ON THE COMPLEX ECONOMICS OF PATENT SCOPE
Robert P. Merges * and Richard R. Nelson
90 Colum. L. Rev. 839
[*839] INTRODUCTION
The economic significance of a patent depends on its scope: the broader the scope, the larger the number of competing products and processes that will infringe the patent. Many theoretical papers have tried to assess the effects of fine tuning various aspects of the patent system to make it more efficient. n1 But only a few have focussed on patent scope, n2 even though scope decisions are subject to far more discretion [*840] than most of the aspects more intensively studied.
Furthermore, most theoretical writing on patents is directed toward issues that as a practical matter are considered largely settled. For example, several economists have explored the question of optimal patent duration. n3 Their work did have a direct impact on the decision to extend patent terms on pharmaceuticals to compensate for regulatory lag. n4 But despite the scholarly attention to patent duration, the term of most patents remains fixed at seventeen years. n5 Likewise, there has been considerable debate over the years on the merits of compulsory licensing of patents under some circumstances, n6 yet the intellectual property community has repeatedly rejected the idea. n7 Thus, while the literature continues to generate interesting questions about bedrock assumptions and practices, it has little bearing on the everyday operations of the patent system. This Article is an attempt to redress this deficiency by analyzing the economic effects of patent scope.
The Patent Office and the courts are constantly making patent scope decisions. The Patent Office does so when it determines the claims it will allow on a specific patent. The courts do so in litigation, where questions of patent infringement are decided. In the former context, the applicant wants to claim as much as she can, and the Patent Office must decide what claims are allowable. While decisions regarding what to allow are constrained by a number of legal principles, and by the invention itself, in many cases the Patent Office has considerable room for discretion. Within that discretionary zone, the Office must [*841] decide which claims should be admitted and which ones pruned back or rejected.
After a patent has been issued, a patentee will often allege that her invention has been copied by competitors. In arguing the case, she will try to demonstrate that the accused infringer's product falls within the boundaries of her invention, as defined in her patent clainms, or that any differences between the infringer's device and her invention are insignificant. The challenger, meanwhile, will argue first that the patent is invalid, and second that her invention or product does not infringe the patent -- that it is different in some material respect from the invention claimed by the patentee. Again, the legal principles and objective evidence often leave considerable room for discretion. There has been surprisingly little theoretical discussion of how to exercise this discretion. This paper is concerned with the effects of these decisions, and with the policies that should influence them.
Several recent cases signal the nature and complexity of the questions involved in patent scope decisions. In 1988 the Patent Office granted a patent to the inventors of a transgenic mouse. n8 The Office accepted the inventor's argument that their procedure could be used to engineer higher order animals, and thus allowed a claim to any "non-human mammal" made with their procedure. n9 Is such a generalization of specific results consistent with what is known about this area of technology? What are the likely consequences of accepting this broad claim? Should this claim cover other transgenic animals, even if it takes a major breakthrough to create them?
In another case a court ruled that a blood clotting protein made with recombinant DNA techniques violated a product patent held by an earlier inventor who had purified the same protein from human blood. But the court later found the original patent invalid on the grounds that the best mode of operation was not revealed by the patent. n10 Was the [*842] first decision a reasonable one? The second? How will these decisions affect future inventions in this field?
Given the issues at stake, the question of appropriate patent scope has attracted surprisingly little attention. There has been some analytic writing on the subject, n11 but in our view most of the papers do not focus on the key issues. The well known paper by Edmund Kitch is an exception. n12 Kitch has argued that our system of granting a patent early in the development process allows an inventor to invest in development without fear that another firm will steal her work, n13 thus encouraging the inventor to coordinate her activities with other firms. n14 Kitch states that this "prospect function," which necessarily implies broad patents, explains many of the doctrines and practices of the patent system. n15 Another commentator has responded that for the most part the Patent Office and courts have resisted granting broad prospects. n16
Our own exploration of the economics of patent scope has led us to focus on very much the same kinds of issues as raised by Kitch. We proceed as follows: We begin by considering the legal doctrines that define a patent's scope, then identify the room for discretion which often exists, and point out areas of consistency and inconsistency in current practice. Next, we develop an economic analysis that illuminates the central issues at stake in varying permissible patent scope. This analysis differs from standard economic models by moving beyond the two-dimensional analysis of incentives and deadweight loss.
[*843] Much of our discussion will center on the post-invention environment for development and subsequent improvements. By contrast, the work of Nordhaus and others is concerned with conditions surrounding the initial invention. n17 One way to describe our approach is to view it as a broadening of what counts as an incentive to invent or as a social cost of issuing patents. The concept of incentives, in our view, should embrace post-invention conditions favorable to the inventor, such as extension of an initial patent to cover subsequently-developed versions of the invention. Likewise, the notion of a patent's social costs should include its potential to reduce competition in the market for improvements to the patented technology.
Like Kitch, then, we see the important question as how patent scope decisions influence the development of a technology, both in the sense of an individual invention and that of a future line of improvements extending from it. However, contrary to what Kitch suggests, we do not presume that granting broad scope to an initial inventor induces more effective development and future invention. We regard this as an open question.
Our analysis differs from the existing literature on patents in a second way as well. This literature tends to assume that invention is the same in all technologies. In contrast, we develop several models of technical advance in industry, models that differ in terms of how various inventions are related to each other. These models are designed to highlight and capture the different ways in which technical advancement proceeds in different fields. One of our major objectives is to show that the issues at stake regarding patent scope depend on the nature of technology in an industry. This dependence includes two characteristics: the relationship between technical advances in the industry, and the extent to which firms license technologies to each other.
Theoretical argument alone, however, cannot resolve the question of whether technical advance proceeds more vigorously and effectively under competition or under a regime where one person or organization has a considerable amount of control over developments. Therefore we follow our theoretical analysis with an empirical-historical examination of the course of technical advance in several industries, guided by the various models we have developed. In each industry, critical rulings regarding the scope of important early patents significantly influenced the subsequent path of the technology. Our focus will be on those critical decisions and their consequences.
We conclude with an attempt to draw lessons regarding appropriate patent scope. Our basic conclusion is this: Without extensively reducing the pioneer's incentives, the law should attempt at the margin to favor a competitive environment for improvements, rather than an environment [*844] dominated by the pioneer firm. In many industries the efficiency gains from the pioneer's ability to coordinate are likely to be outweighed by the loss of competition for improvements to the basic invention. Throughout the article we suggest ways that patent doctrine can be applied to carry out this goal.
I. PATENT LAW DOCTRINES
A. Patent Prosecution: Threshold Issues
During prosecution of a patent, a Patent Office examiner reviews an application to determine what is patentable. To be patentable an invention must meet all the statutory requirements for patentability: novelty, n18 utility n19 and non-obviousness. n20 The claims are crafted to meet these requirements. But another requirement relates more directly to the scope of the claims -- enablement, which largely concerns how the invention is described and claimed in the patent.
A patent application has two main parts. The first is a specification of the invention, which is written like a brief science or engineering article describing the problem the inventor faced and the steps she took to solve it. It also provides a precise characterization of the "best mode" of solving the problem. n21 The second part of the patent application is a set of claims, which usually encompass more than the material set out in the specification. n22 Claims define what the inventor considers to be the scope of her invention, the technological territory she claims is hers to control by suing for infringement.
The specification and claims serve quite different functions. The specification is used by the Patent Office to determine whether the inventor has made a patentable invention and, if so, whether others can make and use it. This fundamental principle -- that legal protection is [*845] premised on an adequate disclosure of the invention - is built deep into the history of patent law. n23 The patent claims serve a different function: Analogous to the metes and bounds of a real property deed, they distinguish the inventor's intellectual property from the surrounding terrain.
Claim breadth is largely a function of two doctrines. The enablement doctrine requires that the specification teach one skilled in the relevant art how to make and use all the embodiments of the invention encompassed by the claims. In appropriate cases, the doctrine of equivalents expands the scope of a patent beyond the literal language of the patent's claims. We consider each in turn.
B. Doctrines of Disclosure and Enablement
One important issue in patent law is how broad the knowledge communicated by the disclosure should be. Under section 112, the disclosure must be sufficient to enable someone skilled in the art to make and use all the embodiments of the invention claimed in the patent. This requirement can at times be applied rather loosely: a specification that describes only one working example of an invention but that supplies less guidance on the subject matter at the fringes of a patent's claims is often sufficient. n24
At first blush it might seem to make sense to limit the rights of a patentee to only those embodiments of the invention she has disclosed in her specification, i.e., those that she has actually created at the time the patent application is filed. But imitators would soon find some minor variation over the disclosed embodiments; with such an ultra-narrow enablement principle, they would then have a nonenablement defense if the patentee tried to enforce the patent. Such a rule would soon render patents useless.
The patent system recognized this danger long ago. For example, in 1904 King Gillette received a patent for the first disposable blade safety razor. n25 One of the problems Gillette faced was how to keep a very thin, detachable blade rigid during shaving. His solution, as [*846] described in his specification, was to "'secure [the] blade to a holder . . . [so that] it receives a degree of rigidity sufficient to make it practically operative.'" n26 Claim two of the Gillette patent reads "'[I claim as] a new article of manufacture, a detachable razor-blade of such thinness and flexibility as to require external support to give rigidity to its cutting edge.'" n27
Gillette's success drew imitators, including the Clark Blade and Razor Company. When Gillette sued for patent infringement, Clark claimed that Gillette's patent did not sufficiently describe all the possible embodiments of the blade and that, in particular, Clark's design fell outside the range of what Gillette's patent had described. The Third Circuit rejected this argument, quoting broad language from the Supreme Court:
[C]laim 2 is not invalid . . . for, if such were the law, patentability must have been denied to Elias Howe for "the grooved and eye-pointed needle" which constituted his seventh claim, and of which it was said [by the Supreme Court] in Deering v. Winona:
"The invention of a needle with the eye near the point is the basis of all the sewing machines used, but the methods of operating such a needle are many; and, if Howe had been obliged to make his own method a part of every claim in which the needle was an element, his patent would have been practically worthless." n28 The Gillette case illustrates that a patent's specification need not point out precisely how to make every device n29 that would fall within its claims. Disclosure of an inventive concept or principle, whose precise contours are defined by the claims, is enough. n30
The infamous Selden patent episode shows the difficulty of cabining a claimed invention. n31 The Selden patent on an automobile design had as its key claim the use of a light, gasoline-powered internal combustion engine. n32 The claim was quite general, failing to specify many [*847] important details about the engine. The Patent Office allowed that claim, and district courts upheld it twice, n33 despite arguments that the broad idea was obvious, and that the engine referred to in the claim was of a particular kind not encompassing all the engines that were claimed to infringe. Eventually, the Second Circuit drastically narrowed the claim, stating that it covered only the particular kind of gasoline engine used by Selden. n34
Another example of enablement at work is the recent patent granted to Doctors Phillip Leder and Timothy Stewart of the Harvard Medical School for their successful work involving transgenic mice. They isolated a gene which is associated with cancer in mammals including humans) and then injected the gene into a fertilized mouse egg, which yielded transgenic mice that are extremely sensitive to carcinogens. n35 This makes the mice excellent animal "models" for studying cancer drugs. Leder and Stewart claimed not only the technique they had used, or the particular transgenic mouse variety they had created, but rather all "non-human transgenic mammals" produced by their technique. It may well turn out that their admittedly important discovery was indeed this broad. n36 On the other hand, significant work may be required to obtain similar results in higher-order mammals. One wonders whether arguments by an accused infringer that she had to do considerable experimenting and problem-solving prior to producing a transgenic dog, or that she created a transgenic cat using a substantially different technique, would be sufficient to take her invention outside the Leder and Stuart claims. In fact, the European Patent Office cited just these concerns when it rejected those claims in the Leder and Stuart patent that went beyond mice and rodents. n37
[*848] It is difficult to resolve issues like these when a patent is filed; at that point, no one knows what future developments will follow or how difficult it will be to achieve them. Thus, there is an argument for granting a broad set of claims for pioneering inventions. Since the inventor may have enabled a broad new range of applications, courts reason, it is unfair to limit her to the precise embodiment through which she discovered the broader principle claimed. n38 As one opinion put it,
To restrict [a patentee] to the . . . form disclosed . . . would be a poor way to stimulate invention, and particularly to encourage its early disclosure. To demand such restriction is merely to state a policy against broad protection for pioneer inventions, a policy both shortsighted and unsound from the standpoint of promoting progress in the useful arts, the constitutional purpose of the patent laws. n39
But surely one can go too far. Although as a general rule, a patentee should be able to claim beyond her precise disclosure, current practice seems to permit a range of claims that may stretch beyond the spirit of the enablement doctrine. If the patent examiner can point to something in the prior art that indicates that some embodiments of the claimed invention will be impossible to make without more information than the inventor has disclosed, then the application may be rejected. But if the examiner cannot point to such an indication in the prior art, patent office policy dictates that even very broad claims may be allowed. n40 This means that claims to pioneer inventions often are allowed to cover ground that examiners believe, but cannot prove, is well [*849] beyond the area actually explored and disclosed by the inventor. n41 The rule puts the burden of disproving enablement on the examiner. The rationale is that any other rule would leave claim scope too much in the hands of individual examiners and their technological forecasting abilities. n42 Narrowing is left to the courts in particular infringement suits.
