The Law and Economics of Reverse Engineering
Samuelson, Pamela, Scotchmer, Suzanne, The Yale Law Journal
Reverse engineering has a long history as an accepted practice. What it means, broadly speaking, is the process of extracting know-how or knowledge from a human-made artifact. (1) Lawyers and economists have endorsed reverse engineering as an appropriate way to obtain such information, even if the intention is to make a product that will draw customers away from the maker of the reverse-engineered product. (2) Given this acceptance, it may be surprising that reverse engineering has been under siege in the past few decades.
While some encroachments on the right to reverse-engineer have been explicit in legal rulemaking, others seem implicit in new legal rules that are altogether silent on reverse engineering, including the Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS) (3) and the Economic Espionage Act of 1996 (EEA). (4) TRIPS is an international treaty that, among other things, obligates member states of the World Trade Organization to protect trade secrets, yet it neither requires nor sanctions a reverse engineering privilege. (5) The EEA created the first federal cause of action for trade secrecy misappropriation. Its lack of a reverse engineering defense has troubled some commentators because rights granted under the EEA arguably implicate certain reverse engineering activities previously thought to be lawful. (6)
Among the explicit legal challenges to reverse engineering are these: In the 1970s and 1980s some states forbade the use of a direct molding process to reverse-engineer boat hulls. (7) In the late 1970s and early 1980s, the semiconductor industry sought and obtained legislation to protect chip layouts from reverse engineering to make clone chips. (8) In the mid-1980s and early 1990s, a controversy broke out about whether decompilation, a common form of reverse engineering of computer programs, was legal as a matter of copyright law. (9) Even after U.S. courts ruled that decompilation was acceptable for purposes such as achieving interoperability, (10) a related controversy broke out over the enforceability of licenses forbidding reverse engineering of software and other digital information. (11) More recently, questions have arisen about whether the decompilation of computer programs infringes upon patent rights in software components. (12) In 1998, Congress outlawed the reverse engineering of technical protections for digital versions of copyrighted works and prohibited both the creation and distribution of tools for such reverse engineering (except in very limited circumstances) as well as the disclosure of information obtained in the course of lawful reverse engineering. (13)
Our objectives in this Article are, first, to review legal developments regarding the right to reverse-engineer, and second, to understand their economic consequences.
We start in Part II with a discussion of the well-established legal right to reverse-engineer manufactured goods. In our view, the legal rule favoring reverse engineering in the traditional manufacturing economy has been economically sound because reverse engineering is generally costly, time-consuming, or both. Either costliness or delay can protect the first comer enough to recoup his initial research and development (R&D) expenditures. (14) If reverse engineering (and importantly, the consequent reimplementation) of manufactured goods becomes too cheap or easy, as with plug-molding of boat hulls, it may be economically sound to restrict this activity to some degree.
In Parts Ill, IV, and V, we consider the law and economics of reverse engineering in three information-based industries: the semiconductor chip industry, the computer software industry, and the emerging market in technically protected entertainment products, such as DVD movies. In all three contexts, rules restricting reverse engineering have been adopted or proposed. We think it is no coincidence that proposals to restrict reverse engineering have been so common in information-based industries. Products of the information economy differ from traditional manufactured products in the cost and time imposed on a reverse engineer. With manufactured goods, much of the know-how required to make the goods remains within the factory when the products go to market, so that reverse engineering can capture only some of the know-how required to make the product. The information-rich products of the digital economy, in contrast, bear a higher quantum of applied know-how within the product distributed in the market. (15)
For so-called digital content (movies, sound recordings, and the like), the relevant knowledge is entirely on the surface of the product, at least in the absence of technical protections such as encryption. Technical protections create costs for reverse engineers. When computer programs are distributed in object code form, a difficult analytical process is required to ascertain information embedded in the program, but it is there for the taking if a reverse engineer is willing to spend the time to study it. (16) For computer chips, the relevant knowledge is circuit design, which is not only embodied within the chip, but also readily accessible using technologies discussed below. (17) The challenge is to design legal rules that protect information-rich products against market-destructive cloning while providing enough breathing room for reverse engineering to enable new entrants to compete and innovate in a competitively healthy way.
