A COST-BENEFIT APPROACH TO PATENT OBVIOUSNESS

INTRODUCTION

The standard for patentability, codified as the "obviousness" standard,1 holds that an invention is not patentable if it "would have been obvious at the time the invention was made to a person having ordinary skill in the art. . . ."2 People, however, have different opinions on what is "obvious," on what is "innovative," and subjectivity has proven impossible to remove from the process. In the 150-year history of the standard, courts have articulated a variety of tests to determine obviousness, all to no avail.3 The Supreme Court's most recent effort, after deriding the prevailing test in the lower courts for the last twenty years as "gobbledygook,"4 managed to replace it only with the admonition that the test for obviousness is "flexible."5

The confusion springs partly from the fact that the obviousness inquiry as framed misses the point. What matters to society is not whether an invention is obvious or innovative per se. Most people would agree that a Rube Goldberg machine is difficult to create, innovative, and not obvious-but it creates no benefit for society. What matters to society is whether allowing a patent for a particular invention brings more innovation benefits to society than the patent costs.6 And unlike debates about whether an invention is obvious or innovative to a hypothetical person, economic benefits and costs can be objectively measured. As such, patentability can and should be determined by balancing the costs and benefits of making a patent available for a particular invention. This Article outlines the assessment of these costs and benefits.

In assessing the benefits of a patent, courts and commentators often casually speak of patents as providing incentives to "create" inventions.7 It is important to emphasize that the social benefit of patents is not the creation of inventions as such. Virtually every invention will be created sooner or later. The true benefit of a patent system is that it speeds up the inventive process, that is, patent incentives accelerate inventions. If the cure for AIDS would have taken 100 years to develop without the prospect of a patent, we hope that by offering a patent and the prospect of monopoly profits, the cure will be developed in, say, 10 years. More specifically still, the benefit is not only that we receive the invention earlier, it is also that we can use the invention during this entire period even though the invention is under monopoly.8 Paying monopoly prices for 100 years is better than having no cure for 100 years.

It follows that the widespread perception that patents are socially costly simply due to their conferring a monopoly is also misguided. Society benefits from inventions even under monopoly conditions-if society would not otherwise have had the invention at all. The cost of a patent accrues only when society could have otherwise had the same invention in perfect competition without a patent. Stated another way, the patent system incurs costs only if the invention would have been independently invented during the life of the patent. Only after independent invention occurs-or rather, would have occurreddoes the monopoly cost of a patent accrue. This monopoly cost must then be balanced against the patent's benefits.

An invention should receive a patent if the accrued benefits before independent invention outweigh the costs after independent invention. The Supreme Court recently moved towards this concept, noting that awarding patents "to advances that would occur in the ordinary course without real innovation retards progress . . . ."9 As I shall demonstrate, the benefits of awarding patents correspond to the economic concept of consumer surplus, and the costs are the difference between the consumer surplus in monopoly (with the patent) and the consumer surplus under perfect competition (with independent invention). Because both are measurable in a given market for a patented invention, patentability reduces to an objective, empirical question.

This Article redefines the patentability standard using economic principles, offering determinable criteria for courts that have struggled to articulate them. Part I outlines the history of the struggle to define a standard for patentability, later codified as the obviousness requirement. Part II discusses the economic costs and benefits of a patent grant. Part III outlines the test, including the important criterion of independent invention. Part D7 discusses some complications to the analysis that do not materially change the test. Finally, Part V suggests some modifications to current law in light of the considerations described.

I. DEVELOPMENT OF THE LAW OF OBVIOUSNESS

A. From Hotchkiss to Cuno

The early American patent statutes did not have an express obviousness requirement-all that was required was novelty and usefulness.10 Even relatively trivial changes satisfied a bare novelty standard, causing predictable problems:

The most frivolous and useless alterations in articles in common use are denominated improvements. . . . Implements and utensils, as old as the civilization of man, are daily, by means of some ingenious artifice, converted into subjects for patents. If they have usually been made straight, some man of genius will have them made crooked. ... If, from time immemorial, their form has been circular, some distinguished [artisan] will make them triangular. . . .11

The requirement of some substantial innovation was a judicial gloss on the bare novelty standard, developing over time. Although some early cases had hinted at a requirement stricter than bare novelty,12 it was Hotchkiss v. Greenwood that finally articulated the requirement clearly.13 In Hotchkiss, the patentee was the inventor of a method of making doorknobs made of clay or porcelain. The problem was that this method had been used on metal knobs, and the patentee was left to argue that "the novelty consisted in the substitution of the clay knob in the place of one made of metal or wood."14 The Court was not impressed:

[U]nless more ingenuity and skill in applying the old method ... to the clay or porcelain knob than were possessed by an ordinary mechanic acquainted with the business, there was an absence of that degree of skill and ingenuity which constitute essential elements of every invention. In other words, the improvement is the work of the skilful mechanic, not that of the inventor.15

By requiring a "degree of skill and ingenuity,"16 the Court established what would be known, for the next hundred years, as the "invention" requirement.

The revolutionary implications of Hotchkiss were not appreciated for some time.17 There were now three requirements for patentability: novelty, usefulness, and inventiveness.18 The inventiveness requirement, however, proved difficult to define in any positive sense. All the Court could issue was a series of pronouncements of what would not be inventive and thus was not patentable.

Following Hotchkiss, the Court had little difficulty in establishing the principle that "use of one material instead of another in constructing a known machine is, in most cases, so obviously a matter of mere mechanical judgment, and not of invention, that it cannot be called an invention."19 Similarly, the Court held that "change[s] of form, proportions, or degree, or the substitution of equivalents, doing the same thing as the original invention by substantially the same means, is not such an invention as will sustain a patent, even though the changes of the kind may produce better results."20 The Court also declared that "application by the patentee of an old process to a new subject, without any exercise of the inventive faculty, and without the development of any idea which can be deemed new or original in the sense of the patent law" would not be patentable.21 Finally, the Court established the principle that the "mere aggregation of separate devices, each of which performs the function for which, when used separately, it was adapted" was not patentable,22 unless the combination produced "a new and beneficial result, never attained before."23 The "new result," however, could not be the "[perfection of workmanship, however much it may increase the convenience, extend the use, or diminish expense."24

Explaining how an inventive "new result" could do more than increase convenience, extend useful life, and diminish expense proved difficult. Eventually, in the 1941 decision Cuno Engineering Corp. v. Automatic Devices Corp.,25 the Court decided that the difference between workmanship and inventiveness was a "flash of creative genius."26 This unfortunate phrase caused endless confusion in the lower courts, and many suggested that corporate sponsored laboratory research was essentially unpatentable because such research necessarily involved multiple scientists and step-by-step experimentation with no single "flash" of genius.27 Finally, in Great Atlantic & Pacific Tea Co. v. Supermarket Equipment Corp.,28 the Supreme Court acknowledged that it had not, could not, and would not articulate a precise test for its inventiveness requirement:

While this Court has sustained combination patents, it never has ventured to give a precise and comprehensive definition of the test to be applied in such cases. The voluminous literature which the subject has excited discloses no such test. ... In course of time the profession came to employ the term "combination" to imply its presence and the term "aggregation" to signify its absence, thus making antonyms in legal art of words which in ordinary speech are more nearly synonyms. However useful as words of art to denote in short form that an assembly of units has failed or has met the examination for invention, their employment as tests to determine invention results in nothing but confusion.29

B. Section 103 and Graham v. John Deere Co.

At the same time that Great Atlantic was pending in the Supreme Court, efforts for a major reform of patent law were pending in Congress. This effort included a codification of the standard of patentability. The first version of what went on to become 35 U.S.C. § 103 read:

Conditions for patentability, lack of invention

A patent may not be obtained though the invention is not identically disclosed or described in the material specified in section 22 of this title [which went on to become 35 U.S.C. § 102], if the differences between the subject matter sought to be patented and said material are such that the subject matter as a whole would be obvious to an ordinary person skilled in the art.

