II. BACKGROUND: THE "IMPOSSIBLE" IS NOW POSSIBLE

III. HOW WE GOT TO WONDERLAND

IV. EVALUATING THE PATENT EXAMINATION SYSTEM

A. WHY IS PATENT QUALITY SO POOR?

B. SKETCHING AN IDEAL PATENT OFFICE

1. THE FUNCTION OF A PATENT OFFICE

2. THE GOALS OF A PATENT OFFICE

3. PROS AND CONS OF A SIMPLE REGISTRATION SYSTEM

C. OPTIMAL PUBLIC EXPENDITURES ON PATENTS

1. THE BENEFITS OF SORTING APPLICATIONS

2. THE PROBLEMS (AND POLITICS) OF SORTING

3. SECOND-BEST SOLUTIONS

A) WHY NOT PENALIZE HOLDERS OF INVALID PATENTS?

B) WHO IS THE CHEAPEST COST-AVOIDER?

D. EXAMINING THE PTO

1. THE PTO'S EXAMINATION BUDGET

2. SETTING THE IDEAL PTO BUDGET

V. SOME SIMPLE SUGGESTIONS FOR IMPROVING THE EXAMINATION PROCESS

A. JOB DESIGN

B. ALTERNATIVE BONUS SYSTEMS

C. REFORMING REEXAMINATIONS: THE COMMON SENSE CASE FOR PATENT OPPOSITIONS

VI. CONCLUSION

"Now I'll give you something to believe[" the White Queen remarked.] "I'm just one hundred and one, five months and a day."

"I can't believe that!" said Alice.

"Can't you?" the Queen said in a pitying tone. "Try again, draw a long breath and shut your eyes."

Alice laughed. "There's no use trying," she said, "one can't believe impossible things."

"I daresay you haven't had much practice," said the Queen. "When I was your age, I always did it for half-an-hour a day. Why sometimes I've believed as many as six impossible things before breakfast."¹

I. Introduction

The White Queen would be right at home in the U.S. patent system today. First software, once thought too purely mathematical, and now business "methods" or concepts, once thought too abstract, have become perfectly acceptable subject matter for patents. For better or for worse, whole new landscapes have been opened to the possibility of patents.

To get right to the heart of the issues surrounding patents for business concepts, log on to <http://www.walkerdigital.com/html/information.html>. This is the website of Walker Digital, Inc., the company that recently "spun off" its Priceline.com subsidiary, a separate company that uses the Internet to match buyers with sellers.² Here is what you will read:

We [Walker Digital, Inc.] conceive, research, and prepare our patented business systems in-house. Our team of specialists prepares cases that solve real-life problems for a wide variety of industries such as retail, telecommunications, credit cards, casinos and more. So far, we've filed over 250 U.S. and international patent applications to create a portfolio that we believe is unlike anything anywhere else in the world.³

Until very recently, Walker Digital would not have existed. The patent system did not embrace the abstract patents on business concepts that are the company's key assets.⁴ There would be no cornerstone patents on internet price-matching, personified by Walker Digital's "Priceline.com" subsidiary.⁵ Without patents, in fact, it is difficult to see how a firm could survive as an independent "idea factory" for Internet commerce.

Walker Digital is therefore a perfect test case. It can tell us whether formerly "impossible" patents on business concepts are a good idea. If there were some way to determine whether this firm had initiated business concepts that no one else would have, or had hurried them into practice faster, we could ask: is the game worth the candle? Alas, no such knowledge has been revealed to us. The instruments we have at hand are simply too imprecise, at least for the time being. We may see an explosion of activity. Or we may hear horror stories about good, solid businesses abandoned in the face of predatory patent extortionists. It is simply too soon to tell.

But there are some positive steps we can take to limit any negative effects from business method patents. The most important is to make sure that the business concept patents that do issue are good, solid patents. It may be too late to argue to a court that business concept patents are universally bad. And it may be too early to ask Congress to rein them in. But it is neither too late nor too early to argue forcefully that bad business concept patents are bad.⁶ In fact, the time is just right: minimizing the number of worthless business concept patents makes a great deal of sense just now. Only by improving the overall quality of these patents can we begin to determine whether or not they make any sense. Once we disentangle the bad from the good, we can see whether the good ones are worth the trouble. If, in the process, this entails improving the overall quality of issued patents, all the better. If it tweaks us into fixing some deep-seated flaws in the way the Patent and Trademark Office ("PTO") examines patents, the advent of business method patents may even turn out to serve a useful purpose.