As we have seen, it is often very difficult to determine whether a patentee has enabled others to make and use all the devices that fall within the claims. One approach that has evolved to help make this determination focuses on the doctrine of "undue experimentation." Under this doctrine, an alleged infringer can argue noninfringement by showing that extensive experimentation beyond what was disclosed in the patentee's specification was required to make the allegedly infringing embodiment. We now turn to some examples of the doctrine at work.
In 1895, Thomas Edison brought a Supreme Court challenge to a very broad patent held by Sawyer and Mann for materials used in light bulb filaments. n43 The patentees had found that carbonized paper worked as an effective light-emitting conductor in light bulbs. Based on this invention, they filed a patent claiming the right to use all carbonized fibrous or textile material as an incandescing conductor. n44 Edison challenged Sawyer and Mann, contending that the claim was too broad: it did not indicate which of the thousands of "fibrous or textile material[s]" would work as conductors in light bulbs, since most do not. Nor did it describe any method for finding out. In effect Edison argued that all Sawyer and Mann had invented was a carbonized paper [*850] conductor for use in a light bulb, not a broad class of materials. Edison pointed to his own painstaking experimentation with a wide variety of materials, arguing that his discovery that a particular part of a variety of bamboo plant performed well as a filament was not made any easier by Sawyer & Mann's disclosure. The Court agreed, stating that "[i]f the description be so vague and uncertain that no one can tell, except by independent experiments, how to construct the patented device, the patent is void." n45 The patent would have been upheld, the Court suggested, if it had claimed only what Sawyer and Mann had actually invented (carbonized paper incandescence); it was invalid, however, since it would take a good deal of additional experimentation to determine whether incandescing conductors could be made out of the many materials they claimed. n46
In an earlier case, O'Reilly v. Morse, n47 the Supreme Court considered a similar issue. The case involved a challenge to the scope of a claim in Samuel Morse's famous telegraphy patent. Morse claimed "the use of the motive power of the electric or galvanic current, which I call electro-magnetism, however developed for making or printing intelligible characters . . . at any distance []."n48 In essence, Morse declared ownership of all methods of communicating at a distance using electromagnetic waves. But since he had not actually disclosed "all methods" in his specification, much less even imagined them, the Court [*851] ruled the claim invalid. n49 As wih light bulb case, the pantantee disclosure was found to be nonenabling.
We turn now to a more specialized scope issue, the patenting of natural products. Although this issue has arisen before in chemical patents, n50 it is of increasing importance because many biotechnology companies are using bacteria and other expression "vehicles" to produce purified versions of naturally-occurring proteins. n51 These patents typically claim purified versions of products that exist in nature. In these cases, it can be argued that it is stretching the concept of inventing greatly to say that the patentee really invented the products. The true invention seems to be a way of producing those products in a desirable form. But because a product claim is typically broader than one simply on a particular way of making that product, patentees seek -- and often obtain -- product patents. n52 Thus the product versus process patent issue [*852] in chemical and biological technologies is an interesting variation on the patent scope issue.
A related question concerns what is patentable when a new use is discovered for a known, patented product, an event relatively common in chemical products. Dawson Chemical Co. v. Rohm & Haas Co., n53 a Supreme Court case, involved a patent for a new application of propanil, a chemical that had earlier been held to be unpatentable over the prior art. n54 The patentee claimed a process for using the chemical as a fungicide, a use that had not been previously known. n55 The case thus illustrates how process patents can be used to protect a newly discovered use for a known compound. It encourages patent applicants to draft claims in the form "the process of applying Old Product X to New Application Y," and thereby protect their discovery -- a new application -- in spite of the fact their applications exploits a well known compound which is not itself patentable. n56
It is difficult to summarize the content of the disparate doctrines that ensure adequate disclosure. As our brief review illustrates, the factual diversity of cases involving disclosure issues leads to generalized standards that must be applied to a wide array of specific technologies. As a result, courts have a large amount of discretion in applying the doctrines. While one might note with caution certain trends in recent allowed patents, n57 our primary point is not to critique doctrine, but to point out that this discretion exists. After describing the effects of patent breadth on technical advance in Part III, we will suggest ways that courts can use this discretion in certain cases to increase the overall benefits of the patent system.
C. Infringement Doctrines
Doctrines relating to enablement have provided a way of determining the appropriate scope of claims. But claims inevitably leave room for interpretation. Even when a claim is not disputed, it is not always clear on its face whether an allegedly infringing device falls within the claim. Further, in many cases an allegedly infringing device may lie lie [*853] outside the literal scope of the claims, yet a court will find that it falls so close to this scope as to be justly included as an equivalent.
1.Literal Infringement and the Interpretation of Equivalents. -- Courts analyze infringement in two steps. First, they ask whether the challenger's product falls squarely within the boundaries of the patentee's claims -- that is, whether there is "literal infringement" of the patent. n58 If the court determines that there is no literal infringement it moves on to the second question: whether the challenger infringes under the "doctrine of equivalents." The doctrine of equivalents developed because of the frequency of cases where, even though the accused product or process does not literally infringe a claim, it may be considered essentially the same device as was patented. Of the many articulations of the doctrine of equivalents, Judge Learned Hand's captures it the best:
[A]fter all aids to interpretation have been exhausted, and the scope of the claims has been enlarged as far as the words can be stretched, on proper occasions courts make them cover more than their meaning will bear. n59 What is such a "proper occasion"? The Supreme Court wrote in 1950, quoting from an earlier case:
[I]f two devices do the same work in substantially the same way, and accomplish substantially the same result, they are the same, even though they differ in name, form, or shape. n60
A good application of the doctrine of equivalents is International Nickel Co. v. Ford Motor Co. n61 International Nickel obtained a patent that "cover[ed] a cast ferrous alloy" called "nodular iron." n62 The patent taught the addition to molten iron of a "'small but effective'" quantity of magnesium, fixed by the patent as "about 0.04%" as a minimum. n63 The magnesium caused "the graphite (crystallized form of carbon) to occur in spheroidal rather than flake form thereby producing a product with vastly improved physical properties." n64 International Nickel accused Ford Motor Company of infringement when Ford began making a nodular iron. Even though Ford's iron contained under 0.02% magnesium -- less than half the minimum required in International Nickel's [*854] patent -- it was judged to be an equivalent substance, and thus to infringe the patent. n65
Courts have determined how broadly they see "equivalents" based on the degree of advance over the art the original patent represents. When the patent is on a "mere improvement" the courts tend not to consider as "equivalent" a product or process that is even a modest distance beyond the literal terms of the claims. n66 On the other hand, a patent representing a "pioneer invention" -- which the Supreme Court has defined as "a patent covering a function never before performed, a wholly novel device, or one of such novelty and importance as to mark a distinct step in the progress of the art" n67 -- is "entitled to a broad range of equivalents. n68 That is, when a pioneer patent is involved, a court will stretch to find infringement even by a product whose characteristics lie considerably outside the boundaries of the literal claims. n69
Of course the question of infringement also turns on the precise characteristics of the allegedly infringing device. Following the test laid down by the Supreme Court in Graver Tank, n70 courts confronted with a [*885] device accused of infringing inquire whether it performs the same function and achieves the same result as the invention in the claims, and whether it does so in the same way. Where the accused device shows only minor or "insubstantial" n71 variations in one of these elements -- such as the small movement of one part or a minor change in structure -- infringement will be found even if the patentee's invention is a "mere improvement." n72 And even a pioneer patent is not infringed by a device that achieves a different result, or achieves it in a different way. n73
One important set of cases under this doctrine has grappled with the question of whether new technologies, unforeseen at the time the patent was issued, can constitute equivalents. This issue arises when a subsequent device that uses new technology is accused of infringing the original patent. The early cases we split, but the prevailing view now is that new technology can be equivalent. n74 This is true despite the statement in Graver Tank that an important determinant in the equivalents inquiry is whether "persons reasonably skilled in the art would have known of the interchangeability of an ingredient not contained [*856] in the patent with one that was. n75 Despite this language in the leading Supreme Court case on the subject, a device performing the same function and achieving the same result in the same way as a patented invention can be found to infringe even if it uses technology developed after the patent was issued. But this observation is subject to two caveats: 1) new technologies can constitute equivalents only so long as they do not perform a different function n76 or cause the device to operate in a substantially different way: n77 and 2) a truly meritorious improvement can escape even literal infringement under the "reverse" doctrine of equivalents discussed below.