Patentability as to this condition shall be determined by the nature of the contribution to the advancement of the art, and not by the nature of the mental processes by which such contribution may have been accomplished.30

This section went through some minor amendments, and was eventually passed as section 103 of the Patent Act of 1952.31 One early amendment was to replace the phrase "lack of invention" with a new "nonobviousness" standard.32 This was no doubt motivated, in part, by a hope on the part of some participants in the legislative process that the codified standard would avoid the confusing morass of "inventiveness" precedent that had accumulated since Hotchkiss.33 At the same time, Congress did not specifically repudiate any prior judicial decisions except Cuno.34 Section 103, in its final form, read:

Conditions for patentability; non-obvious subject matter

A patent may not be obtained though the invention is not identically disclosed or described as set forth in section 102 of this title, if the differences between the subject matter sought to be patented and the prior art are such that the subject matter as a whole would have been obvious at the time the invention was made to a person having ordinary skill in the art to which said subject matter pertains. Patentability shall not be negatived by the manner in which the invention was made.35

The Supreme Court first interpreted the new, codified, obviousness standard in Graham v. John Deere Co.36 The Graham Court first reaffirmed that the newly enacted section 103 was a mere codification of the prior law dating from hotchkiss,31 save for the legislative overruling of the "flash of genius" test.38 Surveying its prior cases, the Court recounted: "[A]s this Court has observed, '(t)he truth is, the word ("invention") cannot be defined in such manner as to afford any substantial aid in determining whether a particular device involves an exercise of the inventive faculty or not.' "39 The Court then proceeded to lay down a new approach to determining patentability:

The Hotchkiss formulation, however, lies not in any label, but in its functional approach to questions of patentability. In practice, Hotchkiss has required a comparison between the subject matter of the patent, or patent application, and the background skill of the calling. It has been from this comparison that patentability was in each case determined.40

According to the Court, obviousness could be determined using a simple four step analysis. First, "the scope and content of the prior art are to be determined."41 Second, the "differences between the prior art and the claims at issue are to be ascertained."42 Third, "the level of ordinary skill in the pertinent art [is to be] resolved."43 Finally, "[a]gainst this background, the obviousness or nonobviousness of the subject matter is determined."44

The Graham formulation gave some structure to the obviousness inquiry. The Court's opinion, however, provided little insight into how, against the background facts, "the obviousness or nonobviousness of the subject matter" was to be determined. The opinion gave all the appearance of expecting a solution to appear out of thin air once the formula was followed.45 The lack of an articulable rule meant that determinations of obviousness took the appearance-and arguably the reality-of resting on judicial whim, where the validity of a patent was heavily dependent on the court that was deciding the case,46 a result that the drafters of section 103 had clearly sought to avoid.47

C. The Teaching-Suggestion-Motivation Test

Partly in response to the widespread perception that the validity of a patent was mostly dependent on the forum of its adjudication,48 Congress created the Court of Appeals for the Federal Circuit in 1982 and vested in that court exclusive jurisdiction over patent-related appeals from both the patent office and district courts.49 With the creation of the Federal Circuit, the Supreme Court largely withdrew from the field of patent law.50 The Court did not decide a single case substantially affecting the obviousness doctrine between 1977 and 2006.51

In the absence of Supreme Court intervention, the Federal Circuit developed its own structure for obviousness analysis. The first step requires the party challenging patentability to prove a prima facie case of invalidity.52 If a prima facie case is made, the patentee can then rebut this case by providing facts supporting non-obviousness, such as demonstrating unexpected results,53 that the prior art teaches away from the claimed invention,54 that the invention proved to be commercially successful,55 or that others had attempted and failed to make the invention.56 When rebuttal evidence is provided, the prima facie case dissolves and obviousness is determined in the totality of circumstances as under Graham.57

The key to this three-step process, however, was always proving the initial prima facie case because the Federal Circuit created a difficult evidentiary hurdle. In order to establish a prima facie case of obviousness, the party challenging patentability must prove a "teaching," "suggestion," or "motivation" to modify prior art references into the claimed invention.58

The teaching-suggestion-motivation test achieved limited objectivity, but at a heavy cost. To the extent that the test required explicit suggestions to combine elements to achieve an invention, it was objective.59 For example, if an invention is the combination of A and B in a single package, people can disagree over whether the idea was ingenious or obvious. On the other hand, if the legal question was transformed to whether there was a piece of paper that explicitly said, verbatim, "combine A and B in a single package," the inquiry became purely factual and objective. The problem with requiring such a degree of explicitness and specificity, however, is that the obviousness inquiry reduces to a simple novelty inquiry. It would require the prior art reference to describe the invention exactly.

To the extent that the patented invention differed from the prior art's explicit teachings, however, the teaching-suggestionmotivation test could do very little to help unless it was to require patent grants for every novel invention. Obviousness, to have any meaning, involves bridging the difference between the prior art and the patented invention through something that is not explicitly in the prior art itself. In a series of decisions after the Supreme Court granted certiorari in KSR International Co. v. Teleflex, Inc.,60 the Federal Circuit emphasized that there may be "implicit motivation" from many sources to provide that bridge.61 The problem is that once this "implicit motivation" standard is accepted, the obviousness question lands back in square one; it is merely a matter of semantics to switch from asking whether substituting a clay doorknob for a metal doorknob was "noninventive," to asking whether it was "obvious," to asking whether it was "implicitly motivated." All of these standards proved to be masks for subjective opinions, either of experts,62 judges,63 or juries.64

D. KSR International Co. v. Teleflex, Inc.

The Supreme Court's decision in KSR International Co. v. Teleflex Inc.65 substantially adopted the Federal Circuit's teaching-suggestion-motivation test, but removed the more bright-line rules that the Federal Circuit had previously laid down. The Court held that there should be "an apparent reason to combine the known elements in the fashion claimed by the patent at issue" and that "this analysis should be made explicit."66 The Court also held, however, that "the analysis need not seek out precise teachings directed to the specific subject matter of the challenged claim, for a court can take account of the inferences and creative steps that a person of ordinary skill in the art would employ."67

The KSR Court quite clearly rejected the strong version of the teaching-suggestion-motivation test, one that would have required the prior art to explicitly teach the differences-and one that would have rendered § 103 dead-letter as there would be no differences to bridge.68 Ultimately, KSR's holding is rather akin to the "implicit motivation" standard that the Federal Circuit had already adopted in response to the Court's grant of certiorari.69 The problem with this "flexible" approach to obviousness endorsed by the Supreme Court is the same as it has always been: Whether something is implicitly motivated and obvious is nothing more than a subjective, inscrutable judgment call. Judges, juries, and experts will have endless debates on this matter of opinion; but it cannot be objectively proved.

E. A Summary of Obviousness Law

Judge Giles Rich of the Federal Circuit once described the pre-codification "invention" standard as the "plaything of the judiciary [that] meant anything the judges chose to make it mean."70 The codified standard is unfortunately no different. The court that decided in 1999 that a pumpkin face on a trash bag was nonobvious71 was a very different court from the one that decided in 2006 that a new method of dyeing textiles that eliminated two steps in the traditional process was obvious;72 yet both explicitly purported to be applying the same test.73 Both the teaching-suggestion-motivation test and § 103 have proven to be extremely malleable.

As a brief survey of the history of the obviousness standard shows, the problem is a lack of objective criteria to determine obviousness. Although § 103 purports to create an objective test by referencing a hypothetical person, the subjective opinion of a hypothetical person is still subjective: Why, exactly, does the substitution of clay for metal doorknobs not demonstrate ingenuity?74 Who is to say that such a substitution required less brilliance than the substitution of carbon for metal filament in the light bulb?75 And if some expert was to so opine, how could he be proven wrong? Even when everyone agrees on the precise difference between clay and metal down to the last atom, people disagree on the significance of such a difference. With subjective judgment comes the concern that, in hindsight, everything looks obvious and insignificant.76 Because a subjective judgment is ultimately inscrutable, the hindsight bias cannot be isolated and removed.77

The problem of subjectivity in the obviousness standard is similar to that in the reasonableness standard of tort law.78 Like the obviousness standard, reasonableness references the hypothetical judgment of a hypothetical person to create semantic objectivity.79 In tort law, courts struggled to give the standard of a "reasonable and prudent person" useful meaning, beyond telling jurors that the reasonable person was just like them and allowing jurors to exercise subjective judgment.80 What appears reasonable and prudent to one person appears reckless to another. And the phenomenon that in the hindsight of an accident everything looks unreasonable remains a widely recognized problem.81

In the tort context, economic thinking led to Learned Hand's formulation of reasonableness in United States v. Carroll Towing Co.,82 which stated that "if the probability [of accident] be called P; the injury, L; and the burden [of avoiding the accident], B; liability depends upon whether B is less than L multiplied by P: i.e., whether B [<] PL."83 The switch from inherently subjective notions of "reasonableness" to a mathematical formulation with determinable (albeit difficult to determine84) variables marked an important advance that forms the foundation of much of modern tort scholarship.85 Similar to the Hand formulation, patentability can be assessed using a mathematical formulation, replacing ill-defined and subjective notions of "obviousness," "ingenuity," or "implicit motivation" with objective measures of benefits and costs, as demonstrated below.

II. THE ECONOMICS OF PATENTS

A. The Benefits of Patents

1. The Acceleration Benefit

The traditional theory of intellectual property was that the rights provided a reward for making investments that "created" inventions. At the time of the Patent Act of 1952, the economic debate over patents reflected this "creation" theory. In a report to Congress surveying the economic literature on patents, Fritz Machlup wrote:

Are the consumers-the non-patent-owing people-worse off for [the patent system]?

"No; they are not," says one group of economists. Patents are granted on inventions which would not have been made in the absence of a patent system; the inventions make it possible to produce more or better products than could have been produced without them ....

"Wrong," says another group of economists. Many of the inventions for which patents are granted would also be made and put to use without any patent system. The consumers could have the fruits of this technical progress without paying any toll charges. Even if some inventions are made and used thanks only to the incentives afforded by the patent system, consumers must pay for all patented inventions ....