II. Background: The "Impossible" Is Now Possible

Before addressing the question of bad business concept patents, let us first consider how we came to patent this subject matter in the first place. Although the older cases do not articulate their reasoning very clearly, they seem to center around one idea: that the patent system was meant to protect technology-actual machines, devices, and new chemical compositions-rather than pure concepts.⁷ Because business methods are not tied to particular machinery or devices, they are clearly not patentable under this view.

This antipathy to patenting mere abstractions actually grew out of older cases which questioned the patentability of processes per se.⁸ How, it was asked, could a list of steps not tied to particular machinery or devices be patentable? In time, the opposition to process patents died away, partly because they came to be understood as physical transformations rather than mere abstractions.⁹ It also did not hurt that they were perceived as crucial to the growing chemical industry of the early twentieth century. Yet, the prohibition on patents for business methods lived on.¹⁰

With the acceptance of patents for software, courts could no longer persuasively rely on the distinction between concepts and machines.¹¹ Even so, for a brief time the rule against business method patents survived. Those who defended this rule justified it on other grounds. Most powerfully, it was argued that such patents were simply not necessary.¹² After all, there seemed to be no shortage of new accounting methods, financial instruments, or financial services techniques throughout the history of the American economy, when business methods were not patentable. Even into the mid-1980s, when business method patents were just beginning to appear, the U.S. was considered the world leader in this service industry. Thus, according to this view, the proper question is: why fix it if it ain't broke?

The conventional answer is dictated by the logic of patent principles and current practices. It holds that there is no sound reason not to protect business methods. The history, logic, and accepted practices of our method of granting patents essentially compels us to allow patents on business concepts, because there is no principled basis on which to distinguish this "industry" from the myriad other industries that routinely obtain patents. Further, we should all have faith that this wave of patenting will unleash an Edisonian tidal wave of inventiveness-that, if we thought entrepreneurs rapidly introduced new ideas such as overnight package delivery and 1-800-Flowers without patents, then Watch Out!, because we haven't seen anything yet in this field!

Certainly Walker Digital sees it this way. Again, their web page:

We're not believers in traditional commercial inventing, where old methods are shoehorned into new technologies. Rather than think outside the box, we seek to reinvent the box. We create practical new ways to do things based on the inherent benefits of new technologies. We then take our core ideas, protect them with patents and establish licensing partnerships with major industry players who bring our ideas to market. These ideas are our intellectual property, our product.

Who are we?

Our team includes entrepreneurs, inventors, technologists, patent attorneys, industry analysts and even a world-renowned cryptographer. It also includes folks who, in previous lives, were some of the country's top CEOs and marketing executives. It's a group of highly intelligent, inventive, business-savvy people, with plenty of room for even more bright people, like you.

We earn profits from our intellectual property through a variety of business strategies ranging from direct licensing agreements-selling an idea to another company-to spinning off new businesses in which we retain an equity stake. Our first business spin-off was a home run-Priceline.com, our patented buyer-driven commerce system.

Priceline.com is the only system, on or off the Internet, where buyers can name their own price for specified goods and services. Like all of our spin-offs, its success has become our success.¹³

Not surprisingly, the patent covering the Priceline.com service is in dispute. A rival inventor-a patent lawyer, in fact-had filed a patent application with somewhat analogous claims earlier. An interference is now afoot.¹⁴ And so commences the inevitable shakeout period when rival patentees jockey for position. This much, at least, is not new. For example, the Bell System had 600 patent infringement suits pending in the late nineteenth century,¹⁵ and the past fifteen years have seen a steady stream of foundational litigation in the biotechnology industry.¹⁶

What is new is this: the shakeout has begun without answers to some important threshold questions. Chief among these is whether Walker Digital and other firms like it are doing anything that would not be done in the absence of patents. Put another way, in an ideal world, society would have addressed whether or not the types of business concept patents sought by these firms contributed any value in excess of what they cost society. If the answer was no, we would deny patents to them; if yes, patents would be allowed.