That these distinctions may not always be easy to make is demonstrated by the case of Hughes Aircraft Co. v. United States. n78 Hughes Aircraft had a patent, developed by an employee named Williams, on a means of controlling the attitude of a communications satellite. The claims called for receiving and directly executing control signals from a ground station on earth. After the patent was issued, advances in semi-conductor technologies permitted satellites to use on-board microprocessors to process and execute control signals without communicating with the ground. "Advanced computers and digital communications techniques developed since [the] Williams [patent]," said the Federal Circuit, "permit doing on-board a part of what Williams [*857] taught as done on the ground." n79 The Court concluded: "[P]artial variation in technique, an embellishment made possible by post-Williams technology, does not allow the accused spacecraft to escape the 'web of infringement.'" n80 Another case found a patented method for laying pipe, calling for a beam of light to align pipe segments, infringed by the use of later-developed laser beam technology. n81
One should note that these decisions, while we discuss them here under equivalents doctrine, come into conflict with the enablement principles discussed earlier. n82 If one adheres to the doctrine that limits claims to what is enabled by the disclosure, one would think that the doctrine of equivalents would distinguish between allegedly infringing devices that used "new technologies" basically to get around the claims from those that used the technologies to do something significantly better. In some cases, this distinction does not seem to have been made.
A recent case involving Texas Instruments' pioneering patent on the hand-held calculator shows the court applying the doctrine of equivalents in a way more consistent with the principles of enablement. n83 The Federal Circuit held that major improvements in all the essential elements of hand-held calculators rendered the improved devices [*858] noninfringing. n84 The specification supporting Texas Instruments' pioneer patent, for instance, described the use of integrated circuits containing bipolar transistors. n85 All of the improvements used integrated circuits having metal oxide semiconductor (MOS) transistors. n86
There were other improvements in the calculators as well. The improved calculators receive input via a device that scans the "matrix" under the keyboard at frequent intervals, whereas the original design had a conductive strip underneath the keypad. This is an example of an improvement that reduced the number of components in the invention. <a href="#n87 ">n87 [*859] Also, the original Texas Instruments display was shown in its specification as a small thermal printer that printed dots on a tape in response to output signals from the processor. The accused devices all used liquid crystal displays (LCDs), the familiar black-on-gray display that does not produce a paper copy, an example of an improvement that increases the efficiency of an individual component. n88 Finally, the internal processing elements of the original calculator were manufactured as discrete components that were electrically interconnected in the final design. The newer calculator, in contrast, had all their logic on one intergrated circuit, eliminating the necessity for many electrical interconnections. n89 This is an example of enhanced overall design.
The court concluded "that the total of the technological changes beyond what the inventors disclosed transcends . . . equitable limits . . . and propels the accused devices beyond a just scope" for the Texas Instrument patent. n90 Although the mode of analysis used in Texas Instruments -- described as the "as a whole" test for equivalents -- was apparently criticized in a subsequent en banc decision, n91 it has surfaced again in more recent cases, and so apparently still lives. n92 In any event this opinion is instructive for its focus on the merits of the accused device. As we note in the Conclusion, the opinion -- especially its emphasis on changes in materials, number and simplicity of components, and increased overall efficiency -- should serve as a model for applying the doctrine of equivalents. n93
[*860]
There are two more limitations on the doctrine of equivalents that should be mentioned: First, just as an applicant cannot claim anything in the prior art when applying for a patent, so are the courts limited by the prior art when "stretching" claim language under the doctrine of equivalents. n94 And second, the doctrine of "prosecution history estoppel" prevents the patentee from recapturing through equivalents claimed subject matter given up during prosecution. n95
2. Blocking Patents and Revese Equivalents. -- the doctrine of equivalents helps the patentee by expanding the scope of her claims beyond their literal boundaries. In a roughly symmetrical way, two similar devices are availabel to the accused infringer: blocking patents and the reverse doctrine of equivalents.
Two patents are said to block each tother when one patentee has a broad patent on an invention and another has a narrower patent on some improved feature of that invention. The broad patent is said to "dominate" the narrower one. In such a situation, the holder of the narrower ("subsevient") patent cannot practice he invention without a license from the holder of the dominant patent. At the same time, the holder of the dominant patent cannot practice the particular improved [*861] feature claimed in the narrower patent without a license. n96
It is of course preferable for an inventor to have her own patent free and clear of anyone else's claims. An inventor therefore will not often voluntarily characterize her invention as subservient. n97 But a court may do so in the course of litigation. Where the court upholds the validity of an accused infringer's patent on some enhanced feature, but nevertheless finds that the accused product infringes a prior, broad patent, it is in effect making the accused infringer's patent subservient to the broad patent. n98
[*862] Even where a court finds a patent subservient to another -- thus creating blocking patents -- the holder of the subservient patent is still better off than if she had never filed a patent at all, for two reasons. First, she can exclude the holder of the broad patent from practicing her improvement. Thus, although the improver may infringe the broad patent, she may gain some bargaining leverage by obtaining the subservient patent. n99 Second, because of this, she may be able to reduce the "lost profits" component of the dominant patentee's damages in an infringement action; the dominant patentee would not have replaced all the infringer's sales, presumably, because the infringer's sales were based at least in part on her improved feature. n100
We turn now to a doctrine that can much more effectively mitigate the impact of literal infringement: the "reverse" doctrine of equivalents. Courts have long recognized that, "[c]arried to an extreme, the doctrine of equivalents could undermine the entire patent system." n101 Scope could be enlarged so far beyond the literal language of claims that patents would take on unlimited power. To check the potentially destructive impact of the doctrine and to preserve symmetry in the rules on infringement, the Supreme Court long ago ruled that
a charge of infringement is sometimes made out, though the letter of the claims be avoided. The converse is equally true. The patentee may bring the defendant within the letter of his claims, but if the latter has so far changed the principle of the device that the claims of the patent, literally construed, have ceased to represent his actual invention, he is as little subject to be adjudged an infringer as one who has violated the letter of a statute has to be convicted, when he has done nothing in conflict with its spirit and intent. n102
An example, drawn from the case just quoted, may help to illuminate the doctrine. In 1869 George Westinghouse invented a train [*863] brake that used a central reservoir of compressed air for stopping power. Further advances in his design, primarily the addition of an air reservoir in each brake cylinder, resulted in a brake that was patented in 1887. An improvement on this 1887 brake, invented by George Boyden, added an ingenious mechanism for pushing compressed air into the brake piston from both the central reservoir and a local reservoir in each brake cylinder. n103 With the added stopping power of the Boyden brake, engineers could safely operate the increasingly long trains of the late nineteenth century.
The Westinghouse patent included a claim for "the combination of a main air-pipe, an auxiliary reservoir, a brake-cylinder, a triple valve [the device that coordinated the airflows from the main reservoir and the individual brake reservoir] and an auxiliary valve device, actuated by the piston of the triple-valve . . . for admitting air in the application of the brake." n104 The Court noted that the literal wording of the Westinghouse patent could be read to cover Boyden's brake, since it included what could be described as a "triple valve." n105 But it refused to find infringement on the ground that Boyden's was a significant contribution that took the invention outside the equitable bounds of the Westinghouse patent:
We are induced to look with more favor upon this device, not only because it is a novel one and a manifest departure from the principle of the Westinghouse patent, but because it solved at once in the simplest manner the problem of quick [braking] action, whereas the Westinghouse patent did not prove to be a success until certain additional members had been incorporated into it. n106
The Westinghouse decision has influenced a number of cases. n107 In [*864] SRI International v. Matsushita Electric Corp. of America, n108 the Federal Circuit reaffirmed the availability of the reverse doctrine of equivalents as a defense to literal infringement. The case involved a patent on a filter used to encode color information in a color television camera. The patent claimed a filter with two sets of parallel stripes of equal width "relatively angularly superimposed" over one another. n109 The accused device used a similar design to achieve the same result, but the stripes in its filters must be at forty-five degree angles to one another. n110
The court unanimously recognized the validity of a reverse equivalents defense:
The law . . . acknowledges that one may only appear to have appropriated the patented contribution, when a product precisely described in a patent claim is in fact 'so far changed in principle' that it performs in a 'substantially different way' and is not therefore an appropriation (reverse doctrine of equivalents). n111
But the court divided sharply on the issue of whether the defendant's camera filter was "so far changed in principle" that it was excused from infringement without more factual proof. n112 It remanded the case with explicit instructions for the trial court to consider the accused infringer's reverse equivalents defense.
These cases demonstrate the use of the reverse equivalents doctrine by the courts to limit the reach of a patentee's claims in the face of substantial technological improvements. However, use of the doctrine is fairly rare. n113 Issuance of an improvement patent, or a holding that a patent is valid but subservient to another patent, is much more [*865] common. n114
At first blush, the technical merits of the allegedly infringing device might seem to be irrelevant where literal infringement is concerned. After all, a patent is the right to exclude; an astoundingly meritorious improvement, while no doubt deserving a patent of its own, ought not escape infringement. The improver can patent the improvement, but this should not affect the original patentee's rights.
This is an appealing argument. An economic rationale for improvement patents would stress their tendency to encourage bargaining between improvers and original patentees. To the extent the improver has a very significant cost-saving technology, it would be in the interest of the original patentee to cross-license with the improver, to gain access to the improved technology.
Unfortunately, the original patentee may use her patent as a "holdup" right, in an attempt to garner as much of the value of the improvement as possible. n115 The chances of this being successful depend on the relative contributions of the original patented invention and the improvement to the "original plus improvement" combination. Where the original invention contributes most of the value, n116 or where the original and improvement inventions contribute roughly equal value, issuing an improvement patent may be a reasonable solution. But where the original patent contributes very little value compared to [*866] the improvement, the holdup problem may be significant. That is, the holder of the original patent may use it to extract much of the value of the "original plus improvement" combination from the improver. n117 The reverse doctrine of equivalents solves the problem by, in effect, excusing the improver from infringement liability -- and therefore removing the original patentee's holdup right. Reverse equivalents, of course, did not evolve in explicit recognition of this problem. n118 But [*867] the fear of the inefficient use of holdup power n119 does provide a rational account of the doctrine and might even assist courts in applying it. Note too that the same rationale could be applied to analysis of infringement under the doctrine of equivalents; the more significant the technological advance represented by the allegedly infringing device, the less willing the courts should be to find it an equivalent of the patentee's device. n120
To see when reverse equivalents might make sense, consider the problem we touched on earlier in the section on enablement: broad claims encompassing embodiments that can be made only after significant additional research is performed. The Westinghouse case is an example; [*868] Boyden's brake involved a triple-valve, and was therefore within the boundaries of the Westinghouse patent. The court, nevertheless, refused to find infringement, since Boyden's invention was "a manifest departure from the principle of the . . . patent." n121
II. THE ECONOMICS OF THE PATENT SYSTEM REVISITED
A. The Social Benefits and Costs of the Patent System
In most analyses of the different aspects of the patent system, concern has centered on a simple tradeoff. The analysis has concentrated on how changing patent coverage affects the balance between incentives to the inventor and underuse of the invention due to patent monopolies. Thus, Nordhaus's analysis of optimum patent life is concerned with the tradeoff between increased inventive effort resulting from longer anticipated patent life and greater deadweight costs associated with longer monopoly. n122 Kaplow uses these two variable to analyze the effects of allowing the patent holder greater freedom regarding licensing agreements. n123 Gilbert and Shapiro's recent work on optimal patent length and breadth builds on the tradeoff model, as does Klemperer's. n124
However, other analyses of the effects of the patent system open [*869] up a much more complex set of issues. These studies recognize that at any time many actors may be in the invention game, and that the game may have many rounds. This broader orientation brings into view the question of how the lure or presence of a strong patent can influence the multiactor portfolio of inventive efforts. n125 It also alerts the analyst to the possible effects of patents on the ability or desire of different parties to stay in the inventing competition over time, and on the efficiency of the inventive effort over the long run. n126
We believe that analysis of the effects of varying patent scope needs to recognize this dynamic multiactor context. One problem with the analysis of Gilbert and Shapiro, and Klemperer, is that this is not done. Both papers treat greater scope as roughly similar to greater duration in terms of its incentive effect on initial invention. n127 We have no real trouble with that. n128 Both treat the social costs of greater scope as precluding a wider range of substitutes covered by the patent. n129 Again, no real argument. However, they treat these substitutes as if they were already in existence or could be made so trivially. n130 It is [*870] is here that we find their analysis inadequate. Our concern is with the effects of patent scope decisions on whether or not, and if so how efficiently, these substitutes are created. More importantly, these papers for the most part ignore what we consider a critical set of "substitutes": subsequent inventions that not only substitute for the initial invention, but also improve on it in some way. Since some of the follow-on efforts of inventors could result in something not simply slightly different but significantly better than the patented technology, broad patents could discourage much useful research. Thus, these papers are not of much help in rationalizing and reforming those aspects of legal doctrine that apply to the economically significant class of improvement inventions.