This is but one of the fundamental conflicts in the economics of the patent system.86

As the italicized portions demonstrate, the "creation" dispute rested on a dichotomy. Patent proponents asserted that patents created inventions whole, opponents asserted that patents created nothing at all. What both sides ignored was that while virtually every invention would be created sooner or later even without a patent system, patents created incentives for additional research investment, leading to inventions being made sooner than they otherwise would be.87 The patent system creates no inventions, it accelerates them.

One of the first authors that contributed to this modern understanding was Yoram Barzel, in his paper Optimal Timing of Innovations.88 As the title itself suggests, innovation is a matter of timing, not creation. Barzel suggested that the patent system created incentives for multiple firms to race for a patent, affecting the timing of when an invention would be developed.89 Implicit in this idea was that patents affected when, not if, an invention would be created. In response to Barzel, Edmund Kitch crystallized this concept:

If the purpose is to reward the inventor for his invention, then why shouldn't he be awarded all of the present value of his invention? The simplicity of this argument breaks down, however, if we take the view that the inventor's contribution is not the invention itself-which eventually would have been made by someone else-but the time of the invention. The patent should reward not for the whole value of the invention, but for the value of being first.90

Kitch, however, did not fully realize the implications of his theory, believing that the social value of an invention affected the term-and scope-of the optimal patent but not whether it should be granted.91 Further, Kitch suggested that the reward of the patent should be for "the whole value" of being first, or, in economic terms, the entire surplus during acceleration.92 What Kitch ignored in this calculation was the very concern that the rest of his article was dealing with: the problem of dissipating producer surplus.93

2. The Problem of Dissipating Surplus

As Barzel demonstrated, the acceleration of an invention in the patent system is accomplished through the mechanism of the patent race.94 A patent race occurs when the prospect of monopoly rent encourages multiple firms to make the investment in research; the first to create the invention obtains the patentand the losers get nothing. To Barzel and Kitch, the patent race was a problem because innovation races between multiple firms led to wasteful duplication of research.95 Wasteful duplication would result in the producer surplus from an eventual invention becoming completely dissipated beforehand.96

Barzel suggested that a government auction system giving one firm complete control of an invention "before resources are committed to the innovating activity" would prevent wasteful duplication from occurring.97 Following Barzel's analysis, Kitch suggested that the patent system could itself grant broad rights and foreclose later research by rivals, thus itself functioning as the auction that Barzel had envisioned.98 Both authors focused on centralizing rights to the invention as a solution to the dissipation of rents.99

The difficulty with the centralization approach is that a broader patent necessarily means even more rents for the patentee and this pushes rival competitors to accelerate an invention even harder to secure the patent, resulting in even more wasteful duplication and rent dissipation.100 The patent system does not, and probably cannot, grant exclusive control over an invention sufficiently early to prevent a race between multiple competitors from dissipating rents.101

As John Duffy later pointed out, however, patent racing does not dissipate all the benefits of an invention, but only the producer surplus.102 In the aggregate, "firms will neither gain nor lose from investing in innovation,"103 and thus the multiple racing firms will spend only their portion of the expected surplus.104 Society's benefit from the invention is what remains: the consumer surplus.105 It is not the social surplus, because the producer surplus is dissipated. By shifting producer surplus to consumer surplus, more wealth is preserved from dissipation.106

3. Calculating the Benefit

The benefit of a patent, economically, is the social wealth that is preserved from dissipation, i.e., the consumer surplus as demonstrated above. Assume, then, that a patentee conceives an invention and files for a patent, and that if the patent is granted today, the invention disclosed would produce a potential total surplus of $10 in this year and that the surplus grows at 5 percent per year.107 Assume that the surplus is evenly divided between consumers and producers in perfect competition but that the consumer surplus shrinks to 30 percent under monopoly conditions, the producer surplus taking 60 percent, leaving a deadweight loss of 10 percent. Assume finally that any patent on the invention would issue immediately and would last for twenty years. Figure 1 would represent consumer surplus, projected indefinitely:

In this hypothetical, the consumer surplus will continue to grow indefinitely. Although the annual consumer surplus receives a significant boost at year 2027 when the patent expires, thus shifting the market to perfect competition, during the period of monopoly there is still consumer surplus. Under the assumptions above, the consumer surplus during the patent term, under monopoly conditions, is $99.20. A monopoly consumer surplus is better than the alternative: no consumer surplus at all.108

The model above is unrealistic because the benefit of a patented invention will not continue increasing forever. This is not important; the annual consumer surplus can vary across time. Indeed, such variations occur not only between years but also within years. Mathematically, the consumer surplus of an invention can be stated as a function of time t, B(t). The variable t can be expressed in any unit of time, whether it be days, weeks, months, or years; the function would be defined as the consumer surplus for any particular value of t. The cumulative consumer surplus over a period is thus the area under the curve, expressed mathematically as the integral, ∫B(t).

B. The Costs of Patents

Patents are most frequently criticized for imposing a monopoly deadweight loss for their entire term; indeed, both courts and commentators frequently assume that this is the cost of patents.109 But unless society has some alternative where it has both the invention and does not pay a deadweight loss, this "cost" is illusory. If the only alternatives are having the invention under monopoly and not having the invention at all, society is unambiguously better off having the invention under monopoly.110

It is only after when the invention would have been dedicated to the public domain by an independent inventor that society receives the potential for a third option: having the invention under perfect competition. The difference in social welfare between perfect competition and monopoly, when competition is a potential option, is the cost of a patent.

What are the differences between competition and monopoly? In standard economics, monopoly is different from perfect competition in two ways. First, monopoly causes a deadweight loss.111 second, monopoly causes a transfer of part of the consumer surplus to the monopolist producer-the monopoly rent.112 Ordinarily, economists do not consider wealth transfers from consumers to producers as a cost because simple wealth transfers do not affect the overall level of social welfare.113 In the context of patent racing, however, the transfer of surplus from consumers to producers will result in dissipation of the surplus.114 Therefore, the cost of a patent is the difference in consumer surplus between monopoly and perfect competition, or, in other words, the sum of the deadweight loss and the monopoly rent. Thus, contrary to conventional thinking, the cost of a patent is not the deadweight loss over the entire patent term. The cost of patent protection is the deadweight loss plus the monopoly rent to producers-I shall call this sum the "monopoly cost"-but only after independent invention would have occurred.

The monopoly cost at a particular time t depends on the elasticity of demand and supply. Variations in monopoly cost over time can be expressed as a function C(t). The cumulative monopoly cost over a period would thus be the integral, ∫C(t).

III. A NEW TEST OF PATENTABILITY

A. Balancing the Benefits and Costs

A patent is a bargain between society and inventors.115 In return for creating and disclosing inventions, society rewards the inventor with patent rents. Society has no reason to offer losing bargains.116 As such, patents should only be available when the patent incentive creates more social benefits than granting the patent exacts in costs.

As shown above, the benefit to society is the consumer surplus, expressed as a mathematical function B(t). This is shown below in Figure 2:

The integral appears as the highlighted area in Figure 3.

The use of "equals or exceeds" is necessary because costs may equal benefits. One such case is where benefits and costs both equal zero; for example, if the patentee can perfectly price discriminate (leading to zero consumer surplus) but independent invention occurs at or after patent expiration (leading to zero cost).118 So long as independent invention would occur only at or after patent expiration, the patent should always be granted.119

B. Defining Independent Invention

It should be evident that the time of independent invention, I, is a critical determinant of patentability.120 The time when society would otherwise have received the same invention through an independent inventor marks the boundary between when a patentee does unambiguous good (giving society something it otherwise would not have) to when the patentee causes social harm by exacting monopoly costs. Independent invention must contribute the same invention as the patentee, but do it for free. This entails three conditions: (1) independence from the patentee; (2) independence from the patent system; and (3) public disclosure and dedication to the public domain. A decision-maker evaluating a patent application under the test outlined in this Article must project the timing of when an independent inventor with these three characteristics will emerge.

1. Independence from the Patentee

The intuition for this requirement is simple: a purported independent inventor who simply copies the patentee's patent verbatim and publishes it has contributed absolutely nothing. A rival who benefits from the patentee's research, even when deriving less than the extreme example given above, cannot claim independent invention because the patentee is still part of the invention process. If society had not offered the patent, the patentee would not have existed121 and the rival's invention would thus have taken more time to independently replicate whatever the patentee contributed. Projecting a time of independent invention necessarily asks, for this requirement, "when would an independent inventor who received absolutely nothing from the patentee have invented the same invention?"

2. Independence from the Patent System

While independence from the patentee is intuitive, an independent inventor must do more: he must not be a participant in the patent race for the invention.

Patent races are an inevitable part of the patent system.122 At the same time, research leading to a patent is a winner-takeall race. The first inventor to an invention receives the patent and the twenty-year monopoly;123 every other budding inventor not only fails to win the patent, but instead suffers the loss of all sunk investments due to the patent's power to exclude them from the market.124 Thus, the loser has a powerful incentive, after the fact of losing, to invalidate the patent that issues on the invention.