Part III focuses on the semiconductor chip industry. When the competitive reverse engineering and copying of semiconductor chip designs became too easy and too rapid to enable innovators to recoup their R&D costs, Congress responded by enacting the Semiconductor Chip Protection Act of 1984 (SCPA) to protect chip makers from market-destructive cloning while affirming a limited right to reverse-engineer chips. (18) The SCPA allows reverse engineers to copy circuit design to study it as well as to reuse information learned thereby in a new chip, but it imposes a forward engineering requirement that inevitably increases a second comer's development time and increases its costs. (19) In the context of the chip industry, we think this restriction on reverse engineering is economically sound.
Part IV focuses on the software industry. Reverse engineering is undertaken in the software industry for reasons different from those in other industrial contexts. The most economically significant reason to reverse-engineer software, as reflected in the case law, is to learn information necessary to make a compatible program. The legal controversy over whether copies made of a program during the decompilation process infringe copyrights has been resolved in favor of reverse engineers. But as Part IV explains, the economics of interoperability are more complex than legal commentators have acknowledged. On balance, however, we think that a legal rule in favor of reverse-engineering computer programs for purposes of interoperability is economically sound.
Part V discusses the emerging market for technically protected digital content. Because technical protection measures may be defeated by countermeasures, copyright industry groups persuaded Congress to enact the Digital Millennium Copyright Act (DMCA), which creates new legal rules reinforcing technical measures used by copyright owners to protect their works. (20) It protects them against most acts of circumvention, against the manufacture and distribution of circumvention technologies, and against dissemination of information resulting from privileged acts of circumvention. (21) In our view, these new rules overly restrict reverse engineering, although the core idea of regulating trafficking in circumvention technologies may be justifiable.
Part VI steps back from particular industrial contexts and considers reverse engineering as one of the important policy levers of intellectual property law, along with rules governing the term and scope of protection. The most obvious settings for the reverse engineering policy lever are "on" (reverse engineering is permissible) and "off" (reverse engineering is impermissible). However, our study reveals five additional strategies for regulating reverse engineering in the four industrial contexts studied: regulating a particular means of reverse engineering, adopting a "breadth" requirement for subsequent products, permitting reverse engineering for some purposes but not others, regulating tools used for reverse engineering, and restricting the dissemination of information discerned from reverse engineering. In this discussion, we distinguish between regulations affecting the act of reverse engineering and those affecting what the reverse engineer can do with the resulting information. Some restrictions on reverse engineering and on post-reverse-engineering activities may be economically sound, although we caution against overuse of restrictions on reverse engineering because such restrictions implicate competition and innovation in important ways. Part VI also considers policy responses when innovators seek to thwart reverse engineering rights by contract or by technical obfuscation.
Intellectual property law in the United States has an important economic purpose of creating incentives to innovate as a means of advancing consumer welfare. (22) The design of intellectual property rules, including those affecting reverse engineering, should be tailored to achieve these utilitarian goals and should extend no further than necessary to protect incentives to innovate. Intellectual property rights, if made too strong, may impede innovation and conflict with other economic and policy objectives.
II. REVERSE ENGINEERING IN TRADITIONAL MANUFACTURING INDUSTRIES
Reverse engineering is generally a lawful way to acquire know-how about manufactured products. Reverse engineering may be undertaken for many purposes. (23) We concentrate in this Part on reverse engineering undertaken for the purpose of making a competing product because this is the most common and most economically significant reason to reverse-engineer in this industrial context. (24) We argue that legal rules favoring the reverse engineering of manufactured products have been economically sound because an innovator is nevertheless protected in two ways: by the costliness of reverse engineering and by lead time due to difficulties of reverse engineering. (25) If technological advances transform reverse engineering so that it becomes a very cheap and rapid way to make a competing product, innovators may not be able to recoup their R&D expenses, and hence some regulation may be justified. An example discussed below is the plug-molding of boat hulls.