The economic models that do try to encompass multiactor dynamics are quite stylized. In some, invention is analogized to fishing from a common pool. n131 There are many competitive inventors, and the first to make an invention gets the patent on it. Each knows that as others catch (invent) there is less in the pool for her. The result is "overfishing": too many people seeking inventions at once. n132 Other economists have modelled technical advance in terms of a multifirm "race to patent," in which many would-be inventors identify a particular goal, and the first to achieve the goal gets the patent. n133 A good deal of variation has been introduced into these models, with different assumptions being made about such variables as the strength of patents and the costs and benefits of innovating versus imitating. n134 Many of the implications of these models are sensitive to particular assumptions, but some are robust. In particular, under a wide range of assumptions rivalrous inventive efforts generate a great deal of inefficiency.
Despite the drawbacks of these models, the authors of this paper [*871] regard that basic conclusion as persuasive. Not only does proprietary control of technology tend to cause "dead weight" costs due to restrictions on use. (We presume here that in general it is not possible to write licensing agreements to completely offset this problem, a matter to which we will return shortly.) Where invention is rivalrous, the process leading to invention is itself inefficient. With exclusive property rights, we pay both kinds of costs in exchange for the benefits of technical advance. But recognition of the costs of rivalrous inventive efforts leads one to speculate about how these costs might be mitigated. This question is the source of Edmund Kitch's prospect theory.
1. The Prospect Theory. -- Edmund Kitch, in formulating his "prospect theory" of patent rights, moved beyond the static tradeoff model mentioned earlier and incorporated into his analysis some of the insights of the common pool models. Kitch analogized patents to mining claims. Like an exclusive claim to the minerals that may be produced from a plot of land, Kitch emphasized that patents are granted after invention but before commercialization. According to Kitch, this has two advantages: (1) it allows "breathing room" for the inventor to invest in development without fear that another firm will preempt her or steal her work; n135 and (2) it allows the inventor to coordinate her activities with those of potential imitators to reduce inefficient duplication of inventive effort. n136 This amounts to granting rights over an unexplored pool, with the right-holder being permitted to charge for access to various parts of the pool. n137 Thus the inefficiencies associated with rivalrous uncoordinated invention, as in the fishing or race models, can be avoided.
Kitch goes further in suggesting that the prospect theory "may clarify the process and conditions under which a monopolistic industry will be more efficient than a competitive one." n138 He states that this enhanced efficiency "turns not upon the size of the firm, but its dominance over a fruitful technological prospect." n139
Reacting to the inefficiencies highlighted by the fishing models, Kitch clearly has a preference for single-firm domination of a technological [*872] prospect. As Kitch recognizes, this can be achieved by licensing, where a number of firms hold patents on components of a key invention. Alternatively, one firm can hold a single dominant patent. In either case, the advantage seen by Kitch is that development is under the control of a single entity. Rivalry is avoided. Planning is possible.
We have trouble with the view that coordinated development is better than rivalrous. In principle it could be, but in practice it generally is not. Much of our case is empirical. But there are sound theoretical reasons for doubting the advantages of centralization.
For one thing, under rivalrous competition in invention and innovation there is a stick as well as a carrot. Block rivalry and one blocks or greatly diminishes the threatened costs of inaction. Kitch assumes a model of individual or firm behavior where if an action is profitable it will be taken, regardless of whether inaction would still allow the firm to meet its desired (but suboptimal) performance goals. Different models of behavior, like Simon's satisficing hypothesis, predict otherwise. n140 As we shall see, there are many instances when a firm that thought it had control over a broad technology rested on its laurels until jogged to action by an outside threat. n141
[*873] More generally, the model of behavior Kitch is employing ignores the limits on cognitive capacity and the tendency to focus on past experience that are characteristics of other models and of organizational behavior as we know it. n142 Once a firm develops and becomes competent in one part of a "prospect," it may be very hard for it to give much attention to other parts, even though in the eyes of others, there may be great promise there. n143 Again, our empirical explorations show many examples of this. Consequently, one might expect that many independent inventors will generate a much wider and diverse set of explorations than when the development is under the control of one mind or organization.
This flags still another limitation of the "pool" or "mining" models. In these models the "fish" or the "minerals" are out there and known (with perhaps some uncertainty) to all parties. But with the technological "prospects," and perhaps even real life mineral prospects, no one knows for sure what possible inventions are in the technological pool. n144 It is not even generally feasible to assign probabilities to possible outcomes on which all knowledgeable people will agree. Indeed different parties are almost certain to see the prospect differently. Because of this uncertainty, development of technology is critically different from other common pool problems. The real problem is not controlling overfishing, but preventing underfishing after exclusive rights have been granted. The only way to find out what works and what does not is to let a variety of minds try. If a property right on a basic invention covers a host of potential improvements, the property right holder can be expected to develop the basic invention and some of the improvements. But we would expect a single rightholder to underdevelop -- or even ignore totally -- many of the potential [*874] improvements encompassed by their broad property right. n145
Of course, Kitch's notions about how a broad patent prospect can be worked out by the patent holder do not preclude involving many minds. However, we regard as fanciful the notion that wider talent can be brought in without real competition through selective licensing practices. A substantial literature documents the steep transaction costs of technology licensing, n146 and there is indirect evidence that these costs increase when major innovations are transferred. n147 Moreover, various studies have indicated that transaction costs tend to be very high if licenses are tailored to particular licensees. n148 It is much [*875] simpler to grant roughly identical licenses to all who will pay a standard rate. In our own research, we have not found a single case where the holder of a broad patent used it effectively through tailored licensing to coordinate the R&D of others.
Although the preceding analysis applies specifically to patents, it is interesting that other types of property rights are limited in scope, perhaps for similar reasons. Kitch himself notes, for instance, that in mining law "[t]he mineral claim system restricts the area that can be claimed through rules that specify maximum boundaries in relation to the location of the mineralization," n149 and that boundaries are required to be clearly marked. n150 While there are no statutory limits on the number of claims an individual can make, n151 the law requires a claimant who has identified a mineral deposit n152 to work a claim actively before property rights will vest. n153 This places practical limits on the number and dimensions of claims. And state law, which establishes limited exclusive rights during the prospecting period prior to the grant of federal rights, similarly requires persistent and diligent work toward discovery on each claim for which protection is sought. n154
The obvious goal of these requirements -- to prevent hoarding and speculation -- is analogous to the goal of patent law doctrines designed to limit the breadth of patents. Both sets of rules recognize that although property rights assignments can make development of an asset more efficient, the scope of rights is crucial. Property rights that are too narrow will not provide enough incentive to develop the asset, while overly broad rights will preempt too many competitive development efforts. Kitch's prospect theory must be supplemented to take account of this important limitation on the breadth of property rights. n155
[*876] An interesting general point about the economic literature on property rights emerges from this analysis. Economists who theorize about property rights do not appear to have analyzed extensively the issue of how broad property rights should be. For the most part this work emphasizes the importance of defining property rights in the first place. n156 The usual assumption is that, with low or nonexistent transaction costs, the parties will bargain to a Pareto superior solution given any initial assignment of entitlements; therefore the "size" of the rights is not important. n157 However, as elaboration of the Coase theorem has made clear, the initial distribution of property rights can make a difference in the equilibrium level of output of the bargaining parties. n158 If one were to look at the patent scope problem from this view, one would conclude that the present authors favor a regime of property rights that limits the scope of a patent in such a way that inventors of significant improvements are in a strong bargaining position with respect to holders of broad patents. n159 This is not a particularly useful way of conceptualizing [*877] the problem, however, since unlike rights that somehow touch tangible property -- the usual subject of this analysis -- the allocation of property rights between technological pioneers and improvers is not a zero-sum game. n160
Undoubtedly our position is open to criticism. Rivalry no doubt causes waste. Yet we have little faith in the imagination and willingness of a "prospect" holder to develop that prospect as energetically or creatively as she would when engaged in competition. We are also skeptical about her ability to orchestrate development. Given the way humans and organizations think and behave, we believe we are much better off with considerable rivalry in invention than with too little. n161
Can we prove it? We can present empirical evidence that the granting of broad patents in many cases has stifled technical advance and that where technical advance has been rapid there almost always has been considerable rivalry. However, we grant that it is possible to see our evidence as not completely persuasive in this regard, or to posit that we have looked at only a few cases and that these might not be representative.
And even if our case is accepted that, up to a point at least, rivalry facilitates technical advance and unified control damps it, one can respond [*878] by saying "Yes, but what about the costs and the wastes?" We can rejoin that, in our cases at least, it is not evident that the waste caused by the presence of a broad patent is very considerable.
