say that while they sometimes disassemble most or all of a program to find information not available elsewhere, they do not advocate complete disassembly of a program to create an infringing clone product. There are tools such as debuggers that software engineers, including those at Apple, use without making intermediate copies. Debuggers allow programmers to monitor programs while they are running and disassemble them one section at a time. It is necessary, however, to make a copy of sections of the disassembled code if the programmer wishes to study more than a few lines of it at a time. Stern (1992, 3) explained, "If you can do your reverse engineering in your head, in RAM, or on an 80-character, 25-line screen, you can stay in business. But put any of the original code or a translation of it into nonvolatile memory and you are liable for copyright infringement." The CBEMA and IACC amicus curiae brief pointed out that intermediate copying is not necessarily a copyright infringement. "We do not suggest that decompilation is improper in all situations. Under appropriate circumstances, decompilation does not prejudice the rights of the software copyright owner and would not be an infringement," CBEMA and IACC stated (1992, 6). Burger maintained that intermediate copying should be completely prohibited where the entire program is disassembled and the results used to create a competing or replacement program. He said that if companies have legitimate reasons to make intermediate copies, they are not likely to be sued. "It's not logical for any company to sue someone even if they are engaged in wholesale disassembly for purely academic purposes or true interoperability -- even though we believe such decompilation unnecessary to achieve interoperability. Why would any company sue if there are no damages? It is only when the real goal of the disassembler is to unfairly capitalize on the work of another, or to create a knockoff, competitive program, where wholesale intermediate copying will assuredly result in legal action." A summary of the key conflicts and concerns surrounding intermediate copying is presented in the following table. [table omitted] POLICY ALTERNATIVES AND RECOMMENDATIONS Policy makers who attempt to resolve the intermediate copying question are faced with the challenge of protecting the rights of developers to a monopoly over their expressions and the rights of the public to access to the ideas contained in computer software. The National Research Council (1992, viii) reported, "The magnitude of the up-front investments required to create competitive software products, and of the downstream investments needed to sell and support them, inevitably gives rise to conflicting desires to encourage innovation while preserving stability in a huge, competitive industry." Several questions should be considered when developing policy options which address the intermediate copying question. - Should intermediate copying ever be permitted? If so, under what conditions should it be permitted? - Should legislative action be taken to explicitly permit or prohibit intermediate copying? If so, what type of legislative action will best achieve the desired goals? - If intermediate copying is permitted, how can software developers be protected from pirates? These questions are discussed below and policy recommendations are proposed. In considering these questions it is assumed that there are legitimate uses for reverse engineering techniques which require intermediate copying and that these techniques are currently practiced by software engineers. Therefore, any restrictions on intermediate copying are considered to be changes to the status quo. SHOULD INTERMEDIATE COPYING EVER BE PERMITTED? IF SO, UNDER WHAT CONDITIONS SHOULD IT BE PERMITTED? Those who oppose intermediate copying usually argue that the intermediate copy can be easily modified to create a competing program, thus depriving software developers of some of the reward for their investment in innovation. However, the mere existence of an unauthorized intermediate copy does not damage the software developer -- the developer is only damaged when the intermediate copy is used in an illegal manner. As Johnson-Laird (1992, 332) explained, "If the resulting computer software is strikingly similar to the original, then it does not matter what the process was -- the program will always be infringing. If the process used to produce the program happened to be reverse engineering, that does not mean reverse engineering is a per se copyright infringement." There are legitimate and generally desirable uses of intermediate copying, including the development of interoperable systems, that could be precluded if intermediate copying is prohibited. The European Community's directive has been viewed by some as a good compromise position, prohibiting intermediate copying unless "information necessary to achieve interoperability has not previously been readily available" (Official Journal of the European Community 1991, 45). However, the definition of interoperability is vague, and whether or not information has been made readily available is debatable. Therefore, the question will still be left largely to the interpretation of the courts. In addition, the directive precludes other uses of reverse engineering including customization of software to work with a particular system, and correction of software problems that surface when the software is used under a special set of conditions. Another possible compromise position would be to only prohibit making an intermediate copy of an entire program for the purpose of developing a competing product. This would help prevent plagiarism while allowing most legitimate uses of intermediate copying. It is conceivable, however, that someone might have need to completely disassemble a program to learn information not available elsewhere in the process of developing a competing -- but noninfringing -- product. Burger argued that prohibiting the intermediate copying of an entire program will not preclude legitimate activities unless they damage a company that holds a software copyright. However, in the United States, freedom, not restriction, is the norm. Therefore, as Stallman stated, "everything should be permitted unless there is good reason to prohibit it." There is no good reason to prohibit intermediate