A. A Legal Perspective on Reverse Engineering
Reverse engineering has always been a lawful way to acquire a trade secret, as long as "acquisition of the known product ... [is] by a fair and honest means, such as purchase of the item on the open market." (26) As the Restatement of Unfair Competition points out, "The owner of a trade secret does not have an exclusive right to possession or use of the secret information. Protection is available only against a wrongful acquisition, use, or disclosure of the trade secret," (27) as when the use or disclosure breaches an implicit or explicit agreement between the parties or when improper means, such as trespass or deceit, are used to obtain the secret. (28) Even when a firm has misappropriated another firm's trade secret, injunctive relief may be limited in duration based in part on the court's estimation of how long it would take a reverse engineer to discover the secret lawfully. (29)
The legal right to reverse-engineer a trade secret is so well-established that courts and commentators have rarely perceived a need to explain the rationale for this doctrine. A rare exception is the 1989 U.S. Supreme Court decision, Bonito Boats, Inc. v. Thunder Craft Boats, Inc., which characterized reverse engineering as "an essential part of innovation," likely to yield variations on the product that "may lead to significant advances in the field." (30) Moreover, "the competitive reality of reverse engineering may act as a spur to the inventor" to develop patentable ideas. (31) Even when reverse engineering does not lead to additional innovation, the Bonito Boats decision suggests it may still promote consumer welfare by providing consumers with a competing product at a lower price. (32)
Further justification for the law's recognition of a right to reverse-engineer likely derives from the fact that the product is purchased in the open market, which confers on its owner personal property rights, including the right to take the purchased product apart, measure it, subject it to testing, and the like. The time, money, and energy that reverse engineers invest in analyzing products may also be a way of "earning" rights to the information they learn thereby. Still another justification stems from treating the sale of a product in the open market as a kind of publication of innovations it embodies. This publication dedicates these innovations to the public domain unless the creator has obtained patent protection for them. (33)
Courts have also treated reverse engineering as an important factor in maintaining balance in intellectual property law. Federal patent law allows innovators up to twenty years of exclusive rights to make, use, and sell an invention, (34) but only in exchange for disclosure of significant details about their invention to the public. (35) This deal is attractive in part because if an innovator chooses to protect its invention as a trade secret, such protection may be short-lived if it can be reverse-engineered. If state legislatures tried to make trade secrets immune from reverse engineering, this would undermine federal patent policy because it would "convert the ... trade secret into a state-conferred monopoly akin to the absolute protection that a federal patent affords." (36) Reverse engineering, then, is an important part of the balance implicit in trade secret law.
No reverse engineering right, as such, exists in patent law. (37) In theory, there should be no need to reverse-engineer a patented invention to get information about how to make it because the patent specification should inform the relevant technical community of how to make the invention, and indeed the best mode of making it. (38) Insofar as a patent does not teach technologists everything they might want to know, it is clear that some reverse engineering activities will not infringe a patent. The purchaser of a machine embodying a patented invention, for example, is generally free to disassemble it to study how it works under the first sale principle of patent law. (39) In addition, a person who tries to make a patented invention to satisfy scientific curiosity may assert an experimental use defense to patent infringement. (40)
Until quite recently, copyright law neither had nor had need for a reverse engineering privilege. The artistic and literary works this law traditionally protected did not need to be reverse-engineered to be understood. (41) Books, paintings, and the like bear the know-how they contain on the face of the commercial product sold in the marketplace. To access this information, one can simply read or analyze the work. Moreover, at least until the admission of computer programs to its domain, copyright law did not protect industrial products of the sort that firms typically reverse-engineer. (42)
B. An Economic Perspective on Reverse Engineering
The economic effects of reverse engineering depend on a number of factors, including the purpose for which it is undertaken, the industrial context within which it occurs, how much it costs, how long it takes, whether licensing is a viable alternative, and how the reverse engineer uses information learned in the reverse engineering process. (43) In this Section, we concentrate on the economics of reverse engineering undertaken for the purpose of developing a competing product. (44)
We argue that a legal right to reverse-engineer does not typically threaten an innovative manufacturer because the manufacturer generally has two forms of protection against competitors who reverse-engineer: lead time before reverse engineers can enter (45) and costliness of reverse engineering. Lead time serves the same function as a short-lived intellectual property right. Costliness may prevent reverse engineering entirely, especially if the innovator licenses others as a strategy for preventing unlicensed entry. Provided that the cost of reverse engineering is high enough, such licensing will be on terms that permit the innovator to recoup its R&D expenses, while at the same time constraining the exercise of market power in order to dissuade other potential entrants.