At the same time, due to the incremental nature of scientific progress in general, and exacerbated by the competition of the patent race in particular, many inventors come very close to each other in achieving the same breakthroughs. This has been witnessed in many important inventions in history: Edison and Swann with the light bulb, Bell and Gray with the telephone.125 Such simultaneous inventions occur with neither researcher stealing the research of the other. But if these simultaneous inventions were to invalidate the resulting patent, no patent would issue on many inventions. Because of the predictability of patent racing, before investing in research, Firm A can predict that its rival Firm B will create nearly-simultaneous breakthroughs with regularity and that the simultaneous breakthrough will destroy the prospect of a patent for both. Firm A will thus not invest in the research; nor will Firm B under the same rationale. Thus, the incentives of the patent system would be seriously harmed, as only a firm with an extremely dominant advantage in research, without fear of rivals entering the race, would pursue the patent.126

Thus in framing the inquiry under Equation 3, the inquiry is not how long it would have taken anyone to develop the same invention. Rather, the inquiry asks, "when would an independent inventor, one not motivated by the prospect of patenting the invention, have developed the same invention?" Of course, it is difficult to discern the subjective motivations of purported independent inventors. But an inventor who is interested in filing for patent protection will presumably not voluntarily subject himself to a statutory bar if he believes he can still patent the invention.127 Thus, an invention disclosed into the public domain and voluntarily subjected to a statutory bar, without the purported independent inventor knowing of any limitation on patentability, i.e., without knowing that another party has beat him to the patent, is likely to be genuinely independent of the patent system.

This test may be under-inclusive in practice. Once we have a patent system, there may be independent inventors who are not motivated by the patent system-in the sense that they would do the exact same thing if there were no patent system-but who nonetheless choose to apply for a patent once they develop an invention.128 On matters of practical proof,129 these independent inventors would be excluded from eligibility even though, considering independent invention in theory, they should be included. But it is difficult to devise a practical test for determining non-reliance on the patent incentive without requiring abstention from the patent system itself.

3. Dedication of the Invention to the Public Domain

Lastly, it is important to note that an independent inventor must contribute the same thing as the patentee, which is to disclose the invention into the public domain. It is not enough that the same invention would be independently conceivedsociety does not receive the invention in perfect competition unless it is disclosed. Thus, an inquiry into independent invention must further ask: "When would the independently conceived invention have been fully disclosed into the public domain?"

The public disclosure requirement serves an additional function of preventing strategic use of the trade secret system. The trade secret system is another form of incentive for innovation.130 If General Motors invented a new method of manufacturing cars cheaply, it can enjoy rents either by patenting the method or by using the method in secret and gaining a competitive advantage-both serve to encourage research into new assembly methods. Disclosing the invention in a patent, however, means the invention will fall to the public domain in 20 years. A trade secret, by contrast, can last indefinitely. For this reason, an inventor who files for a patent will have priority over one who keeps the invention a trade secret, even if the trade secret holder was the first to conceive.131 This creates the effect that a prior inventor who keeps an invention a trade secret will not only lose the trade secret, but will also be excluded from the market if a patent issues to a competitor.132

The trade secret holder thus has the same incentive as a patent race rival-to disclose his trade secret and pass it off as an "independent" invention to invalidate the competitor's patent and avoid being excluded from the market. The problem with this type of strategic disclosure is the same as in the patentracing context: If the patent had not been available, there would have been no incentive to disclose the trade secret. We thus need to separate the purported independent inventor who is really a trade secret holder in disguise from the true independent inventor who would have dedicated the invention to the public domain regardless of whether a patent was available.

In reality, this inquiry is difficult because an invention starts as a trade secret by default, it remains a trade secret absent disclosure,133 and no one but a mind-reader can determine with certainty why a purported independent inventor is disclosing his invention.134 But just as a patent seeking inventor is unlikely to voluntarily subject himself to a statutory bar if unaware of a rival winning the patent race, so too an opportunistic trade secret holder is unlikely to voluntarily disclose his trade secret if he is not aware of an imminent patent on the same invention. Independence from the trade secret system can thus be proven by requiring disclosure of the invention while the independent inventor is ignorant of the patentee.

In summary, independent invention is the time when another inventor will disclose the same invention to the public domain, while ignorant of the patentee and the patentee's disclosure, and without intending to file for a patent himself on the invention.

IV. OBJECTIONS AND COMPLICATIONS

A. The Affect on Incentives

An important consideration for any change to the patent system is how patent incentives would be affected. By this I do not mean whether it will generally enhance or reduce the incentive for research per se. Patents generally increase the incentive for research.135 The question is whether the incentive is optimal in light of the costs of patents.

A patent is in the nature of a unilateral offer for contract between society and inventors.136 Society promises that if an inventor accelerates and discloses an invention that creates more social benefits than costs, society will reward the inventor with a patent of fixed term and scope. In many ways, this is no different than the government promising that a contractor who builds a fighter plane capable of flying at 2,000 mph will receive $100 million dollars. Society must offer sufficient reward-in longer and broader patents that create more monopoly rent, or simply in cash-such that a firm can justify the investment in research-in fighter planes or anything else.137 At the same time, society has no reason to subject itself to a losing bargain. A condition of the patent contract should be that social benefits must exceed social costs. I shall call this the "patentability condition," and it corresponds to the formula in Equation 3.

The patent contract offered to inventors has additional terms, most importantly the patent length and breadth. Optimizing the patent term and scope is complex.138 Longer patents may create such enormous incentives that research funds are diverted from other places where they could have been more socially efficiently invested (a cost); at the same time, those same incentives may cause inventions to be created inventions earlier such that the patent expires earlier (a benefit).139 The test for patentability, however, is not concerned with the optimization of the patent contract. The patent term and breadth are fixed under other provisions of the statute.140 Welfare effects flowing from these additional conditions, such as the benefit of early patent expiration,141 are therefore not considered under Equation 3.

Determining the validity of patents-either in the patent office or in litigation-is concerned only with whether the patent contract (including the patentability condition) has been satisfied. Validity analysis should not alter the overall patent bargain itself. Determining whether an invention should be patentable is thus no different from determining whether any seller has satisfied the terms of his contract, for example, whether a proffered plane really travels at 2,000 mph. Society has the option to reject unpatentable inventions just as a buyer who receives goods delivered by a seller has the option to reject nonconforming goods.142 Assuming that the statutory patent term and breadth achieves a given cost-benefit balance, the question is whether subsequent application of the patentability test distorts the balance already achieved.

In a world of certainty, market-participant research firms will themselves determine the correct allocation of resources for research in light of the bargain offered by the patent system.143 A research firm will not undertake research unless it can achieve the invention at a sufficiently early time for P and E (expiration being dependent on filing) such that the resulting invention satisfies the patentability condition.144 Enforcing the patentability condition of Equation 3 then offers optimal incentives, as does enforcement of any other freely negotiated contract. Including the patentability condition as part of the contract was also efficient, as above, because society has no reason to grant patents that diminish welfare, while it should certainly encourage patents for inventions that enhance welfare.

The complication is uncertainty. Firms must predict the eventual values of the variables in Equation 3: the cost of research and the corresponding monopoly rent they will get. These predictions must be made at the time of deciding to make investments in research-but the payday comes when the invention is complete. Inevitably some predictions will be wrong. How is this risk to be allocated?

As a general rule, the risk of mistake should be allocated to the party best able to avoid the mistake, or, if the mistake cannot be reasonably avoided, to the superior risk-bearer.145 Here, the mistake can be of two kinds. First, the research firm may misestimate any of the variables, resulting in the research being more costly, the invention created being unpatentable under Equation 3, or the monopoly rent being less than projected.146 All of these mistakes are more easily avoided by the research firm. On the other hand, the mistake can also be made by the Patent Office or the court deciding whether the invention is patentable; that is, one could wrongfully reject the patent. The research firm can do relatively little about Patent Office mistakes. If it were possible, this risk should be allocated on the Patent Office.

It is not possible, however, to make the Patent Office or its examiners pay directly for their mistakes.147 The risk can only be directly allocated to the research firm or to consumers.148 While the research firm can do little to correct Patent Office errors, it can probably do more than consumers can-by exerting political pressure on the Patent Office ex post as described in Part IV.B.1. As for the superior risk-bearer, risk for research firms is ultimately borne by the firm's shareholders, who can diversify the risk in capital markets. The risk for consumers depends on the nature of the patent at issue; while consumers are usually diversified, a single patent may affect only a segment of consumers. At a minimum, there is no reason to prefer reallocating the risk from diversified shareholders to consumers. On balance, the entire risk is more appropriately allocated to the research firm.

B. Difficulty of Determining Variables

1. The Advantage of Ex Post Verification

As with any theoretical standard, applying the theory to reality requires empirical data that is not usually available. The key variables of benefit, cost, and timing of independent invention are difficult to predict before the patent is issued. It is important, however, to emphasize that a key advantage of the objective test described in this Article over the traditional "ingenuity" test is that the variables can be determined at all. The traditional test of whether a hypothetical person would have found a given invention "obvious" is an inquiry inherently incapable of verification. One may disagree with the opinion of a judge, juror, examiner, or expert on the issue, but that opinion cannot be objectively disproved.