2. Clarifying Assumptions. -- Our argument rests on a simple premise: when it comes to invention and innovation, faster is better. What proof do we have that this is true? The answer lies in the work of scholars who have examined intensively the interrelation among research and development expenditures, invention, and productivity growth. Although there are still a great many unanswered questions in this field, n162 the following general points seem to be widely accepted: First, increases in research and development expenditures yield more inventions. n163 Second, the larger numbers of inventions from increased research and development have a positive effect on future productivity growth. n164 And third, productivity growth is important for economic well-being. n165
These findings support our argument only if there is a link between the speed with which innovations are introduced and the overall number of innovations. The research just sketched is of no help if the timing of an innovation is unrelated to the number of subsequent innovations. Here we must rely on simple economic reasoning. For the same reasons people prefer to have money in hand now, as compared to the same amount (and more, depending on the interest rate) later, society prefers to have improvements now, rather than later. Again, keeping in mind that at some point increased incentives for improvement will reduce the incentive to create a pioneer invention in the first [*879] place, the sooner improvements can be introduced the sooner the cost-saving (and welfare-enhancing) effects of those improvements will be felt by consumers. And, of course, the early availability of improvements will accelerate the pace with which second-and third-generation improvements -- that is, improvements on the improvements -- will be introduced. n166
But it is perhaps not enough to demonstrate the consistency of our thesis with basic economic concepts. Another means of validating our assumptions is by looking for consistency with the goals and purposes of the patent law itself. In its grant of priority to the first to invent; n167 its preference for an early reduction to practice; n168 and its provisions designed to encourage early filing of patent applications, n169 patent law favors not just invention, but early invention. Thus our concern that improvements be introduced as quickly as possible simply carries out a basic policy evident throughout the patent system. While there are those who may challenge the propriety of these goals from the standpoint of economic policy, it is both realistic and necessary for courts and the patent office to pursue the goals implicit in both constitutional and statutory provisions. Our assumptions, therefore, while not completely unassailable, are consistent with both economics and policy considerations.
In the remainder of this section we will consider several historical examples of how technical change proceeds in an industry. In the following section we explore industries whose development followed the different models, testing the relative efficacy of a pluralistic rivalrous system versus one in which technical advance is under the control of one or a few organizations. Our goal is to determine whether our theoretical understandings regarding the effect of patent scope in various industry types are consistent with the historical record.
[*880] B. Differences in Industrial Patterns of Technical Advance
We have noted earlier that, while most analyses of the effects of the patent system on invention assume implicitly that technical advance proceeds similarly in all industries, this assumption is mistaken: the pattern of technical advance varies significantly from field to field. One of the authors, Nelson, has concluded that at least four different generic models are needed. n170 The first describes discrete invention. A second concerns "cumulative" technologies. Chemical technologies have special characteristics of their own. Finally, there are "science-based" technologies where technical advance is driven by developments in science outside the industry. In each of these models patent scope issues take on a special form. In any industry one or another of these models may be applicable at any given time, or appropriate characterization may require a mix. But the mix differs from industry to industry, and so too, therefore, the salient issues involving patent scope.
What we call the discrete invention model corresponds to much of the standard writing about invention. It assumes that an invention is discrete and well-defined, created through the inventor's insight and hard work. In the standard discussions it may be recognized that the original invention can be improved, or even that improvement or complementary advances may need to be made if the invention is to be of much use. The basic invention may be amenable to tailoring for different uses or customers. But it is implicit that the invention does not point the way to wide ranging subsequent technical advances. It does not define any broad prospect. There are many inventions that fit this model, and these may be of considerable economic and social value. Two examples are King Gillette's safety razor n171 and the ball point pen, n172 and many new pharmaceuticals may also fit this model. n173 And in other industries technical advance appears largely to proceed through inventions of this kind. The consumer goods packaging industry is likely of this sort, n174 as is the toy industry. n175 For inventions and industries like these, while tight and broad control of a particular invention [*881] may enable a firm to profit handsomely, possession by that firm of a proprietary lock on the invention is not a serious hindrance to inventive work by many other firms. This stems largely from two features of these industries, one having to do with inventive inputs and the other with inventive outputs. As to inputs, discrete inventions do not typically incorporate a large number of interrelated components; they stand more or less alone. n176 On the output side, the products of discrete technology industries tend not to comprise integral components of some larger product or system; they therefore do not enable the development of a wide array of ancillary products.
However, in a number of technologies, the above characterization is quite inappropriate. In industries like those producing automobiles, aircraft, electric light systems, semiconductors and computers, technical advance is cumulative, in the sense that today's advances build on and interact with many other features of existing technology. n177 This by no means implies that technical advance is slow or inconsequential. Over time dramatic advance occurs in these technologies from improvements to one aspect or another, adding this new feature or that. n178 In many cases the technology in question defines a complex system with many components, usbcomponents and parts, and technical advance may proceed on a number of different fronts at once. n179 In these industries inventions may enhance some feature of a prior "dominant [*882] design," n180 or they may be incorporated into subsequent inventions, n181 or both.
There is much more at stake regarding allowed patent scope in these cumulative technologies than in those where inventions are discrete and stand separately. Particularly when the technology is in its early stages, the grant of a broad-gauged pioneer patent to one party may preclude other inventors from making use of their inventions without infringing the original patent. Two such examples are the Selden patent, which was used to control the development of automobiles, and Edison's successful attack on a broad patent covering light bulb filaments. n182 Thus, a broader pioneer patent may give one party legal control over a large area. Alternatively, in multicomponent products, broad patents on different components held by several inventors may lead to a situation in which no one can or will advance the technology in the absence of a license from someone else. As we shall see, these are not just theoretical possibilities; they describe the development of several important technologies.
Despite the nature of technical advance in cumulative-technology industries, improvement patents (discussed earlier) are no more common in these industries than in others. n183 This is because an improvement patent is undesirable for the reasons discussed above, n184 and because patent lawyers prefer to claim a new or improved component or subcomponent as a distinct product. Accordingly, it is important not to confuse the patent-law concept of an improvement patent with the commercial reality that, in some industries, technical advance proceeds cumulatively, i.e., via a series of improvements.
Technical advance in the chemical industries has some attributes that fit the discrete invention model, some that fit the cumulative technologies model, and some particular characteristics of its own. A new chemical product is in most cases a discrete entity, or it may encompass a particular class of products, like penicillin. But particular chemical product innovations seldom are the keystones to the development of large numbers of other chemicals. Although there are recognizable families of chemical products, the invention of one chemical species seldom gives more than general guidance in the development of other species. This is primarily a function of the complex and unpredictable [*883] relationship between chemical structure and function, most clearly evident in the pharmaceutical industry. n185 Sometimes, however, a new chemical entity turns out to have a wide variety of applications. n186 Because of this, not every chemical product invention shares all the features of a true discrete invention. At the same time, chemical processes tend to be improved along the lines of the cumulative technology model, and licensing and cross licensing are well-established practices in these industries. This tradition of licensing mitigates the potential impact of broad patents. As a result of these special features, scope decisions affect the chemical industries differently from others, a point we return to in Part III.
An invention in any of the three regimes described above may be assisted by recent developments in science. But technologies whose advance is predominantly driven by such developments, while rare, warrant special recognition. In these science-based technologies, n187 of which modern biotechnology is a prominent example, research and development efforts attempt to exploit recent scientific developments. n188 These scientific developments tend to narrow and focus perceived technological opportunities in the industry and concentrate the attention of inventors on the same things.
Such science-based technologies warrant analytic distinction for [*884] several reasons. In the first place, this is a context that engenders inventive races of the sort described earlier, particularly if it is anticipated that the first to apply a scientific finding will get a patent of considerable scope. Many are rushing toward the same objective that all see as feasible and several will get there, but only the first receives a patent. Second, new scientific and technological developments "in the air" open the possibility of a major advance over prior practice, and the contribution made by the individual or firm who first makes these possibilities operational may be relatively small. n189 The invention may diverge from "prior art," in the sense of actual technological accomplishments, and sweep the market, yet still be only a successful application of knowledge that is apparent to the scientifically sophisticated. n190 When this is a possibility, the patent system should be particularly careful in awarding patents of broad scope. Third, and this is where our focus will be, there is a real danger that allowing patent scope to be overbroad may enable the individual or firm who first came up with a particular practical application to control a broad array of improvements and applications.
We now turn to a more detailed discussion of these models of technical advance, with an eye toward what they can teach us about the effects of patent scope.
III. EFFECTS OF PATENT SCOPE IN VARIOUS INDUSTRIES
Because we are concerned with the effects of patent scope decisions on the subsequent development of technology, we are not interested in the cases of discrete invention. We deal with what we have called cumulative technologies, chemical technologies and science-based technologies, in that order.
A. Cumulative Technologies
We have asked two questions about the effects of broad patents on cumulative technologies. One concerns the consequences of "pioneer" patents. We wish to test the validity of the hypothesis that the granting of broad patents is likely to make subsequent invention and development more orderly and productive. The second question is how the presence of broad patents on components of a cumulative technology affects subsequent development.
One must keep in mind, however, what we are not testing. We do not [*885] ask whether any patent should have been granted in the following cases. We take it as axiomatic that some degree of patent protection is necessary and desirable. And we do not ask whether the scope of the patents discussed should have been limited to the precise embodiments the inventor had developed when the patents were filed. We accept that patents claiming the general inventive principle were justified; and we focus on the impact of broad scope on the environment for subsequent development and improvement.
1. Electrical Lighting Industry. -- The chain of reasoning in our critique of the prospect theory, and our view of the patent system, is consistent with most of the historical evidence on cumulative technologies. The early electrical illumination industry illustrates this most clearly.
Patents played a very important part in this industry from the beginning. In the field of incandescent lighting, Edison's early patent gave his company, later General Electric, a dominant position. But in certain other sectors, most notably arc lighting and the production of dynamos, efforts to establish dominance via a single broad patent failed. The contrast between these sectors, where entry was easy and competition for improvements was intense, and the incandescent lighting field is noteworthy for our purposes. Most importantly, the history of the early electrical industry supports the notion that broad pioneering patents can play a pivotal role in the evolution of industry structure.
No single patent better illustrates this than Edison's U.S. Patent 223,898, issued in 1880. This was "the basic patent in the early American incandescent-lamp industry," covering the use of a carbon filament as the source of light; n191 it proved to have a profound effect on the industry until it expired.
Although the Edison General Electric Company had some difficulty establishing the validity of its basic patent, once it did the industry changed drastically. In 1891, U.S. Patent No. 223,898 was held valid and infringed by a competing design. n192 General Electric officials then quickly obtained a series of injunctions that shut down a number of competitors. n193 As the aptly-named industry historian Arthur Bright stated, "For twelve years [after the issuance of the 223,898 patent] competition had been possible; it suddenly became impossible." n194 The company's market share grew from 40 to 75 percent; entry into the industry slowed from 26 new firms in 1892 to 8 in 1894, the last year of [*886] the patent's life; n195 and the steady downward trend of lamp prices slowed until the patent expired. n196
More importantly for our purposes, the validation of Edison's broad patent slowed the pace of improvements considerably.