Our economic assessment of reverse engineering recognizes that this activity is only one step in what is typically a four-stage development process. The first stage of a second comer's development process is an awareness stage. (46) This involves a firm's recognition that another firm has introduced a product into the market that is potentially worth the time, expense, and effort of reverse engineering. In some markets, recognition happens very rapidly; in others, it may take some time, during which the innovator can begin to recoup its R&D costs by selling its product and establishing goodwill with its customer base. (47)
Second is the reverse engineering stage. This begins when a second comer obtains the innovator's product and starts to disassemble and analyze it to discern of what and how it was made. (48) The reverse engineering stage may be costly, time-consuming, and difficult, (49) although this varies considerably, depending mainly on how readily the innovator's product will yield the know-how required to make it when confronted by a determined and skilled reverse engineer. (50) However, a reverse engineer will generally spend less time and money to discern this know-how than the initial innovator spent in developing it, in part because the reverse engineer is able to avoid wasteful expenditures investigating approaches that do not work, (51) and in part because advances in technology typically reduce the costs of rediscovery over time.
Third is the implementation stage. (52) After reverse-engineering the innovator's product, a second comer must take the know-how obtained during the reverse engineering process and put it to work in designing and developing a product to compete in the same market. This may involve making prototypes, experimenting with them, retooling manufacturing facilities, and reiterating the design and development process until it yields a satisfactory product. It may be necessary to return to the reverse engineering stage again if it becomes apparent in the implementation phase that some necessary know-how eluded the reverse engineer the first time. Information obtained during reverse engineering may, moreover, suggest possibilities for additional product innovation that will be investigated in the implementation stage. (53) For these reasons, the second comer's implementation stage may take considerable time and require significant expense.
The fourth stage in the second comer's development process is the introduction of its product to the market. How quickly the new product will erode the innovator's market share and force the innovator to reduce prices to be competitive with the new entrant will depend on various market factors. (54)
In the chart and discussion below, we use four criteria to assess the social welfare effects of the law's recognition of a right to reverse-engineer. The criteria are the effects on the following: incentives to innovate, incentives to engage in follow-on innovation, prices, and socially wasteful expenditures of resources. At first glance, these considerations seem to cut in opposite directions in the manufacturing industry context. On the negative side, the right to reverse-engineer seems to decrease incentives for first comers to introduce new products and to encourage wasteful expenditures on reverse engineering. (55) On the positive side, a right to reverse-engineer can increase competition in the marketplace, lead to lower prices, and spur follow-on innovations by second comers.
However, the argument against reverse engineering based on wasted costs is misleading because the cost of reverse engineering can be avoided by licensing. (56) Licensing should be in the interest of both the innovator and potential reverse engineers as they can share the saved costs.
The key question, however, is how the threat of reverse engineering affects incentives to innovate. If reverse engineering actually occurs, it will erode market power and reduce the innovator's profit to an extent determined by the costliness and time required for reverse engineering. With licensing, the threat of reverse engineering will reduce the innovator's profit to a similar extent. In order to avoid reverse engineering by unlicensed entrants, the licensor must make sure that reverse engineering by unlicensed entrants is unprofitable. He can do this by allowing some measure of competition from licensees (e.g., by licensing with low royalties). (57) How much competition he authorizes will depend on the costs that unlicensed entrants would have to bear in reverse engineering and how long it would take them. (58) The profit earned by the innovator will depend on the relative costs of the innovator and potential reverse engineers, and on the time required for reverse engineering, but not very much on whether reverse engineering is avoidable by licensing. (59)
Table 1 illustrates the social welfare effects of two possible reverse engineering rules in the context of traditional manufacturing industries: one allowing it and one disallowing it. As to each criterion, the effects of permitting reverse engineering are compared with the effects of forbidding it.
On balance, we conclude that a legal rule favoring reverse engineering of traditional manufactured products is economically sound. A prohibition on reverse engineering would, in effect, give firms perpetual exclusive rights in unpatented innovations. (60) Given that the costs and time required for reverse engineering already protect most innovators, a ban on reverse engineering is unnecessary. On the positive side, a right to reverse-engineer has a salutary effect on price competition and on the dissemination of know-how that can lead to new and improved products.
C. Anti-Plug-Mold Laws: An Exception to Reverse Engineering Rules?
In the late 1970s through the 1980s, twelve states adopted laws to prohibit plug-molding of manufactured products. (61) These laws typically forbade use of a manufactured item, such as a boat hull, as a "plug" for a direct molding process that yielded a mold that could then be used to manufacture identical products in direct competition with the plugged product. Florida's legislature had apparently been convinced that plug-molding of boat hulls was undermining incentives to invest in innovative boat designs, thereby harming a significant Florida industry. (62) California passed a more general anti-plug-mold law.