copying. If the American legal system is ill-equipped to identify strikingly similar programs, then that is the problem which needs to be addressed -- not intermediate copying. Recommendation: The problem of companies creating infringing programs through intermediate copying should not be addressed by prohibiting intermediate copying, but by prosecuting those who use intermediate copies to create infringing programs. Intermediate copying should be permitted unconditionally. SHOULD LEGISLATIVE ACTION BE TAKEN TO PERMIT INTERMEDIATE COPYING? IF SO, WHAT TYPE OF LEGISLATIVE ACTION WILL BEST ACHIEVE THE DESIRED GOALS? As long as the laws concerning intermediate copying remain uncertain, software developers must gamble when deciding whether or not to use techniques which require intermediate copying. If they make intermediate copies, they may be liable for copyright infringement in one court and not in another. If they choose a more expensive and time-consuming development method to obtain information not otherwise available or do without the information, it may prove to be a wasted effort if the courts determine that intermediate copying is legal. Therefore it is important that the issue be resolved quickly and unambiguously. Some say they are content to wait for the Ninth Circuit Court to rule on Sega v. Accolade before they consider lobbying for legislative action regarding intermediate copying. This ruling may provide a clear interpretation of the law; however, it may take many more court cases to make the interpretation of the law clear. Samuelson (1992) explained, "As a society, Americans like the common law process of decision-making where the details of the law get filled in on a case-by-case basis. We've been waiting for a case to resolve the reverse engineering conflict....That's a nutty way of going about things." Mitchell Kapor, chairman of ON Technology, Inc. stressed the importance of not blindly maintaining the status quo and leaving the decision to the courts. "[S]aying it is okay to do nothing because things will work out strikes me as analogous to the response given by the man who jumped off the top of the Empire State Building. When asked as he passed the 50th floor, 'How's it going?', he replied, 'So far, terrific,'" Kapor said (National Research Council 1991, 87). Even if the courts clarify the law, their decision may be based on precedent set by cases having nothing to do with the circumstances facing the computer industry. Therefore it is important to consider Congressional action to explicitly permit intermediate copying. There are a variety of options for Congressional action. These include amending the Copyright Act and enacting a sui generis law -- a law that applies to a specific technology -- that would replace copyright or patent protection for computer software. Amending the Copyright Act is likely to be the easiest and quickest option to enact; alternatively, replacing existing intellectual property protection with a sui generis law would enable lawmakers to address a variety of intellectual property issues specific to software. A new law may be accompanied by a highly uncertain transitional period, in which the law will be tested in the courts to establish precedents for its enforcement. Further a new law may not fit under established international treaty structure (U.S. Congress 1992, 31). However, a law that is carefully drafted with input from both computer and intellectual property professionals could be a significant improvement over current intellectual property laws for computer software. Recommendation: The question of whether to take legislative action to explicitly permit intermediate copying requires further study by both software engineers and intellectual property professionals. This study should include an assessment of what types of activities would be restricted by the courts' rulings on intermediate copying. If legislative action is to be taken, a two-tiered approach should be considered in which Congress first amends the Copyright Act to explicitly permit intermediate copying and later enacts a sui generis law which addresses intermediate copying as well as other intellectual property issues specific to the software industry. HOW CAN SOFTWARE DEVELOPERS BE PROTECTED FROM PIRATES? If intermediate copying is to be permitted under all conditions, it is likely that some companies will attempt to take advantage of the law and reverse engineer with the intent of creating infringing programs. However, reported the National Research Council (1991, 11) infringement of this kind is difficult to prove. "Already available technology can in certain circumstances decompile, or translate, computer code into a higher-level language, mechanically restructure the program, and generate new computer code that, by appearance, is substantially different from the original....if abuses do arise, several legal experts pointed out, intellectual property law may not be an effective means of redress." Therefore, if software developers are truly to be protected, they must be guaranteed that intellectual property laws will be enforced and that people who attempt to plagiarize their products will suffer civil prosecution. If intermediate copying is permitted under current conditions, said Burger (1992), "There's no copyright protection except against slavish copying." Supporters of intermediate copying who feel the current legal system is ill-equipped to deal with technical issues advocate a variety of solutions including establishing a "technical court," introducing more technical experts into the judicial system, or enacting sui generis legislation for computer software. A technical court, judged by technical experts, would be better qualified to rule on issues such as alleged software copyright infringement than the current courts are. Stern explained, "The [Accolade] decision is a loud argument for creating a specialized court to handle software cases -- or at least for referring such cases to technical experts who will not make gross blunders because they cannot understand the technology." The Carnegie Commission's Taskforce on Science and technology in Judicial and Regulatory Decision Making is considering the possibilities of educating judges on scientific and technical issues, and encouraging judges to call on scientific and technical experts for advice. In