Even as the courts were passing on the Edison lamp patent in 1891, the Edison General Electric Company . . . . [recognized that it] gradually had been slipping backward in its commercial position, particularly since 1886 . . . . Its technological contributions were becoming relatively smaller than they had been during the early [eighteen] eighties. n197
This was especially true in Great Britain, where the Edison Company's patent position was even more commanding, due to its control of a basic patent on a process for producing carbon filaments. A series of court victories over its largest competitors gave the British "Ediswan" company "a practical monopoly of incandescent-lamp production." n198
Given the lack of competition, it is perhaps not surprising that the pace of technical advance slowed. According to the historian Bright:
After the introduction of the incandescent lamp and its first rapid changes . . . . the Edison Electric Light Company did not introduce many important new developments. Edison himself turned to other problems, and the company's technical leadership in incandescent lighting was not revived until after the merger [that formed General Electric in 1896]. n199
Prior to the enforcement of the patent, Edison's competitors were quickening the pace of technical advance:
Despite the improvements in the Edison lamp, a number of its competitors had improved their lamps even more rapidly . . . . Efficiency advantages permitted many of the other American concerns to compete very successfully with the Edison lamp after 1885 . . . until the corporate reorganizations and the establishment of patent supremacy regained for the Edison lamp commercial supremacy as well. n200
The same was true overseas: "In England, filament improvement was almost entirely halted during the period of Edison patent monopoly [*887] from 1886 to 1893." n201 Bright concludes:
The lengthy and expensive patent struggle in the lamp industry from 1885 to 1894 was a serious damper on progress in lamp design, although process improvement continued. The Edison interests concentrated on eliminating competition rather than outstripping it. . . . After 1894, when it was no longer protected by a basic lamp patent, General Electric devoted more attention to lamp improvement to maintain its market superiority. n202
Thus the broad Edison patent slowed down progress in the incandescent lighting field. The lesson, however, is not that this patent should not have been granted. It is rather a cautionary lesson: broad patents do have a significant impact on the development of a technology and hence on industry structure, and this should be reflected in those doctrines that collectively determine patent scope.
Two other sectors of the electrical industry -- ones where broad patents were invalidated -- demonstrate what can happen in the absence of dominant patents. The first was in the production of power generation dynamos, where the Brush Company attempted to establish patent dominance. Brush, together with several other companies, acquired a patent they thought "would give absolute control of all dynamo manufacture in the United States." n203 But the courts thought otherwise; the patent was found to have lapsed when a foreign counterpart patent reached the end of its term. n204 As an historian of the industry describes it:
The effect of the decision was to free the dynamo from patent control. Anybody could manufacture it. It was only minor details in dynamo design and construction -- such as particular coil windings or commutators -- which were patentable. n205
Because there were no broad patents to discourage entry, entry was easy and competition for improvements was intense. n206
[*888] The second failed attempt to establish dominance by way of a broad patent also involved the Brush Company. In 1884 the company brought a test infringement case involving its basic patent on a double-carbon arc light. n207 Unfortunately for the Brush Company, the court found that the defendant's design did not incorporate a key feature of the Brush patent, and thus held that there was no infringement. n208 This ruling had an important effect: "[p]atents were consequently not a handicap to entry into the industry. Firm after firm was organized to manufacture its own arc-lighting system. At one time, nearly fifty different firms were making arc-lighting equipment." n209
2. Automobiles and Airplanes. -- We move now to two infamous cases regarding pioneer patents: the Selden patent in the development of automobile technology, n210 and the Wright patent's influence on the growth of aircraft technology. As we have seen, the Selden patent claimed a basic automobile configuration, one using a light-weight internal combustion engine as the power source. The Wright patent was on a broadly defined airplane stabilization and steering system. In both of these cases, the holders of the pioneer patent engaged in extensive litigation against companies that did not recognize the patent, n211 and [*889] the Wrights refused to license theirs. n212 Our question is how the presence of these patents affected the evolution of the technologies.
The Selden patent had as its key claim the use of a light gasoline-powered internal combustion engine. This claim was extremely broad and covered a myriad of possible embodiments. n213 Contrary to the prospect theory, however, neither Selden nor his assignee used the patent to orchestrate the efficient improvement of automobile technology; there was no policy of "developing the prospect." n214 They were willing to license anyone who would acknowledge the validity of the patent and pay royalties; to this end they formed the Association of Licensed Automobile Manufacturers. n215 But the Association's purpose was to collect royalties, and perhaps control competition in the industry, n216 rather than to facilitate orderly technological development.
[*890] But did the presence of the Selden patent actually hinder technological progress in the industry? That is perhaps a bit more speculative. Law suits based on it surely did absorb considerable time and attention of people like Henry Ford, whose production methods revolutionized the industry. n217 Perhaps more importantly smaller firms may have been put off by the threat of suit. At this early stage in the history of the technology, those that left the industry or chose not to enter may well have taken valuable improvements with them.
An interesting result of this experience with patent litigation was that, even before the Selden patent was pruned back in 1911, n218 the automobile industry, through the Association, developed a procedure for automatic cross licensing of patents. While formal agreements to cross license all new patents no longer exist, the practice of relatively automatic cross licensing has endured to the present.
The Wright brothers patent is different in a number of regards. First of all, the achievement described in the patent -- an efficient stabilizing and steering -- system was in fact a major one, and it did enable a multiplicity of future flying machines. n219 Second, the Wright brothers were very interested in producing aircraft and in improving their design, and they did so actively. However, there were other important people and companies who wanted to enter the aircraft design and manufacture business. They had their own ideas about how to advance the design of aircraft, and they strongly resisted being blocked by the Wright patent. In this case, and others, it turned out to be extremely difficult to work out a license agreement that satisfied both the holder of a broad patent and an aggressive potential competitor who believed that there was a lot of his own work in his design. The early attempts by the Wright Brothers and Glenn Curtiss, who was the most prominent such potential competitor, came to naught. Litigation followed. n220
There is good reason to believe that the Wright patent significantly [*891] held back the pace of aircraft development in the United States by absorbing the energies and diverting the efforts of people like Curtiss. The aircraft case is similar to that of automobiles in that the problems caused by the initial pioneer patent were compounded as improvements and complementary patents, owned by different companies, came into existence. The situation was so serious that at the insistence of the Secretary of the Navy, during World War I, an arrangement was worked out to enable automatic cross licensing. n221 This arrangement, like the licensing of automobile patents, turned out to be a durable institution. By the end of World War I there were so many patents on different aircraft features that a company had to negotiate a large number of licenses to produce a state-of-the-art plane. n222
3. Radio. -- The case of radio in the United States warrants at least a brief recounting, for it is an excellent example of what happens when several companies each hold patents of broad scope. The earliest radio patent was a broad patent granted to the British inventor Marconi in the field of radio transmission. n223 Marconi also invented and acquired rights to the basic technology for tuning, n224 which he controlled until 1914, and the basic Fleming patent on the two element vacuum tube, or diode. n225 These patents helped the Marconi Wireless and Telegraph Company establish an imposing presence in the early radio industry, which was dedicated primarily to large-scale commercial uses such as ship-to-shore communications.
AT&T, as part of its radio operations, n226 acquired rights to two very fundamental patents on the triode vacuum tube, an early radio wave amplification device patented by Lee De Forest. n227 While technically [*892] only an improvement on Marconi's diode, the triode was in fact a very significant advance; it was called "the heart and soul of radio." n228
Several other firms had important patent positions. General Electric entered radio as a natural extension of its expertise in electricity generating systems. It controlled the important Alexanderson patents on the electric alternator, the signal generation invention that made long-range transmission possible. n229 Westinghouse also joined the industry, mostly on the strength of patents on receiving technology, which served as the basis of the firm's successful entrance into the inexpensive home receiver market. n230 Other companies also held American rights of varying breadth over other important radio technologies. n231
The situation soon became similar to that in the aircraft industry, where different companies could block each other from using key components. A good example is the deadlock between the Marconi Company and the De Forest interests, a classic instance of blocking patents. Marconi's diode patent was held to dominate De Forest's patented triode, n232 yet neither party would license the other. n233 As a consequence, no one used the admittedly revolutionary triode for a time. This is a good example of a case where the reverse doctrine of equivalents might have been invoked to permit De Forest to practice his improvement. n234
In at least one other area as well -- long-range transmission -- licensing proved difficult. n235 The upshot was that no one could produce state-of-the-art radio technology without being threatened by litigation. n236 Radio is thus a canonical instance where the presence of a [*893] number of broad patents, which were held by different parties and were difficult to invent around, interfered with the development of the technology. n237 The various pioneers formed RCA to break the deadlock; the new company promptly acquired the American rights to the Marconi patents. n238 The companies that owned most major radio patents became RCA shareholders. n239 With all the constituent radio technologies under one roof, RCA established itself as the technical leader in radio and dominated its advance for many years. n240
As shown by the cases of autos, airplanes, and radios, the many early inventors in cumulative technologies often perform overlapping research. This may lead to blockages unless basic patents are not present, or routine licensing and cross licensing is instituted. We next consider the former possibility by describing important post-World War II technologies that have advanced rapidly because no one held a pioneer patent that was used to restrict access. At the end of this section, we discuss the licensing solution and its impact on our analysis of patent breadth.
4. Semiconductors and Computers. -- Semiconductors are a good example of a technology that developed without patent blockages. There are two instances in the history of this technology where a broad-gauged patent was issued which could have given its holder control over a large "prospect," but in fact did not. One involved the initial transistor patents held by AT&T. Because of an antitrust consent decree, [*894] AT&T was foreclosed from the commercial transistor business. n241 Some have argued that it is not clear whether AT&T would have gone into the merchant transistor business even in the absence of a consent decree. n242 In any case, given that it was not going to do so, AT&T had every incentive to encourage other companies to advance transistor technology because of the value of better transistors to the phone system. AT&T entered into a large number of license agreements at low royalty rates. n243 Many companies ultimately contributed to the advance of transistor technology because the pioneer patents were freely licensed instead of being used to block access.
The second instance involved the parallel inventions of the integrated circuit (by Texas Instruments) and the Planar process for producing them cheaply (by Fairchild Instruments). Both of these companies obtained patents on their own inventions, which meant that each had to license the other to produce integrated circuits effectively. n244 Cross licensing was favored by the government; the Department of Defense, which for some time had provided the lion's share of the market for semiconductors, had a strong interest in seeing these important technologies become broadly available throughout the industry. n245 Again, the absence of a single, broad patent assisted the rapid development of an industry.
The second recent cumulative technology developed without strong, broad patents is electronic computers. Although original computer inventors Eckert and Mauchley did file for and receive a patent on their basic ENIAC design, the patent was ruled invalid because of a judgment that the prior art included much of what they claimed. n246 Since this ruling, patents have played only a very minor role in the computer industry, and where patents are concerned, cross licensing is common. n247 As a result, the pace of technical change has been rapid. n248
5. Licensing and Cumulative Technologies. -- In many of the cases we have examined, licensing and industry consolidation emerged as solutions to patent blockages caused by patents. This would appear to have a bearing on our study. Does the consolidation of the radio industry in RCA, for example, support the position that development would have [*895] been more efficient had control been in the hands of one party from the beginning, in the form of one super-patent? Or does it imply that patent breadth was irrelevant -- consolidation would have happened even with narrow patents?