In Interpart Corp. v. Imos Italia, Vitaloni, S.p.A., (63) a firm charged with violating California's anti-plug-mold law defended against the claim in part by challenging the consistency of this California statute with federal patent policy. The Court of Appeals for the Federal Circuit rejected this challenge, characterizing California's anti-plug-mold law as a regulation of a certain use of chattels (i.e., don't use another firm's product as a plug in a direct molding process). (64) California perceived no conflict with federal patent law because its state law did not confer a right to exclude others from making, using, or selling the product. (65) Anyone could reverse-engineer and copy a manufactured product by conventional means; they just couldn't do so by plug-molding. (66)
Four years later the U.S. Supreme Court overruled Interpart in Bonito Boats, Inc. v. Thunder Craft Boats, Inc. (67) One reason the Court gave for striking down Florida's anti-plug-mold law was that it "prohibit[ed] the entire public from engaging in a form of reverse engineering of a product in the public domain." (68) The Court said that it was "difficult to conceive of a more effective method of creating substantial property rights in an intellectual creation than to eliminate the most efficient method for its exploitation." (69) Drawing upon earlier preemption rulings, the Court said they protected "more than the right of the public to contemplate the abstract beauty of an otherwise unprotected intellectual creation--they assure its efficient reduction to practice and sale in the marketplace." (70) It went on to say that "[w]here an item in general circulation is unprotected by patent, `[r]eproduction of a functional attribute is legitimate competitive activity.'" (71)
The economic consequences of plug-molding deserved more serious consideration. (72) The plug-mold process dramatically reduces the costs of, and time required to engage in, reverse engineering and reimplementation of an innovation. If plug-molding undermines incentives to invest in innovative boat hulls or other manufactured goods, a ban on the use of the plug-mold process might be economically sound, at least for some period of time. (73) The germ of an argument that plug-molding might have market-destructive effects can be found in Bonito Boats. The Supreme Court noted that Bonito Boats had expended substantial resources in developing the boat hull that it sought to protect in the litigation against Thunder Craft Boats, (74) and that the very purpose of the plug-mold process was to "`provide a method for making large molded boat-hull molds at very low cost, once a prototype hull has been provided.'" (75) Yet the Court gave very little attention to these details in its lengthy legal and policy analysis of the case.
The Supreme Court suggested in Bonito Boats that plug-mold duplication of boat hulls was "an essential part of innovation in the field of hydrodynamic design." (76) Professor Heald has questioned this assertion, pointing out that the Florida law "primarily discriminates against those interested in reproduction rather than innovation" (77) and implying that plug-molding might well "result in less innovation." (78) Heald's is the more economically sound view of the effects of plug-molding on follow-on innovation. (79)
Of course, this does not mean that the laws enacted in Florida or California were adopted on the basis of economic merit. Some features of the Florida law suggest that it was the product of a rent-seeking special interest group lobby. Consider, for instance, that the law applied retroactively to boat hulls already in existence. (80) Moreover, it did not require any showing of originality, novelty, or improvement as a criterion for the grant of protection. (81) Nor was there any durational limit to the protection. (82) It is difficult to believe that perpetual rights are necessary to enable boat-hull designers to recoup their R&D expenses. (83) An economically sound anti-plug-mold law might, then, apply only prospectively, have a minimal creativity requirement and a durational limitation aimed at providing a reasonable amount of lead time to enable innovators to recoup their investments, but not more than that. (84) In 1998, Congress enacted a sui generis form of intellectual property protection to protect boat hulls from unauthorized copying, not just from plug-molding. (85)
From an economic perspective, anti-plug-mold laws illustrate that even in the context of traditional manufacturing industries, a form of reverse engineering and reimplementation that produces cheap, rapid, identical copies has the potential to have market-destructive consequences. "[Q]uick imitation robs innovation of value." (86) Insofar as market-destructive effects can be demonstrated, it may be economically sound for the law to restrict a market-destructive means of reverse engineering and reimplementation for a period of time sufficient to enable the innovator to recoup its R&D expenses. Plug-molding is only one example of technological advances that have changed the economic calculus of reverse engineering rules, as subsequent Parts show.