addition, they are considering proposals to improve judges' access to scientific information in the pre-trial process and to legislatively remove certain types of scientific and technical cases from the jurisdiction of the courts. The Commission plans to release a report on this subject in spring 1993. It is clear that the current legal system cannot adequately handle technical cases. The addition of technical experts into the judicial system will improve the system's ability to decide software cases fairly. It is important that the U.S. legal system be prepared to handle all technical cases with which it will be faced now and in the future. Recommendation: Congress should modify the judicial system to ensure that judges have sufficient technical background to decide technical cases. CONCLUSIONS How to best promote innovation while maintaining the U.S. software industry's competitive edge is a debatable question. Some maintain that intermediate copying techniques are essential for innovation, while others counter that permitting these techniques will lead to a loss of competitiveness. The intermediate copying debate focuses primarily on the scope of copyright law, foreign competition, and interoperable systems. The amount of intermediate coping performed by software engineers is uncertain; however, it is clear that there are legitimate uses for intermediate copying techniques and that they are currently practiced by software engineers. After considering the arguments, it is clear that reverse engineering through intermediate copying should be permitted under all conditions. The problem of companies creating infringing programs through intermediate copying should not be addressed by prohibiting intermediate copying, but by prosecuting those who use intermediate copies to create infringing programs. The possibility of legislative action to explicitly permit intermediate copying under all conditions should be studied further by software engineers and intellectual property professionals. This study should include an assessment of what types of activities would be restricted by the courts' rulings on intermediate copying. If legislative action is to be taken, a two-tiered approach should be considered in which Congress first amends the Copyright Act and later enacts a sui generis law which addresses intermediate copying as well as other intellectual property issues specific to the software industry. Furthermore, to help protect against software plagiarism and to ensure the U.S. judicial system is prepared to handle technical cases now and in the future, Congress should modify the judicial system so that judges can acquire sufficient technical background to decide technical cases. REFERENCES CITED Accolade, Inc. 1992. Appellant's reply brief submitted to the United States Court of Appeals Ninth Circuit, No. 92-15655. American Committee for Interoperable Systems (ACIS) et al. 1992. Brief of Amicus Curiae submitted to the United States Court of Appeals Ninth Circuit, No. 92-15655. Bjornsson, E. Steuart (Independent Software Developer, Hardware Manufacturer, and IEEE-USA Intellectual Property Committee Member). 1992. telephone interview. 29 June. BNA. 1992a. News & Comment: Reverse engineering software by disassembly may be infringing. BNA's Patent, Trademark & Copyright Journal. 43(23 April): 545-546. BNA. 1992b. News & Comment: Whelan approach to idea and expression is too simplistic. BNA's Patent, Trademark & Copyright Journal. 44(11 June): 122-123. Burger, Jim (Director of Government Law, Apple Computer, Inc.). 1992. telephone interview. 29 June. Byrne, Chris (Senior Intellectual Property Counsel, Storage Technology Corporation). 1992. telephone interview. 24 June. Chandler, James (Professor of Law, George Washington University National Law Center). 1992. personal interview. 8 June. Computer and Business Equipment Manufacturers Association (CBEMA) and the International Anticounterfeiting Coalition, Inc. (IACC) , et al. 1992. Brief of Amicus Curiae submitted to the United States Court of Appeals Ninth Circuit, No. 92- 15655. Copyright Law Professors. 1992. Brief of Amicus Curiae submitted to the United States Court of Appeals Ninth Circuit, No. 92-15655. Cutler, Norton (Assistant Chief Council, NCR). 1992. telephone interview. 26 June. Johnson-Laird, Andrew. 1992. Reverse engineering of software: separating legal mythology from actual technology. Software Law Journal. 5(2): 331-334. _______. 1992b. Technical demonstration of "decompilation." In the proceedings of Reverse Engineering: Legal and Business Strategies for Competitive Product Design in the 1990s. 23- 24 April. Sanfrancisco. Prentice Hall Law & Business. Mansfield, Edwin. 1988. Intellectual property rights, technological change, and economic growth. In Intellectual Property Rights and Capital Formation in the Next Decade. Charles E. Walker and Mark A. Bloomfield, ed. Lanham: University Press, 3-26. Markoff, John. 1992. Ruling may restrict copyrights for software. The New York Times. (June 24): D1, D5. National Research Council. 1991. Intellectual Property Issues in Software. Washington, DC: National Academy Press. Official Journal of the European Communities (English edition). 1991. Council directive of 14 May 1991 on the legal protection of computer programs. 34(L 122, 17 May): 42-46. Samuelson, Pamela. 1990. Reverse engineering someone else's software: is it legal? IEEE Software. (January). Reproduced electronically with permission. Samuelson, Pamela (Professor of Law, University of Pittsburgh School of Law). 1992. telephone interview. 23 June. Samuelson, Pamela, Michael Denber, and Robert J. Glushko. 1992. Developments on the intellectual property front. Communications of the ACM. 35(6): 33-9. Siber, Victor (Senior Corporate Counsel, IBM Corporation). 1992. telephone interview. 24 June. Stallman, Richard (Founder, League for Programming Freedom and Free Software Foundation). 1992. personal correspondence. 2 June, 14 July. Stern, Richard H. 1992. No accolades for Accolade court. IEEE Micro. (June): 3-6. U.S. Congress, Office of Technology Assessment. 1992. Finding a Balance: Computer Software. Intellectual Property, and the Challenge of Technological Change. Washington, D.C.: U.S. Government Printing Office. U.S. Department of Commerce. 1992. U.S. Industrial Outlook 1992. Washington, D.C.: U.S. Government Printing Office.