The first possibility seems remote, and there is indirect evidence that the second is wrong as well. The fact that many inventors and firms made important advances in various components of radio technology indicates that no one firm had the inventive firepower to develop radio on its own. And there is no reason to believe that one firm could have orchestrated the development of the technology, since there was no way to know in advance which inventors would cultivate expertise in each component, or which inventor's approach would work. There were no "proven" experts in transmission or reception that a firm could have granted licenses to, for example; experts emerged only when their inventions turned out to work. n249 And it would have been impossible to identify all the potential experts, since everyone was working on the various components simultaneously. n250 In any event, the inventive scramble that in fact resulted, while by no means optimal, did result in the fairly rapid commercialization of a complex, multicomponent technology. It also resulted in a patent tangle, one that might have been lessened if some of the key patents had been narrower. But it is difficult to see how a single broad patent would have led to more rapid commercialization. The ex post consolidation, in other words, simply does not imply that a broad ex ante "prospect" would have been effective in this case. n251
As to the second objection to our analysis -- that the radio industry would have consolidated regardless of patent breadth -- two points [*896] seem relevant. First, narrower patents might have made consolidation unnecessary. If one or more firms could put together a complete radio system without infringing any patents, consolidation would not have been essential, at least for patent-related reasons. One candidate is General Electric: the only essential component for which patent blockage was a problem was the triode; n252 if De Forest's patent had been narrower, n253 or if inventor Edwin H. Armstrong had won his interference with De Forest, n254 General Electric might have put together a noninfringing system.
Second, even if narrower patents would not have prevented the deadlock, they might have helped break it sooner. Perhaps without the value of a "holdup right" on an essential component of radio technology, the firms would have been content to contribute their patents to a pool and compete on the basis of improvements and price competition. n255 Finally, even if this industry eventually consolidated into one firm, there is no promise that all industries will do so. n256
There is also no reason to assume that when blockages arise industries will always turn to the deadlock-breaking solutions we have seen, patent pooling and cross licensing. Though we saw the emergence of cross licensing among aircraft manufacturers, the impetus was war-time government pressure. n257 In the case of the light bulb industry, the government stayed out; when the firms finally pooled technology, it was only to effectuate the operation of a cartel. n258
There is therefore no guarantee that pooling, cross licensing, or consolidation will always emerge to break an industry impasse. And without these solutions there is nothing to mitigate the effect of broad [*897] basic patents in cumulative technology industries. Earlier we saw that theory offered a number of reasons to be concerned about these patents. The historical evidence available is consistent with this theory. In most instances this evidence can be read as supportive of our concerns about the effects of broad patents on cumulative technology industries. Chemical industries, which we turn to next, tell a different story.
B. Chemical Industries
Chemical industries produce an incredibly diverse range of products, from bulk chemicals like sulfuric acid, to synthetic materials like plastics, to pharmaceuticals. n259 Despite the diversity of products, however, invention in the chemical industries shares several key attributes. To a large extent chemical product invention tends to fit the "discrete invention" model described earlier. n260 Thus product patents tend to define a well delineated class of substances. n261 Valium is valium and, although subject to some variation, sulfuric acid is sulfuric acid. However, research and development on new chemical products is subject to an unusual degree of uncertainty and costly experimentation, both because it is difficult to predict the precise chemical structure needed to achieve a given end and because the effects of using a new chemical substance in a particular way can be startling. n262 Further, once a new product or use is discovered, it is easy for a competitor to replicate. [*898] Thus patent protection on products or novel ways of applying them is vital if the inventor is to reap returns. n263
In contrast with product technology, most chemical production processes evolve cumulatively in the sense discussed earlier. The first versions of new chemical processes tend to be amenable to a wide range of improvements. Thus one might expect to see the same kinds of problems regarding chemical process patents as we have seen in our examination of other cumulative technologies in the section above.
To analyze the importance of process and product inventions in the chemical industries, it is helpful to disaggregate those industries into three groups: bulk chemicals, synthetics, and pharmaceuticals.
1. Bulk Chemicals. -- Bulk chemicals consist of products like sulfuric acid, ammonia, ethylene, and other substances that have been known and widely used for some time. Many are natural substances. In any case there are no effective product patents on bulk chemicals. n264
As a consequence, most research and development is concerned with creating new or improved processes. The development of chemical process technology tends, as noted, to be cumulative; at any time there tends to be one process that is the dominant mode of production. From time to time a dominant process is superseded by a new one. And the early patent or patents on that new process have the characteristics of "pioneer" patents. However, these patents have not generally been used to control subsequent development, which by and large has proceeded with multiple sources of initiative. n265 This is due primarily to the inherently limited power of control conferred by patents in the bulk chemical field. n266 Pervasive cross licensing in chemical industries confirms this. n267
[*899] Thus, until 1861 the Leblanc process dominated the production of alkalis. n268 This process was widely licensed and a number of different companies contributed to its improvement. n269 When, in 1861, the Solvay process was developed and patented, the original patent holder also had a chance to control future development of the process. n270 However, here too the policy of the original patent holder was one of reasonably wide licensing of the basic patent. A number of different companies made improvements; these were also cross-licensed. n271
Of course there are patent suits and short-term hold ups in the field of bulk chemical process technology, but these problems are usually settled and licensing is a general practice. n272 The recent development of new processes for making acrylamide is a good example. Acrylamide is an organic chemical commonly used to make polymers for "water treatment, pulp and paper processing, textile treatment, food processing and other applications." n273 Until the 1960s it was made in a two-step process using sulfuric acid and ammonia. In the mid 1960s, researchers at several different companies all began investigating ways to improve the traditional process. n274 Both Standard Oil and American Cyanamid came up with processes using copper as the catalytic agent. Dow Chemical also made several patentable inventions in this field. n275
Lawsuits were filed. Standard Oil sued American Cyanamid arguing that American Cyanamid's process infringed Standard Oil's patent. The court ruled against Standard Oil in this case. n276 On the other [*900] hand, Dow successfully sued American Cyanamid for infringing its patents. n277 However, after this round of legal scuffling, the companies cross licensed each other. No single company tried to hold to itself the right to use the new technology or control its future development. n278
In short, the pattern of development in bulk chemical process technology is similar to several of the cases of cumulative technologies considered earlier. It is sometimes possible to obtain a fairly broad patent when a new technology is invented. n279 This has the potential to give its holder a measure of control over subsequent development. However, by and large the chemical companies have not used their patents that way, partly under the pressure of competing inventions. These firms choose instead to license or cross license. Thus several companies tend to be involved in the subsequent development of the technology.
2. Synthetics. -- Product patents are slightly more important in the field of synthetic materials, where they are sometimes connected closely with process inventions; research on a new process for making an established product may yield a distinct and patentable version of the product. n280 But just as in bulk chemicals, reasonably liberal licensing is common in the synthetic chemical industry. n281
When Du Pont wanted to enter the business of producing Rayon it took out licenses on the product and the key processes from the French firm that held them. n282 Du Pont similarly took out a license on Cellophane technology. n283 Subsequent research and development at Du Pont on both of these products significantly improved them. n284 In turn, Du Pont licensed Nylon to both Imperial Chemical Industries of Great Britain and IG Farben of Germany. n285 Both of these companies [*901] later came up with variants on the original Nylon.
The fact that product patent claims are narrowly bounded keeps the advance of synthetic material technology competitive. Thus Du Pont's Nylon provided a superior alternative in many uses to the earlier Rayon. n286 And newer fibers like Dacron and Orlon subsequently replaced some of Nylon's market. n287
Another good example of the interdependence of product and process technology in synthetic materials is the effort to develop an improved process for the manufacture of polyethylene. Research teams at several firms worked on this project simultaneously. In the 1950s researchers at the Max Planck Institute, led by a chemist named Karl Ziegler, invented a superior process, based on a new understanding of catalytic compounds. n288 Not only was the new process patentable, but due to the relatively restrictive claims on the older polyethylene patent, the product it produced was outside the scope of Imperial Chemical's basic patents. n289
In turn, work by an Italian chemist, Giulio Natta, led to significant improvements in the Ziegler process. Natta's group also discovered a way to produce polypropylene, another important polymer. n290 Groups at other companies and research institutes were following the same trail. At least five different companies filed product patents on a version of polypropylene between 1953 and 1955. n291
Needless to say, the customary round of law suits resulted, and [*902] dragged on for some time. n292 However, the result was not that a single company controlled the basic technology and improvements, but rather a series of cross-licensing agreements which kept the technology open to a number of firms. n293
3. Pharmaceuticals. -- We turn now to the two matters regarding the scope of chemical patents which are especially important in the pharmaceutical industry: what to do when someone discovers a new use for an established product, and how to treat a process invention that yields a much purer form of a natural substance than was available earlier.
Earlier we observed that chemical products have a surprising range of uses. Often some of these cannot be foreseen when a product is invented and patented. In a number of cases researchers looking for a [*903] way to meet a new need will discover that an old product can do the job. In other cases, the discovery of a new use may be accidental -- a byproduct of looking for something else. In either case, this is important inventive work that ought to be encouraged and rewarded. How to do this? The Patent Office and the courts have been struggling with this issue for some time.
The general solution has been to award a process patent to the discoverer of a new use. We cited a nonpharmaceutical example earlier, the case of Rohm & Haas v. Roberts Chemical Co.. n294 In this case the defendant's patent on use of a well-known product as a fungicide was upheld because this use was not anticipated or claimed in the original patent. n295 This process patent would not enable the patent holder to produce the product in question, but rather only to control its new use. If the use is an important one, such a process patent can provide a substantial reward to the patent holder. This is an important doctrine in the pharmaceutical industry, where new uses are often discovered. n296
Another special problem that crops up in the chemical patent field involves inventing a synthetic version of a substance found in humans or animals. Typically the discovery involves enhancing purity or lowering cost. Today this issue arises mainly in the field of biotechnology, but the problem has existed for some time. Thus in 1911 Learned Hand upheld a product patent on purified human adrenalin made via a new process. n297 The patent was not simply on the process, but also on the purified natural substance.
The problem with this practice is that it grants patents of unnecessarily wide scope. The adrenalin patent would be infringed by the use of a radically different, and better process for making the same natural product unless the characteristics of the product were judged substantially different. Yet the argument is not convincing that what the original inventor invented was the product, in addition to her particular process for making it.
The recent case involving Genentech, which we mentioned earlier, illustrates the issue. Genentech had invented a recombinant DNA method for producing the human blood clotting protein factor [*904] VIII:C. n298 That process had major advantages over an earlier, patented technique of purifying the substance drawn from natural blood. Genentech's process was not only better; it was completely different. Yet in the first part of the case, the court upheld the earlier patent, held by the Scripps Institute, on the ground that it was a legitimate product patent and thus Genentech's new method of producing it was an infringement. n299
In a later ruling, the court invalidated the Scripps patent, saying that it did not adequately disclose the purification method that Scripps itself judged best. n300 But the court did not retract its earlier judgement that a product patent was quite legitimate in this case. We think this is unfortunate social policy. It might well inhibit technical advance in biotechnology, where much invention involves improving ways to produce purified natural products. If the initial patent is granted on the product, rather than the process for making it, subsequent process research by others will be discouraged. This is a good example of a prospect that will likely reduce competition for improvements. While licensing by firms can mitigate this problem, there is no guarantee that this will take place at such an early stage in the industry. n301
The doctrine of reverse equivalents might be employed to limit the blocking power of product patents in appropriate cases. Under this doctrine a court could rule that an important process invention yielding a more purified form of the product escapes infringement. Although compulsory licensing might be another option, our law does not for the most part permit judicially mandated licensing. n302
C. Science-Based Industries
Earlier, we discussed what makes an industry science-based. n303 Perhaps the most dramatic contemporary example is the biotechnology industry. Scientific advances, especially in molecular biology and biochemistry, created this industry, and continue to feed it ideas, theories, discoveries, and techniques. n304 Other examples include the chemistry [*905] of catalysis and semiconductors during the 1950s, n305 and the burgeoning new field of superconductivity. n306 Because science-based industries rely so heavily on scientific discoveries, one relevant patent issue is the appropriate scope of patents in the face of the (usually published) science that makes invention in these industries possible.