In contrast, the costs, benefits, and timing of independent invention are facts capable of real-world verification. If the Patent Office predicts that independent invention will occur in two years for an invention and rejects the patent on that basis, in two years we can see whether the Patent Office predicted accurately. In the long-run, the Patent Office will be able to adjust its projections in light of past experiences. While we may not have perfect information for every patent application even after patent expiration-for example, we cannot be sure of the precise timing of independent invention if a patent is granted since the patentee's disclosure might preempt the emergence of actual independent inventors-we can use statistical studies across large samples to determine the overall performance of the patent system. Are there too many patents that later prove invalid? Or are patent applications routinely rejected using projections that later turn out to be incorrect?

Verification allows political accountability because repeat-players in the patent system have a keen interest in correct outcomes. If the Patent Office and courts systematically under-predict benefits, over-predict costs, and estimate independent invention occurring sooner than what eventually turns out to be the case, research firms such as pharmaceutical companies will surely bring political pressure to correct this trend using empirical evidence of the errors. On the other hand, if the Patent Office and courts systematically over-predict benefits, under-predict costs, and estimate independent invention occurring later than what eventually turns out to be the case, frequent accused infringers such as software firms and generic drug companies will likely bring the same political pressure to bear.

Of course, the patentability test is not just a matter of ex post verification. The validity of a patent must be decided before the evidence is fully available. It is useful to separately discuss the problems at the two stages of an obviousness determination: the initial patent office determination of whether a patent should be granted and subsequent disputes over the patent's validity.

2. Validity Determinations in Initial Examination

A prospective estimation of the likely benefits and costs of a patent will be difficult for the overwhelming majority of patents before they issue.149 In practice, this means that the Patent Office should not even attempt such an estimation for most patents because the average patent is likely to have no value or cost of any significance that justifies such expenditure of administrative resources to examine it.

The consensus among economists and antitrust scholars is that the overwhelming majority of patents grant no economic rents of any significance.150 This is because while patents confer a legal monopoly, they do not necessarily confer an economic monopoly.151 Stated simply, most patents are on inventions that are either entirely useless, or at least readily substitutable. And without economic monopoly a patent has no value or cost. Obviousness analysis that is costly should not be undertaken for the run-of-the-mill-patent with low expected cost and the patent should simply be granted.152 Such a laid-back attitude towards examination, of course, will result in the occasional wrongfully issued patent that, contrary to initial expectations, eventually becomes very costly.153 But unless the costly patent can be cheaply identified ex ante for serious examination, society will be better off overall with the occasional costly patent and low administrative costs for patent examination.154

This does not necessarily mean, however, that we should disband the Patent Office and grant every patent application that comes. There are some patent applications that can be easily identified as likely to create large monopoly costs because they will confer true economic monopolies: A miracle drug that cures cancer is likely to fall into this category. These are the patent applications where an effort should be made to estimate future benefits and costs.155 Luckily, within this narrow category of patent applications, there is reason to think that they generally will be capable of at least a rough estimation of their future benefits and costs. After all, the future potential monopoly cost has been roughly estimated as large already.156 Undoubtedly in some cases there will be an invention where we can only generally predict that demand for an invention will be strong, without any capability of even beginning to guess as to how strong, but such cases will likely be rare. Thus, the initial daunting problem of complete inability to determine the future costs and benefits of a patent poses a problem only rarely for the patents where it matters.

The other key variable-the timing of independent invention-also requires prospective estimation. Because an independent inventor necessarily does not seek a patent for his invention, the Patent Office will rarely have actual evidence of subsequent independent invention.157 In the usual case, the only thing a patent examiner can do is make the hypothetical inquiry described in Part III: When would an independent inventor have disclosed the invention to the public domain?

Relevant factors to consider in this hypothetical inquiry will likely include the following: (1) the resources available to a hypothetical independent inventor-general knowledge, printed publications, and patents, except those derived from the patentee and strategic disclosures by rivals;158 (2) the number of potential independent inventors-people of skill in the relevant field, excluding those who would seek a patent or trade secret for such an invention; (3) the differences between the prior art in (1) and the invention; (4) the difficulty and amount of work that would be required to achieve the invention; and (5) the general pace of scientific progress in the field and any special progress leading to the invention, independent of the patent system. The parallel with the Graham factors is evident.159 The underlying evidence is therefore not materially different from that which is currently used to determine obviousness. Of course, the analysis is very different because the timing of independent invention is an objective fact whose accuracy can be verified later.

For most inventions, even very valuable ones, a rough estimate of costs and benefits is all we need, depending on when independent invention occurs. If independent invention will occur within one year of patentee invention, the benefits in that one year must outweigh nineteen years of monopoly cost-a difficult proposition to meet. Similarly, if independent invention occurs nineteen years after invention, even rough calculation will likely show the benefits over nineteen years greatly outweighing one year of monopoly cost. And if independent invention is projected to occur after the patent expires, a further calculation of cost is not necessary. While there will be close cases where the rule is difficult to apply, an objective test that reaches the correct result in most cases is in all events better than the current "ingenuity" test, which is misaimed from the beginning and likely equally difficult to apply in practice.

3. Validity Determinations in Litigation and Reexamination

The validity of a patent can be contested after issuance by three primary routes: litigation,160 ex parte reexamination,161 and inter partes reexamination.162 A patent can be challenged as many times-by different parties-as necessary before it is declared invalid,163 though a party that previously lost in litigation cannot re-litigate validity under res judicata principles.164 Repeated validity challenges may appear unfair to the patentee; however, it is no different than the unfairness arising from any other form of non-mutual collateral estoppel.165 The patentee also receives offsetting advantages.166

Determining the patent's validity after its issuance differs from the initial analysis in the Patent Office in two chief ways. First, we know that the patent is valuable because someone is willing to devote significant resources to contest its validity; most typically, this "someone" will be an accused infringer.167 Second, in the time between the initial grant and the second look, more information becomes available. The existence of an accused infringer will usually provide a wealth of data about the benefits of the invention. For example, an invention that allowed General Motors, as an infringer, to lower the cost of making a car by $100 will create that amount of consumer surplus, less the cost of a license.168 Moreover, in this period there may be an actual independent invention.

As an initial matter, we may question whether such subsequent information should be considered as part of litigation. Courts routinely struggle with the difficult question of whether subsequent events that show error in an initial projection should be admissible.169 This has particular salience in the obviousness test because validity can be challenged repeatedly. If subsequent information can be considered, a patent granted based on information available in 2000 may be invalidated by information available in 2007; on the other hand, a patent rejected based on information available in 2000 will remain rejected forever-absent still-pending appeals, the patent application cannot be "revived" even if the information in 2007 indicates that it should have been granted.170 This asymmetry leads to a bias toward invalidity if subsequent information is considered.

While the invalidity bias is a real problem, the reason to consider subsequently available information is compelling: obviousness analysis likely cannot be done otherwise. An accurate determination of future costs and benefits based solely on the information available prior to patent issuance will be difficult and uneconomical for the vast majority of patents.171 Few inventions have a readily foreseeable market at the time of patent issuance, one that would permit estimations of consumer surplus and monopoly cost. A rule of evidence that excluded all post-issuance-or, more extremely, all post-conception-information will likely reduce the patent system to little more than a random lottery since the data necessary to make an informed judgment of patentability simply is not available at that early time for most inventions.172

Accepting that subsequent events informing the valuations of benefit, cost, and independent invention must be considered, how should they be analyzed? If the patentee practices the invention, the patentee's own sales and profits will provide important information on supply and demand for the invention, laying the foundation for estimates of B(t) and C(t). The presence of an accused infringer provides more of the same type of information. While we cannot have a perfect estimate of the cost of monopoly unless the patentee actually manages to exclude competitors (or charge them license fees), having established markets allows economic experts to make estimates of the relevant variables; such expert market analysis is routine in antitrust cases.173

The most important subsequent development occurs if actual independent invention takes place. As defined previously, an actual independent invention takes place if another inventor creates the same invention (1) without contribution from the patentee or anything derivative of the patentee, (2) without intending to patent the invention, and (3) dedicates the invention to the public domain while remaining ignorant of the patentee.174 The occurrence of actual independent invention sets the maximum value for I.175 The earlier I is, the less time there is to accrue benefit and the more time there is to accrue cost. Thus if the invention is unpatentable using the time of actual independent invention as I, it cannot be patentable if I was an earlier value. An accused infringer, therefore, should strive to demonstrate independent invention-including by pointing to his own activities if they were done without contribution from the patentee. The challenge is to define the criteria for admissible evidence.