III. How We Got to Wonderland

It would certainly be nice to have a theory that would tell us when one type of invention is unpatentable, while another is patentable. But the problem with such a normative theory of patentable subject matter is no less vexing for its familiarity. Where do you draw the baseline? One conventional candidate, historical practice, is not helpful. There would seem to be little hope in constructing an "originalist" interpretation of the Intellectual Property Clause of the Constitution¹⁷ to limit its subject matter: By definition, the clause envisions the creation of unanticipated inventions and writings. It provides no built-in limits. Hence it does little good to argue that the patent law traditionally protects only conventional "hardware" inventions. It is quite true that the canonical patented technology in the eighteenth century was a simple agricultural tool (an axe or a plow) which then became a more complex implement (a cotton gin or reaper) in the nineteenth century; even later, it became a machine, electrical device, or chemical process. These are true, but useless, historical facts; they say nothing about the appropriateness of patenting modern business concepts.

Indeed, following in the general spirit of the Intellectual Property Clause, Congress early on seems to have embraced a kind of blind technological optimism, ignoring economic costs. The value of a limited and well-administered patent system was little debated in the early Republic. There were no detailed cost-benefit analyses when the first few patent acts were passed. And there was consequently no real effort to separate patentable subject matter from things that were not to receive patents. Perhaps this was in part a function of an understanding-shared widely among legislators, courts, patent office officials, and inventors-about what patents were meant to protect. Everyone knew that manufactures and machines were at the core of the patent system.¹⁸ Agricultural and industrial machinery was almost synonymous with "patents." For Thomas Jefferson and his cohorts, a piece of technology was readily identifiable: it had substance, and moving parts, and did something out in the practical world of farming or manufacturing. At the very least, for Jefferson, if you put technology in a bag and shook it, it would make some noise.

Against this background, it would have been seen as absurd for an entrepreneur to file a patent on a new finance technique such as publicly traded corporate shares, techniques for obtaining private financing for a bridge to compete with an existing bridge, or a security interest in uncut timber. These were the earmarks of commerce, of enterprise; laudable, surely, but something altogether distinct from the realm of "invention" and "the useful arts."

Indeed, it might well have been argued that patents on such things were precluded by the British Statute of Monopolies-a statute which itself grew out of abuses in the grant of exclusive franchises in various lines of business such as trading cards, alehouses and various staple products.¹⁹ The Statute of Monopolies, after all, prohibited the grant of monopoly rights in various lines of business, rights which had been used as a device to raise public revenue and reward court favorites.²⁰ And the line that the Statute drew precisely reflected the eighteenth century view that technology was special: only new and useful inventions could receive patents, because these were the only property rights that could enhance social welfare.²¹ The old, discredited monopolies in everyday household items were understood to contribute nothing new, except higher prices.²² A reasonable extension of the underlying logic of the Statute would have been to preclude patents on new business techniques, on the ground that the statute prohibited any special privilege to a commercial enterprise, as opposed to a new technology.

Alas, no such consensus on what patents are meant to protect exists today. Computer software-a sort of quasi-machine constituted out of written programs, or a written code that does machine-like work-has clouded and confused our working definition of "technology." Program writers, or "software engineers," labor to solve complex, demanding problems with standard toolkits, much as a determined inventor worked to design a new textile machine or seed drill in Jefferson's day. Just because the end product of today's engineering mind is manifested in a string of bits, it is no less a piece of "technology" than practical solutions of old, expressed in wood or steel.

Patent lawyers, paid to push the outer limits of what is protectable, have responded to the new technological realities with remarkable creativity. In the realm of financial instruments and Internet business concepts such as Priceline.com, the ubiquitous presence of computer technology permits inventors and their lawyers to characterize new businesses as essentially new combinations of hardware and software, and in some cases as new software packages per se. Once the Wall of Jericho holding back the forces of software patents was breached-and there can be no doubt anymore that the breach has occurred²³-the way was open for computer-related business concepts to be patented. When these software-embedded concepts are characterized as novel computer programs, there is little to separate them from any other computer program. They are therefore just as patentable as any other software.²⁴ QED.