The modern biotechnology industry is built around two different sets of technologies: recombinant DNA and monoclonal antibodies. Both of these are based on prior, more general advances in molecular biology and both were initially discovered and employed by scientists concerned with pure research. One of these technologies was originally developed in 1975 by Kohler and Milstein, who discovered that individual immune system cells, which generate antibodies to a specific antigen, could be fused with immortal cancer cells, to create a small "factory" for producing antibodies. n307 They did not take out a patent on their discovery. They were awarded a Nobel prize. n308
The pathbreaking Kohler-Milstein research almost immediately was recognized as opening up a myriad of commercial possibilities. n309 Hybritech was an early entry into the race to develop applications. It was the first to use monoclonal antibodies in diagnostic kits sold to doctors and hospitals to identify the presence of diseases (e.g., AIDS) or heightened hormone levels (e.g. pregnancy tests). It received a patent covering this whole family of diagnostic kits.
Other companies saw exactly the same opportunity, if not so quickly. Monoclonal Antibodies, Inc. was one of these, and it created a similar technique after Hybritech. Monoclonal Antibodies, Inc. made and sold these kits, and Hybritech sued. n310 Monoclonal defended by claiming the Hybritech patent invalid, at least in its broad scope, because given the work of Kohler and Milstein the generic technique was obvious. The trial court recognized the argument, and acknowledged:
[T]he major advance was the invention of Kohler and Milstein [*906] in the making of monoclonal antibodies . . . . Onece the scientific community had the monoclonal antibody it was obvious and logical to those expert in the field to use them in known assays as substitutes for . . . polyclonal antibodies . . . of inferior qualities. n311
However, on appeal, the patent was held valid. Granted, the call was not an easy one: Hybritech clearly invented something. The question was, given that it was building on public science, what was the limit of its contribution? The Patent Office allowed Hybritech a broad prospect and the court concurred.
While a case has not come to court yet, the Patent Office also allowed Genentech a very broad prospect on the second major technology of the new biotechnology industry, expression of recombinant proteins. The basic genetic technique was developed earlier by two scientists, Cohen and Boyer. n312 The two scientists involved saw their basic technique -- the insertion of a specific gene into a host cell and subsequent expression of the protein product for which the gene codes -- primarily as a contribution to ongoing public science. Their universities urged them to take out a patent, which they did, but the patent is licensed to all comers. n313
Genentech's patent is an extension of the Cohen-Boyer work. n314 It covers the basic technology of gene expression, where the firm clearly made a major early contribution. n315 Their contribution was to refine existing gene expression techniques to achieve the first successful expression of a human protein in a bacterium. n316
In their specification, the inventors describe one particular technique for expressing and recovering proteins and apply this technique to the production of two polypeptides. n317 The technique disclosed in the patent no doubt legitimately covers many more specific embodiments than those expressly disclosed. But it can be argued that they simply were the first to bring to practice techniques that persons "skilled in the art" knew could be made to work. It is difficult to tell yet whether the breadth implicit in this patent will hold up, but it has created a good deal of trepidation in the industry. n318 According to the [*907] head of a rival biotechnology firm, "If interpreted most narrowly, there are certain bacterial [production] systems that wouldn't even be covered. If interpreted most broadly, it could cover all production systems in bacteria, yeast and cells." n319
Fortunately for the industry, an even broader patent on gene expression was rejected on obviousness grounds because several of the inventors published results prior to the invention. n320 The investigators had discovered that a gene for a non-operational protein taken from a frog could be inserted into a bacterium and expressed there. n321 On the basis of that research they filed a patent claiming a process for producing proteins comprising "linking a natural or synthetic heterologous gene [i.e., one from a foreign source] . . . to [an] indigenous [bacterium] gene portion." n322 It is worth noting that there is no indication that these claims would have been rejected because of their breadth. Thus if the prior publication had not been judged to render the claimed invention obvious, it might have received a patent. Judging from the quoted claim language, this would have been very broad indeed.
Biotechnology is not the only industry where scientific breakthroughs [*908] sparked a scramble to obtain broad patents. The current rush to obtain patents over superconductors demonstrates that patent positioning often is important at the birth of science-based industries. n323
As a new science-based technology matures, the issues relating to patent scope change largely because particular technologies become established. Thus the early work on catalysis was science based. But as catalysts were developed, further innovation became more cumulative than science-based. Step-by-step process improvements now dominate the field, succeeding the early advances that came quickly on the heels of the Ziegler and Natta research. As a result, the issues involved in setting appropriate patent scope change as an industry advances.
V. SUMMARY AND CONCLUSIONS
1. Cumulative Technologies. -- Our general conclusion is that multiple and competitive sources of invention are socially preferable to a structure where there is only one or a few sources. Public policy, including patent law, ought to encourage inventive rivalry, and not hinder it. As the "race to invent" models show, a rivalrous structure surely has its inefficiencies. But such a structure does tend to generate rapid technical progress and seems a much better social bet than a regime where only one or a few organizations control the development of any given technology.
While there are exceptions, where a few organizations controlled the development of a technology, technical advance appeared sluggish. The company with the inside track has often failed to move aggressively; the Edison light bulb patent is perhaps the best example. At the same time the history of many industries -- beginning with the steam engine -- show that outsiders with promising approaches have been held back. n324 In what we have called cumulative technologies, particularly when the product in question was a multicomponent system, broad patents on components led to blockages. These were resolved, in some cases, by the development of more or less automatic (if elaborate) cross-licensing schemes. These should not be understood as mechanisms to achieve orderly development of the "prospect" but rather as mechanisms to cancel out the blocking effects of broad patents. There is no evidence, for instance, that firms coordinated the development of a prospect by licensing the cultivation of particular [*909] applications of a broad technology to particular licensees; indeed, patents were often pooled and cross licensed en masse to all firms seeking to enter the field. While sometimes these have come about privately, in other cases patent logjams have been broken only with the powerful force of government intervention. These episodes testify to the blocking power of broad patents as well as social creativity in working around them; they do not argue for the social efficacy of broad patents.
In addition to these general conclusions, we wish to comment on another problem involving patent scope: pioneer patents and the doctrines of enablement and equivalents.
We have seen a number of examples of patents granted on major new discoveries, or pioneer patents. The Wright brothers, for example, received an important and well-deserved pioneer patent on airplane stabilization. This created problems because other inventors such as Glen Curtiss, following close on the heels of the Wrights, were blocked for a time from introducing their advances into the fledgling industry. Our proposal would not mitigate these problems. However, in many cases we have described there was no justification for the broad scope granted and upheld. There was certainly no justification for the broad Selden patent which caused such difficulties. A similar argument could be made with regard to the Genentech "expression" patent. In each of these cases the actual or potential harmful effects of the pioneer patent could have been mitigated had the Patent Office paid closer attention to what the inventor actually disclosed in his specification as an indicator of what the inventor actually achieved, and the broad nature of what was already known and in the public domain, and restricted the allowed scope accordingly. Likewise, the doctrine of equivalents and its reverse can be applied to restrict pioneer patents' scope when necessary. n325 Courts should be encouraged to hew more closely to the substance of the inventor's disclosure when deciding whether an accused device infringes her patent. Because these issues are so important, we make a number of rather detailed recommendations.
In an earlier section of this essay, we suggested an important addition to conventional equivalency analysis. Once a court completes its assessment of the significance of the patented device, it should consider in addition the importance of the advance represented in the accused device. This was in essence the approach taken by the court in the Texas Instruments case, n326 in which the Federal Circuit denied infringement because the accused calculators showed significant improvements in [*910] many respects over the patentee's design. In light of our comments on the interconnection between enablement and equivalents, we approve of the strict attention the court paid to the specification of the pioneer patent in that case. n327 The equivalents inquiry, even for a pioneer patent, should be centered around whether the improved structures of the accused device show major differences from the structures disclosed in the original specification. n328 specifically, the court should look for differences in the following areas:
Materials;
Changes in the number of components;
Greatly improved efficiency in individual components;
Increased efficiency in the way components work together, i.e., overall design improvements.
This was the approach followed in the Texas Instruments case. In Texas Instruments, the accused devices incorporated improvements in all these areas. What makes Texas Instruments worthy of emulation is its focus on the degree to which the accused device represents an advance over the patented device. In a way, this simply brings some symmetry to the equivalency issue. Just as a court looks to the degree of advance over the prior art in deciding whether the patented device is a pioneer or only a minor improvement, so too should the courts be encouraged to examine the accused device. If it represents a significant advance, this is a factor that should weigh in its favor when the issue of infringement by equivalency is decided. Analysis along these lines might have reduced the blocking effect of pioneer patents in certain fields we have [*911] studied. n329
The same point should be borne in mind when a claim covers embodiments that turn out to be well beyond the teaching of the patent's disclosure. This is the case of so-called reverse equivalents. If an improvement represents a very significant advance, it should be held not to infringe -- even if it is within the literal bounds of the patentee's claim. This is the lesson of the Westinghouse case. n330 As long as compulsory licensing remains anathema, the possibility of a "hold up" by the basic patent holder makes reverse equivalents the best alternative under these circumstances. n331
A more liberal use of reverse equivalents would be especially valuable when the allegedly infringing improvement embodies new technology not available when the patent was issued. As long as adequacy of disclosure is measured as of the filing date, enablement doctrine will be of no help to the infringing improver. n332 The Hughes Satellite case n333 and the polypropylene patent n334 are good examples of the need for a reverse doctrine of equivalents in this situation.
The essential point to grasp is that here, as with regular equivalents, courts have their only opportunity to review patent scope in light of later technological developments. They should make good use of the opportunity, with an eye toward preventing the kinds of blockage we have described.
2. Chemical Industries. -- As we saw in our review of the chemical industries, invention in this field has some of the features of discrete and some of cumulative technologies. For the reasons just described, the latter similarity leads us to counsel cautiion in the awarding of broad patents in this field. But there are two factors that mitigate our concerns somewhat. First is the relative rarity of very broad patents in the chemical field, primarily because of the unpredictability of chemical research. n335
Second is the very well established practice of licensing in these industries. Some of the examples we have explored bear this out. Because Ziegler was an academic scientist, he had to license his catalyst [*912] patents to make money on them. n336 And even a huge chemical company like Imperial Chemical found it necessary to license several competing producers of polyethylene before a 1952 antitrust consent decree made licensing mandatory. n337 The reason is probably the same as for bulk chemicals: no one pro