Of the three conditions of independent invention, dedication to the public domain is easiest to ascertain. An invention can be dedicated to the public domain in manifold ways, such as by publishing the idea or publicly demonstrating the idea to others. Because there is a required element of publicity, the evidence is usually easily verified. The requirements of independence from the patentee and the patentee's disclosure, and demonstrating ignorance up to the time of dedication, are more difficult to prove because the purported independent inventor must demonstrate a negative. This difficulty can be overcome by rigorously documenting the sources of information received by researchers and publication committees. In this respect, it is notable that firms already routinely instruct their scientists not to search patents for fear of willful infringement liability.176

The biggest problem is determining whether the purported independent inventor is truly independent from the patent system; that is, whether he had intended to patent the invention and only abandoned that plan upon discovering that another inventor had already won the race. This problem arises because patent applications can be filed up to one year after disclosure.177 Thus, the mere fact of disclosure does not necessarily reflect an intent to forgo patenting; only the expiration of the one-year grace period raises a 102(b) statutory bar.

To conclusively establish non-reliance on the patent system, the independent inventor would have to disclose the invention without knowledge of the patentee's disclosure and fail to file a patent for one year while still remaining ignorant of the patentee's disclosure for the entire period. In cases where this occurs, there is no problem; we have a genuine independent inventor. But a per se requirement that the purported independent inventor remain ignorant for one entire year after public disclosure would permit strategic behavior by the patentee: A patentee could strategically prevent evidence of an otherwise genuine independent invention from emerging by specifically disclosing his patent to the independent inventor during the one year period.

The assessment of non-reliance in circumstances where the subsequent inventor receives knowledge of the patentee within one year after disclosure must depend on a factual determination. If the subsequent inventor appears to have made the disclosure as a prelude to patent filing (for example, the disclosure touts the invention as a tremendous improvement over the prior art), that cannot be considered a genuine independent invention. In contrast, a firm intending to market a product that it independently invented-and wishing to prevent an unknown patentee from strategic ex post disclosure within one year-might publicly renounce the option of filing a future patent application to preserve its independent inventor status.178 Another method would be filing a statutory invention registration, which discloses the invention but waives the right to patent protection.179 Firms uninterested in patent protection might also publicize a corporate policy that the firm will not seek patents; a longstanding policy of such nature may provide sufficient reliability to be given credence.

Against the potential strategies outlined above for independent inventors wishing to preserve their status, a strategy remains available to the patentee to preempt independent invention. The patentee can make it impossible to remain ignorant of his patent by saturating the field with knowledge of his patent. This strategy will effectively prevent the emergence of harmful evidence of actual independent invention because any conception will not be demonstrably independent of the patentee. This patentee strategy is less problematic, however, because society receives many collateral benefits from widespread publicizing of a patent. For example, a patentee who publicizes his patent will find it more difficult to practice the widely-criticized "troll" strategy, where the holder of a patent over a small component waits until an industry sinks large investments in a product incorporating the component and then asserts the patent to extort the value of avoiding forfeiture of that investment.180 Widespread knowledge of the patent impedes this strategy because potential infringers would not, ex ante, sink large investments into a product if it anticipated being excluded from the market later.181

C. The Confiscation Problem

Once the variables are determined, inevitably some patent applications will result in rejection. For example, the Patent Office may determine that independent invention will occur in five years and the benefits in the meantime do not justify a patent due to large costs after independent invention occurs. What happens in the five years until independent invention arrives? Do we publish the application so that we can have the invention today?

Initially, the temptation to confiscate may be strong. Because we are rejecting the patent, the incentive effect of the patent system is already lacking. The next inventor contemplating researching a similar invention-in the sense of similar economic risk, cost, and benefits rather than technical similarity-will have a lower incentive to seek a patent whether or not we publish the rejected patent application. Even if the rejection was incorrect, the risk of wrongful rejections is best placed on the patentee to provide incentives for patentees to apply political pressure on the Patent Office to minimize future errors.182

The problem with this analysis is that it ignores the incentives and benefits of the trade secret system. A rejection of the patent under the test in Part III indicates that the acceleration benefits do not outweigh the costs of a patent, nothing more. The inventor, however, may still create more benefits for society with a trade secret than the cost of a trade secret.183 By disclosing the application while denying the patent, we would be confiscating the trade secret rents and thus the trade secret incentive.

Should the patentee have the option of returning to the trade secret system if his foray into the patent system is unsuccessful? In an environment of uncertainty, permitting an inventor to revert to the trade secret system if the patent application is rejected can be efficient, as the following hypothetical demonstrates. Assume that an invention costs $100 to develop by time P, that independent invention will occur 5 years after P, and that the consumer surplus between P and I is $70 whether the invention is used as a trade secret or patented. Assume also that the monopoly rent between P and I will be $75 whether the invention is patented or kept secret, but that the monopoly rent between I and E is $60 and is only available if the patent is granted (if the patent is rejected, the rent after I is zero). Finally, assume that the patentee believes that there is a 50% chance the examiner will correctly determine the cost after I to be $65 (allowing the patent), and a 50% chance the examiner will determine the cost to be $80 (rejecting the patent). The patentee intends to file an application upon conception, and the Patent Office will make a decision immediately.

If we allow the patentee to revert to the trade secret system in the event of patent rejection, the patentee's expected payoff is $105.184 The patentee will therefore undertake the research and produce the invention, creating an expected social benefit of $37.50.185 On the other hand, if we do not allow the patentee to revert to the trade secret system-forcing the patentee to choose-the expected payoff is $75.186 The patentee will thus not develop the invention at all.187

The numbers can be changed to produce a situation where the confiscation of trade secret rents yields social benefits in the short run. If the value of the patent is high enough, and the risk of patent rejection low enough, and the value of the trade secret in comparison low enough, an inventor will surrender his trade secret rents for a ticket in the patent lottery. This potential situation, however, essentially increases the cost for inventors to enter the patent system, a cost that the same inventors will expect to recoup through higher monopoly profits.188 Under Equation 3, a patent that is socially costly will already be rejected and forfeiting the firm's patent-related research costs maintains the overall incentive structure of the patent system. Forcing firms to surrender even more value, i.e., the trade secret rents not attributable to patent system incentives, reduces incentives and distorts the balance struck by Equation 3.

D. Delay Between Conception and Disclosure

In Part II.A, I defined the incentive benefit as accelerating the "invention." The benefit to society is the accrual of consumer surplus between the time of receiving the patentee's invention and the time when society otherwise would have received the same invention through an independent inventor. In more practical terms, however, society receives very little benefit from simple conception by the patentee. An invention usually requires a path of several steps: (1) the idea is conceived; (2) a working model is built; (3) a patent application is filed; (4) the idea is published in an article; (5) the patent is issued; and (6) the idea is embodied in a commercial product.189 The delay between conception and eventual patent issuance creates the problem of defining a precise point for determining the value of P. When should the patentee receive credit for the accrual of consumer surplus?

Initially, the answer may appear that P should start at the time of conception. After all, the patentee cannot affect the time of independent invention, I, and thus any consumer surplus from the moment of conception is a benefit to society. The problem, however, with beginning the consumer surplus "clock" at conception is that it creates two perverse incentives: the incentive to delay disclosure and the incentive to delay filing a patent application.

The patent system is designed to encourage early disclosure and early filing. The most important reason is that filing sets the time for patent expiration.190 Disclosure provides an incentive to file because filing must occur within one year of disclosure or the patent is forfeited.191 Incentives are needed to encourage early filing and disclosure because patentees have many reasons to delay filing as much as possible. For many inventions, the first few years are the least profitable because preparations must be made before a commercial embodiment can be sold: investors must be solicited, factories must be built or reconfigured, and, for pharmaceuticals, FDA approval must be obtained. More importantly, when an idea is first conceived, the inventor has a de facto monopoly over it as a trade secret. Absent disclosure or filing, this monopoly can last indefinitely through the trade secret system.192 Therefore, before either disclosure or filing occurs, the consumer surplus generated by the patentee's invention is not attributable to the patent system's incentives-it is a benefit of the trade secret system.193 Accordingly, the consumer surplus accruing before either filing or disclosure has occurred should not be considered in the balancing test of Equation 3, which is concerned with the benefits accruing from patent incentives.194

It is also clear that after filing and disclosure have both occurred, any consumer surplus generated by the invention should be counted as part of Equation 3, even if the patent has not yet issued. The acceleration benefit of the patent system accrues even when the patent office is slow. The complication is in the two situations between the extremes: (1) when there is disclosure but no filing; and (2) when there is filing but no public disclosure.195

What should the rule be when the patentee discloses the invention in an article but has not filed? Initially, one may very well conclude that this is no different from the situation of the inventor who neither discloses nor files; that is, the inventor is attempting to stretch out his monopoly by delaying the eventual expiration of the patent. Indeed, allowing the consumer surplus to begin accruing at the time of disclosure notwithstanding a later filing would create a strong incentive for an inventor to publish quickly after conception,196 but also creates an equally strong incentive for the inventor to then wait exactly one year after publication before filing the patent application. The inventor would thereby receive an effective 21-year monopoly because competitors will not invest in entering a market if they expected a pending patent to soon force them to forfeit that investment.197

The difference is that the effective 21-year monopoly is the product of congressional choice. By enacting 35 U.S.C. § 102(b) and creating a one year grace period, Congress has essentially condoned the utilization of this mechanism in defining the patent term.198 If Congress wishes to set the patent monopoly truly at 20 years-instead of the current effective monopoly of 21 years-it can do so in many ways: it can reduce the statutory term to 19 years and keep the one year grace period in section 102(b); it can keep the statutory term at 20 years and eliminate the grace period in section 102(b); or it can even change the statutory term to 19½ years and reduce the section 102(b) grace period to six months. There is no reason to use the test for obviousness to achieve indirectly what Congress could easily legislate directly and has chosen not to enact. Therefore, the time of disclosure would be the appropriate value for P when it is earlier than the patent application filing date.