III. REVERSE ENGINEERING IN THE SEMICONDUCTOR INDUSTRY
The semiconductor industry is in many respects a traditional manufacturing industry. However, we give it separate treatment here for two reasons. First, semiconductors are information technology products that bear a high quantum of the know-how required to make them on their face. (87) This made them vulnerable to rapid, cheap, competitive cloning that industry leaders asserted undermined their ability to recoup the very high costs of R&D necessary to produce new chips. (88) Second, Congress responded to these industry concerns about "chip piracy" (89) by creating a new form of intellectual property protection for semiconductor chip designs. (90)
The Semiconductor Chip Protection Act (SCPA) (91) is noteworthy for a number of reasons. (92) First, it is one of the few intellectual property laws (93) with an express reverse engineering privilege. (94) Second, the privilege permits the copying of protected chip designs in order to study the layouts of circuits, and also the incorporation of know-how discerned from reverse engineering in a new chip. (95) Third, the SCPA requires reverse engineers to engage in enough "forward engineering" to develop an original chip design that itself qualifies for SCPA protection. (96) This is in contrast to the predominant legal rule for manufacturing industries that permits reverse engineers to make and sell products identical or nearly identical to those they have reverse-engineered. (97) The economic rationale for the forward engineering requirement was not articulated with precision during the SCPA debate, but we think it is fundamentally sound as applied to this industry.
A. Perturbations in Product Life Cycles in the Chip Industry
The typical product life cycle in the semiconductor industry was relatively constant in the 1970s and 1980s. (98) A pioneering firm, usually Intel Corp., would develop an innovative new product and introduce it to the market priced handsomely so that the firm could recoup its investments. "Later, as the manufacturer [became] more efficient it [would cut] prices to expand its market and discourage competition. Nonetheless, second-source products--chips electrically and mechanically compatible with the pioneering product--eventually appear[ed] on the market. The arrival of competition precipitate[d] further rounds of price cuts." (99) Toward the end of this life cycle, the pioneer's profit margins would trail off, and it would have to hope that the next round of innovation would allow it to regain market share and profits.
Semiconductor firms have historically relied on lead time and secrecy far more than on patents to protect their intellectual assets. (100) An innovator could rely not only on being first to market to provide some lead time, but also on being further along the yield curve than imitating second comers. (101) Trade secrecy protection was especially important in the chip manufacturing process because considerable know-how was required to make commercially acceptable chips. However, trade secrecy law obviously could not protect the layout of chips sold in the marketplace, as this information was readily ascertainable from examination of the marketed product (that is, it could be readily reverse-engineered). (102)
Several factors contributed to the fact that patents did not play a crucial role in the early and mid-development phases of this industry. (103) For one thing, semiconductors are a cumulative system technology in which the interrelatedness of inventions requires extensive cross-licensing of patents in order for industry participants to make advanced chips. (104) Second, some major customers of this industry, notably the U.S. government, insisted on "second-sourcing," that is, attracting competitive suppliers of compatible chips to reduce the risk of unforeseen supply problems. (105) This, too, contributed to widespread cross-licensing. Third, the rapid pace of innovation and short life cycles of many chip products lessened the utility of patents in this industry. (106) Fourth, during the 1970s, when the semiconductor industry was becoming a major American industry, there was a widespread perception that courts were hostile to patents, and patents had, as a consequence, less economic significance than at other times. (107) A fifth limitation of patents, much emphasized in the legislative history of the SCPA, was that under then-prevailing standards, the overall layout of chip circuits was rarely if ever patentable. (108)
While the U.S. semiconductor industry thrived for years under these conditions, the life-cycle pattern of chip products was so disrupted during the late 1970s and early 1980s that leading chip producers sought legislative help. Several factors contributed to this disturbance. First, there was a steep rise in the cost of developing and marketing new chips. (109) Second, advances in chip manufacturing technologies dramatically reduced the cost and time required to make exact or near-exact competing chips, thereby shortening considerably the lead time innovators could expect and reducing the costs of copying. (110) Third, American firms were losing out to foreign--and in particular, to Japanese--competitors, raising the specter of a diminished U.S. presence in this very significant sector of the national and global economy with potentially serious national security consequences. (111)