What of the reverse situation, where the patentee has filed the patent application but has not publicly disclosed the invention? As stated previously, the incentives of the patent system are optimized when the entire consumer surplus before an independent invention is counted as a social benefit. Perverse incentives are not apparent: The patentee has opted out of the potentially indefinite trade secret monopoly because the invention will fall into the public domain when the patent expires. Although there is a benefit to having the patentee disclose the invention early for the progress of basic science, the specific invention will remain under a de facto monopoly whether or not there is disclosure.

A second look, however, reveals countervailing considerations. There are at least three potential reasons why a patentee may wish to keep the invention secret despite filing. First, the cost of entry and exit in the market may be so low that competitors will choose to enter the market even if they must exit upon patent issuance. Thus, secrecy grants the patentee a temporary monopoly before issuance. Second, the patentee may conclude that it is more profitable to sue his rivals for infringement rather than to exclude them from the market, and secrecy encourages rivals to make sunk investments in an infringing product in the period before patent issuance. Third, the patentee can plausibly fear that his patent may be rejected and wish to retain the option of keeping the invention as a trade secret if no patent issues.199

The first reason supports allowing the patentee to keep the invention secret until issuance. Were the rule otherwise, the patentee would be forced into a choice of either risking invalidity due to nothing more than the patent office's delay, or disclosing the application and compromising his exclusivity-effectively shortening the patent term due to the patent office's delay. Both options reduce the patentee's incentive for research.

The second reason is the modus operandi of so-called "patent trolls."200 Because the recovery from an ex post infringement suit can often exceed the royalty that is available through ex ante licensing,201 a patentee can profit from accidental infringers who unknowingly sink investments in a product that is patented. By keeping the invention secret during the pendency of the patent application,202 a patentee increases the chance that such accidental infringement will transpire.

There is little positive that can be said about allowing the patentee to keep his invention secret solely to lay a trap for the unwary. It is unlikely, however, that an incentive for early disclosure-by denying credit for consumer surplus generated before disclosure-would do much to alleviate the problem. The standard for a public disclosure in patent law is notoriously low, and a patentee can make a public disclosure under this standard with full confidence that an accidental infringer will never find it.203 Even published patents are rarely read,204 and it is by no means clear that articles disclosing inventions give notice that a patent is forthcoming on the same invention. And if published articles did have the effect of alerting readers to pending patents, the most likely result is that potential infringers would cease reading articles for fear of willful infringement liability, making a disclosure requirement ineffective.205

The third reason, that the patentee wishes to keep the backup option of reverting to the trade secret system if the patent is rejected, initially smacks of opportunism. Providing an option of reverting to the trade secret system, however, is likely beneficial.206 Although a rejection of the patent indicates that the invention did not create sufficient benefits to justify the cost of a patent, it may justify trade secret protection with lower costs. By rejecting the patent, the incentives for the patent system remain optimal even without disclosure.

In sum, the policy considerations generally favor defining P as the earlier of filing or disclosure. While the policy of deterring patent trolls gives some justification for a mandatory disclosure requirement, it would be difficult to separate the patent trolls from patentees who simply wished to have the option of reverting to a trade secret in the event of a rejection. Without persuasive evidence that a disclosure requirement would be effective against patent trolls, mandatory disclosure would likely do more harm than good.207

V. SUGGESTIONS FOR REFORM

The language of § 103 directs courts to consider "the differences between the subject matter sought to be patented and the prior art,"208 and codifies the standard as whether the invention is "obvious at the time the invention was made to a person having ordinary skill in the art."209 Given this standard, a wholesale adoption of the test in Part III would be difficult to fit within the statutory formulation. Nevertheless, several doctrinal modifications are possible within the current statutory framework to conform patentability to the overall interest of the patent system in advancing long-term social welfare.

A. Invalidating Patents Based on Subsequent Independent Invention

1. Doctrinal Basis for Considering Independent Invention

Prior to the Federal Circuit's formation, its predecessor, the Court of Claims, had held that "[t]he fact of near-simultaneous invention, though not determinative of statutory obviousness, is strong evidence of what constitutes the level of ordinary skill in the art."210 In one early decision, the Federal Circuit extended this doctrine so that even subsequent independent invention, just after the patentee, could be relevant to assessing obviousness.211 Regional circuit decisions pre-dating the Federal Circuit's exclusive jurisdiction had similarly placed weight on subsequent independent invention.212

The relevance of independent invention was drastically curtailed by the Federal Circuit's 1984 decision in Lindemann Maschinenfabrik GMBH v. American Hoist & Derrick Co.213 In Lindemann, the district court invalidated a patent primarily on the basis that "three individuals independently created the designs which resulted in development" of the claimed invention.214 The Federal Circuit was not impressed:

Because the statute, 35 U.S.C. § 135, (establishing and governing interference practice) recognizes the possibility of near simultaneous invention by two or more equally talented inventors working independently, that occurrence may or may not be an indication of obviousness when considered in light of all the circumstances. In this instance, it clearly is not. Two of the three individuals were . . . the co-inventors listed on the '315 patent [the patent-in-suit]. The third was an Amhoist employee [the accused infringer] who claimed at trial to have proposed the split ram in January of 1979, more than five years after the invention was made by [plaintiff] Lindemann's assignors, nearly three years after the '315 patent issued, and well after Amhoist's employee Bleeland had in England observed and photographed a Lindemann shear embodying the claimed invention. Accepting, as we must, the district court's crediting of the testimony respecting independent suggestion by an Amhoist employee, that suggestion was simply too late to have been relevant to a determination of whether the invention would have been obvious at the time it was made . . . .215

Although the Federal Circuit concluded with the statement that independent invention "may or may not be an indication of obviousness,"216 it soon became clear that subsequent independent invention would rarely, if ever, become relevant to the obviousness determination.217 In the next year, the Federal Circuit again held:

Development by others may also be pertinent to a determination of the obviousness of an invention; but the evidence presented was of activities occurring well after the filing date of the '926 patent application, and was not shown to apply to the time the invention was made, as required by 35 U.S.C. § 103.218

Of course, independent invention that occurred prior to "the time the invention was made" would already invalidate the patent under 35 U.S.C. § 102, and thus independent invention ceased to be an important consideration in obviousness determinations.

Moreover, at the same time, the Federal Circuit began to insist upon a teaching, suggestion, or motivation as prerequisite proof for a prima facie case of obviousness.219 As such, independent invention would only be relevant after the prima facie case had been successfully rebutted, when obviousness would be judged on the totality of the circumstances.220 A case where the prima facie case was established, rebutted, and the record contained relevant evidence of subsequent independent invention was predictably rare.

The exception that illustrated the rule finally arose in the 2000 case Ecolochem, Inc. v. Southern California Edison Co.221 In Ecolochem, the district court found that there was a sufficient motivation to combine prior art references for one claim,222 and the question considered by the Federal Circuit was whether the prima facie case of obviousness had been rebutted by secondary evidence.223 Against the patentee's secondary evidence of non-obviousness, the accused infringer submitted evidence of subsequent independent invention, upon which the district court relied.224 The Federal Circuit recognized that "[essentially, the district court found that this secondary consideration factor favors obviousness"225 and affirmed with respect to the one claim where the prima facie had dissolved.226 At the same time, however, the Federal Circuit reversed the finding of obviousness with respect to other claims where it held there was no motivation to combine, without considering the evidence of independent invention.227

A rigid requirement of a teaching, suggestion, or motivation in every case relegated independent invention to a fringe doctrine. The Supreme Court in KSR, however, held that there need not be a rigid adherence to this requirement in every case, and in particular there is no need for a teaching to appear only in published articles or published patents.228 Rather, "design incentives and other market forces can prompt variations" in the prior art.229 Consideration of subsequent independent invention as a secondary factor would illustrate the real-life effects of such design incentives and other market forces.

2. Independent Invention as a Secondary Consideration

As the prior section demonstrates, independent invention is part of obviousness law, though its role has been largely forgotten due to the Federal Circuit's emphasis on proving a prima facie case of invalidity solely through prior teaching, suggestion, or motivation. If independent invention is accorded significant weight, the policy goals embodied in Equation 3 can be accomplished in many cases without substantially revising existing patent law. After all, if actual independent invention were to occur within one year, the consumer surplus within the first year would have to outweigh the monopoly cost in the 19 subsequent years to justify a patent, and such a lopsided benefit-cost ratio is unlikely for most inventions. Thus, the mere fact of independent invention within a short time of the patentee's conception would suggest a high likelihood of invalidity, without the necessity of delving into costs and benefits in many cases.

Independent invention, as defined in this Article, also assuages the concerns expressed by the Federal Circuit in Lindemann.230 Specifically, the Lindemann court articulated three concerns: (1) the patent statute contemplates that rival inventors could simultaneously invent, and establishes the interference procedure when two patent applications claim the same invention; (2) independent invention could be falsely claimed, even after the accused infringer has observed and photographed the patentee's product; and (3) independent invention may occur many years after the patentee files for the patent.231

There is no inconsistency between the interference procedure and independent invention as defined in this Article. The interference procedure governs when a subsequent inventor files a patent application for the invention.232 An independent inventor, however, must not file a patent application. Indeed, to qualify as an "independent invention" as defined in Part III.B, the independent inventor must have demonstrably relinquished the option of filing a patent-through measures such as voluntarily creating a statutory bar to patenting-while remaining ignorant of the patentee and the patentee's disclosure.233

The possibility of false claims of independent invention is overcome by the rigor of proof demanded. In Lindemann, the Federal Circuit displayed obvious skepticism that the subsequent invention was truly independent when the accused infringer had observed and photographed the patentee's commercial product embodying the claimed invention.234 Indeed, such a claim would likely not meet the clear and convincing evidence standard required to prove invalidity.235 A rule that requires actual ignorance of the patentee and the patentee's disclosure is probably inescapable.236 As noted in Part IV.B.3, the difficulty of demonstrating a negative-actual ignorance-can be overcome by carefully documenting the sources of information received by researchers and publication committees to show that the conception and the decision to dedicate the invention to the public domain were made while those involved had access to neither the patentee's disclosure nor materials derivative of that disclosure.

Finally, the amount of time permissible between patentee disclosure or filing, i.e., P, and independent invention, I, is a matter of empirical research as to the typical distribution of benefits and costs over the life of an invention. It would be surprising, however, if many inventions accumulated sufficient benefit in their first or second years to outweigh the cost of eighteen or nineteen years of monopoly. On the other hand, the question becomes much closer when five years of accumulated consumer surplus is weighed against fifteen years of monopoly.237

The dedication of the same invention to the public domain-preferably with a foreclosure of the possibility of patent filing-by a subsequent independent inventor, with clear and convincing evidence that the subsequent inventor was ignorant of the patentee the entire time, is evidence of actual independent invention that can constitute an objective consideration described by Graham.238 When such independent creation occurs proximate in time to the patentee's own disclosure or filing-for example, within one year-there is a strong likelihood that society is paying more in monopoly costs than it could receive in consumer surplus from the patent. As such, the patent should be invalidated.

B. Reconsidering the Definition of "Prior Art"

Although section 103 prohibits patents on inventions that are "obvious" in light of "the prior art," it does not define what the prior art is.239 The Federal Circuit and its predecessor have defined the world of "prior art" by borrowing from select portions of Section 102,240 resulting in "prior art" being defined by the earlier of the patentee's conception or one year prior to the filing date. Events subsequent to this critical date are essentially irrelevant to the obviousness inquiry.241 At the same time, almost every patent and printed publication that is published before the critical date is considered "prior art," with the exception that printed publications that do not meet a minimal threshold of public accessibility are excluded.242

The classification as prior art of all public knowledge prior to the date of conception raises the possibility of wrongful rejections for the patentee. As demonstrated in Part III.B.2, a losing competitor who subsequently publishes the results of his research is not truly an "independent inventor," because this type of strategic behavior by a rival confers no favors on the public in the long-term. Invalidating the patent because of a subsequent disclosure by a rival who lost the patent race destroys patent incentives because firms can predict ex ante that multiple rivals are likely to finish close to them in the quest for a patent. If such a close finish between patent-seeking rivals means no patent issues to the winner, then no one will enter the race to begin with, and the patent incentive loses its force.

This intuition applies equally to disclosures prior to the patentee's conception. Research, by its nature, tends to be incremental, and incremental steps in isolation appear to be obvious. To take a simple example, if the research objective is a car that has a fuel economy of 100 miles per gallon ("mpg"), research breakthroughs might occur in one mpg increments, with each one mpg increment appearing "obvious."243 Thus, the patentee's 100 mpg car would be "obvious" because a rival created a 99 mpg car before the patentee, and the rival's 99 mpg car would have been "obvious" because another rival had created a 98 mpg car even earlier, and so on. The problem then is that no patent will ever issue on the 100 mpg car even if, cumulatively, the successive breakthroughs allow us to progress from 20 mpg to 100 mpg. Multiple rival firms racing for a 100 mpg car, publishing their interim findings, will preempt each other and make the patent ultimately unavailable.244 This harms the patent system because mutual preemption is a predictable result of patent racing, and patent racing is inevitable for valuable patents. The result is that nobody will make the investment in researching a 100 mpg car-at least not at the optimal level that the patent system is designed to induce. In a world where rivals race for patents, the ultimate patent survives this preemption gauntlet and remains available only when: (1) researchers do not publish their interim findings; or (2) there is a single, sudden large breakthrough that itself crosses a "nonobviousness" threshold. The first runs contrary to the purpose of patent laws in encouraging scientific disclosure and is unrealistic to expect in patent racing contexts where a losing firm has the strategic imperative to publish and preempt the patent.245 The second-requiring a single, sudden large breakthrough for patentability-is to restore the reviled "flash of genius" test in all but name.246

This concern is not merely theoretical. A recent Federal Circuit case, PharmaStem Therapeutics, Inc. v. ViaCell, Inc.,241 illustrates the phenomenon, and a dissenting judge accused the court of restoring the "flash of genius" test.248 In PharmaStem, accumulated research had created a suspicion "that umbilical cord blood is capable of hematopoietic reconstitution,"249 but the patentees were the first to achieve the result.250 The Federal Circuit nonetheless held that the invention was not patentable because "the inventors merely used routine research methods to prove what was already believed to be the case."251 The relegation of "routine research" to non-inventive and non-patentable status, of course, was the precise problem with Cuno252 and the "flash of genius" test.253

The solution in cases of prior disclosure is not different from subsequent disclosure: A strategic disclosure by patent-seeking rivals designed to invalidate a patent should not be considered prior art for obviousness purposes.254 To do otherwise leaves firms at the risk that no patent will issue in the end-due to mutual preemption-forfeiting the firms' research investment and deterring the ex ante investment that patents are designed to induce.255

The analysis differs, however, between pre-conception patents and pre-conception publications. Unlike the post-conception scenario where a disclosure in a patent should never be considered as an independent invention, disclosures in a patent prior to the patentee should be considered prior art. This is because the prospect of an early and broad patent right that preempts later patents does not diminish incentives for research-it increases them.256 The increased patent racing incentive has beneficial effects for society.257

The interpretation of "prior art" under section 103(258) should be modified such that a printed publication is not prior art if it was produced by a patent-racing rival to preempt a later patent by a competitor.259

CONCLUSION

The traditional approach to patentability has been to compare the patented invention to the prior art, and to ask whether the differences were sufficiently material to be patentable. This requirement has been variously described as asking whether the differences (and solutions for bridging the differences) were "inventive," "ingenious," "obvious," "taught," "suggested," or "motivated." Except where the prior art was required to explicitly describe the invention such that there were no differences, all of these tests were nothing more than semantic variations of essentially the same subjective inquiry. Objective science can describe the difference between a metal doorknob and a clay doorknob down to the last atom, but it will never be able to determine if substituting a metal doorknob with a clay doorknob is "obvious."

The key to reframing the obviousness inquiry is to realize that the goal of patents is not to encourage invention in the abstract, but to encourage innovations that create more benefit to society than the cost of a patent. Whether the patent creates more benefits than costs depends on what the alternative is-whether, if the patent were not available from the very beginning, society would nonetheless have received the same invention. More precisely, because virtually every invention will eventually be independently created without a patent, the inquiry is when society would have received the same invention had a patent never been available. Before such independent invention would have offered the same invention to society, the patent reward creates benefits in the form of consumer surplus. After the time of such independent invention, the patent exacts monopoly costs in the form of deadweight loss and dissipated surplus.

Determining when independent invention would have occurred, however, is a tricky inquiry because it must reconstruct a hypothetical world that entirely removes the incentives of the patent reward. If the patent had never been available, rival inventors in the patent race would not exist, and competitors holding trade secrets would not disclose them to preempt future patents. Removing these influences from the determination results in the requirements that an independent invention must be independent of the patentee, independent of the patent system, and result in a dedication of the invention to the public domain.

This Article redefines the patentability standard, offering determinable criteria for courts to determine whether the availability of a patent for an invention creates more consumer benefit than the patent exacts in monopoly costs. Because patents are ultimately offered for the purpose of enhancing overall social welfare, the obviousness standard should be reconsidered to incorporate these economic calculations. Even without a change in the statute, however, many of the economic policy objectives can be incorporated through greater consideration of independent invention as a secondary consideration, reinvigorating a dormant doctrine that courts have overlooked and underutilized.