B. Copyright or Sui Generis Protection for Chip Designs?
Intel Corp. initially sought to combat "chip piracy" with copyright law. It obtained copyright registration certificates for drawings of chip circuitry, (112) and then it sought to register masks (that is, stencils used in manufacturing chips) and chips themselves as derivative works of the drawings. This would have provided a basis for claiming that manufacturers of identical or near-identical chips were infringing copyrights in protected drawings, masks, or chips. Intel's strategy was derailed when the U.S. Copyright Office rejected its application to register chips because of their utilitarian function. (113) Although Intel sued the Register of Copyrights to compel registration. (114) it soon dropped the litigation and turned to Congress for legislative relief. (115)
Intel's second strategy was also based on copyright. It asked Congress to amend the copyright law to add "mask works" to the subject matter of copyright. (116) Intel argued that innovative chip designs, like literary works, were very expensive to develop and very cheap to copy, and unless the law intervened to stop rapid, cheap copying, innovators would be unable to recoup their R&D expenses and justify further investments in semiconductor innovation. (117) A nearly identical argument was made by the congressional Commission on New Technological Uses of Copyrighted Works (CONTU), which supported the use of copyright law to protect computer programs. (118) Because programs and chips are both utilitarian information technology products that are expensive to develop and cheap and easy to copy, one might have thought that copyright should be used for both or for neither. Yet, the copyright argument was successful as to programs, (119) but not as to chips.
During the first set of legislative hearings on the chip protection bills, some industry witnesses expressed concern about the use of copyright for chips or mask works because copyright's fair use doctrine seemed too uncertain a basis for ensuring that the common and competitively healthy industry practice of reverse engineering could continue. (120) An explicit reverse engineering privilege was added to a later bill. However, it allowed reproducing a chip design for study and analysis without expressly allowing reverse engineers to use the results in designing a new chip. (121) Industry representatives pointed out that in order to comply with second-source form, fit, and function compatibility requirements, the chips resulting from reverse engineering would necessarily be quite similar to the chips being reverse-engineered, although not necessarily in a competitively harmful way. (122)
Lack of industry consensus stalled movement on chip protection bills until 1983. By that time, a fairly large number of compromise provisions had been added to the bills to satisfy various semiconductor industry concerns. (123) Yet those compromises so deviated from traditional copyright rules that a new and different kind of opposition arose. (124) As a representative of the Association of American Publishers explained at a 1983 hearing:
[T]he AAP is not questioning the creativity, skill, labor, or investment of chip designers, or their need for and entitlement to appropriate protection.... Our concern lies ... with the fundamental departures from the copyright system that accompany the proposal, e.g., the extension of Copyright Act protection to utilitarian objects that, it is acknowledged, may not be "writings" under the Constitution ...; the limitations on remedies against infringers and the extension of compulsory licensing; and, most notably, the limitation imposed on the duration of protection of this particular class, and the distortion of the fair use doctrine to accommodate reverse engineering. (125)
It would be better, he argued, to develop sui generis legislation (126) to protect semiconductor chip designs (127)--which is what Congress ultimately did in 1984.
The SCPA regime resembles copyright in significant respects. (128) One conceptual holdover from Intel's copyright strategy was the subject matter chosen for SCPA protection, namely, mask works. (129) As with copyright, mask works must be "original" to qualify for protection. (130) Rights attach automatically by operation of law, but registration with the Copyright Office brings benefits unavailable to nonregistrants. (131) The legislative history demonstrates that copyright-like concepts of substantial similarity and substantial identity were to be used in judging infringement of SCPA rights. (132) And the SCPA relies, as copyright does, on a grant of exclusive rights to control reproductions and distributions of products embodying the protected work. (133)
A notably sui generis feature of the SCPA (134) is its reverse engineering provision:
[I]t is not an infringement of the exclusive rights of the owner of a mask work for-- (1) a person to reproduce the mask work solely for the purpose of teaching, analyzing, or evaluating the concepts or techniques embodied in the mask work or the circuitry, logic flow, or organization of components used in the mask work; or (2) a person who performs the analysis or evaluation described in paragraph (1) to incorporate the results of such conduct in an original mask work which is made to be distributed. (135)
Industry witnesses distinguished "legitimate" and "illegitimate" reverse engineering:
A reverse engineering firm should be allowed to analyze the chip, draw a circuit schematic of the chip, and then lay out a different pattern. This pattern could be used to fabricate a version of the semiconductor chip which is functionally equivalent to the original chip but has different visual patterns on it. The reverse engineering firm could then improve the performance of the chip, reduce the size of the chip and reduce the overall manufacturing cost of the chip. (136)
A "legitimate" reverse engineer would not, for example, reproduce inefficiencies or mistakes in the innovator's layout of circuits, because careful study and analysis of the chip would identify these problems. (137)
The House Report on the SCPA explained the impact of this and similar testimony:
Based on testimony of industry representatives that it is an established industry practice to ... make photo-reproductions of the mask work in order to analyze the existing chip so as to design a second chip with the same electrical and physical performance characteristics as the existing chip (so-called "form, fit and function" compatibility), and that this practice fosters fair competition and provides a frequently needed "second source" for chip products, it is the intent of the Committee to permit such reproduction by competitors ... [and to make illegal] mere wholesale appropriation of the work and investment in the creation of the first chip. It is the intent of the Committee to permit, under the reverse engineering limitation, the ... creation of a second mask work whose layout, in substantial part, is similar to the layout of the protected mask work--if the second mask work was the product of substantial study and analysis, and not the mere result of plagiarism accomplished without such study or analysis. (138)
One commentator characterized the SCPA as "accept[ing] copying as the industry norm of competition. The industry spokespersons, while seeking protection from piracy as they perceived it, were insistent on preserving and encouraging the industry practices of creative copying, a practice known to them as reverse engineering." (139)
C. An Economic Rationale for the SCPA Rules
Part II argued that reverse engineering does not unduly undermine incentives to invest in innovation as long as it is costly, time-consuming, or both. During the time that the SCPA and predecessor bills were pending in Congress, reverse engineering of chips could be done very cheaply and quickly by peeling away layers of a purchased chip, one at a time, photographing each layer, making a mask from these photographs, and then using these masks to manufacture identical chips. (140) The SCPA rules made this cheap and rapid route to competitive entry illegal and required reverse engineers to design original chips in order to avert infringement liability. The forward engineering requirement lengthened second comers' development time and increased their costs, thereby giving the innovator more lead time to recoup its R&D expenses and more protection against clone-based pricing. The forward engineering requirement also increased the likelihood that second comers would advance the state of the art in semiconductor design. (141) As long as second comers had to make their chips different, they might as well make them better.
Table 2 uses the same social welfare criteria as Table 1 to illustrate our assessment of the economic effects of pre-SCPA rules as compared with post-SCPA rules.
Incentives to invest in innovative chip designs were too low before enactment of the SCPA because cloners rapidly eroded lead time advantages for innovators. In the short run, this may have brought low prices and few wasted costs, but prices were too low to allow innovators to recoup R&D expenses as long as cloning was legal. Incentives to innovate were restored once cloning was no longer an option. Incentives to invest in follow-on innovation were also very low in the pre-SCPA era because firms capable of investing in improved chips chose instead to clone while it was still legal. When chip cloning became illegal, firms had strong incentives to invest in improvements. Although consumers may have initially benefited from lower prices in the pre-SCPA era, prices were so low that innovators could not recoup their costs. The SCPA may result in more socially wasteful costs because some second comers may spend resources making chip circuitry different to satisfy the originality requirements. However, some of these wasted costs are avoidable by licensing.
From an economic standpoint, the anticloning rules of the SCPA are designed to achieve much the same result as the anti-plug-mold rules discussed in Part II, although they do so by a different technique. Chip cloners were no more engaged in innovation-enhancing discovery of applied industrial know-how than were plug-molders. The SCPA rule inducing second comers to join the ranks of innovation-enhancing firms is similar to the anti-plug-mold rule that induced second comers to engage in more conventional forms of reverse engineering likely to advance the state of the art of boat-hull design. The SCPA achieves this result by establishing a kind of "breadth" requirement for subsequent products in contrast to the anti-plug-mold laws that instead outlawed a particular means for making a competing product. (142)
D. Post-SCPA Developments
There has been very little litigation under the SCPA rules. Yet the one reported judicial decision under the SCPA is instructive because it involved a failed reverse engineering defense. In Brooktree Corp. v. Advanced Micro Devices, Inc., (143) AMD produced a prodigious paper trail in support of its reverse engineering defense and pointed to the considerable time and expense it had spent on developing a chip compatible with the Brooktree chip. (144) It …
Questia, a part of Gale, Cengage Learning. www.questia.com
Publication information: Article title: The Law and Economics of Reverse Engineering. Contributors: Samuelson, Pamela - Author, Scotchmer, Suzanne - Author. Journal title: The Yale Law Journal. Volume: 111. Issue: 7 Publication date: May 2002. Page number: 1575+. © 2009 Yale University, School of Law. COPYRIGHT 2002 Gale Group.