The acceptance of business concept patents is not due simply to the underlying technology. Another important cause is the shifting baseline in the intellectual property field. Beginning in the earliest days of the patent system, and extending until perhaps as late as the early 1980s, the legal system assumed that intellectual creations were not protectable unless (very) good cause was shown. Today, it often seems the opposite. We now ask: why not protect a new form of intellectual creation? We're protecting everything else like it.²⁵

This leads us back to the Constitution. At the practical level of this essay, we ask: does Article I clause 8 tell us what to do about the Walker Digitals of the world? Can we constitutionalize the implicit understanding of the framers and early patent system actors that patents are at their core about machines and manufactures-about nineteenth century technology, in other words? The question seems to answer itself. Given a constitutional provision rooted in a blind faith in "progress," we cannot read in historically contingent limitations on patentable subject matter.²⁶ Put simply, there are no plausible subject matter limits, express or implied, in this broad, enabling clause.

If we want limits, we must look to Congress and the courts to provide them. While Congress is still a possibility, it has shown little inclination to limit intellectual property rights in recent years. And as for the courts, we have their definitive answer in such cases as State Street Bank, mentioned earlier.²⁷ In upholding claims to software inventions, the court has supplied a broad interpretation of the statutory classes of "process" and "machine." Unless something changes in the statute or the courts' interpretation of it, we can expect few subject matter limits on patents. This is one important reason we can expect to see continued increases in the number of patent applications, and hence, continued pressure on patent quality, in the coming years.

IV. Evaluating the Patent Examination System

At first glance, it ought to be easy to predict whether the whole idea of business concept patents makes any sense. All we need is a simple theory of when patents are necessary to call forth innovation, and when they are not. Once we have understood the category of things that will not be created in the absence of business method patents, we then ask the simple question: is it worth the social costs of granting exclusive property rights so that those things will be created? In essence, can we design a property right so that we gain more than we give up by granting such a right to those who qualify for it?

Probably not. Put simply, there is no easily-identified "ideal" menu of property rights for a given economy at a given moment in time. While it is clear in theory that only efficiency-enhancing property rights ought to be granted,²⁸ it is not always so simple in practice to tell what they are.

It is virtually impossible to determine-at least at this time-if truly valid business concept patents are a net drag on the economy, a net plus, or neutral. So I am not going to argue about that. But I will argue that we need to pay very close attention to the process by which these patents are granted, because, where the net effects are possibly negative, there is even more reason than usual to be concerned about improperly granted patents. I will therefore focus my attention on improving patent quality, generally.

My proposals are directed primarily at the PTO, the courts, and Congress. Because there is very little chance that any of these entities will act on them, I can be bold. My goal is to convince these people that while we may not be sure whether business concept patents are good or bad, we do know that bad business concept patents are bad. We must take steps to limit the damage from the ones already out there, and prevent more of them from issuing.

Prevention can be achieved best by revamping the patent examination system in the PTO. Business concept patents are not the only reason to make these changes, but they are certainly a sufficient reason. And they might be just the straw that tips the balance in favor of much-needed reforms without which the proud tradition of the U.S. patent system is sure to continue its slow decay.

A. Why Is Patent Quality So Poor?

There are persistent reports that patents in the software area, and perhaps especially, patents for "business methods" implemented in software, are of extremely poor quality.²⁹ People familiar with the technology involved and the history of various developments in it report that patents in this area are routinely issued which overlook clearly anticipating prior art.³⁰ The average number of prior art references cited in software-implemented business concept patents has been said to be fewer than five.³¹ Three out of the five, on average, are citations to other U.S. patents, leaving an average of two non-patent citations per patent. What is disturbing about this figure is that patents have only recently become available for this technology. Consequently, we would expect that most of the prior art in this field would be of the non-patent variety. There is every reason to believe that there is a vast volume of non-patent prior art in the software-implemented business concept field, as is widely believed to be the case with software patents in general. Given that businesspeople have been pioneering new concepts since commerce began, and that Internet commerce has seen exponential growth in recent years,³² very few of the developments in this area have found their way into patents. They are reflected instead in actual businesses, business plans, the financial services industry literature, and the like.³³ It therefore seems likely that many of the patents being issued in this area overlook highly relevant prior art. Thus, the error rate for these patents is likely to be quite high.³⁴

No doubt part of the problem is that the patent system has only recently begun to issue patents in this field. Thus, perhaps we can expect some low quality patents now, until the patent system has time to adjust. This has certainly been our experience in other fields: there were numerous complaints in the early years of biotechnology and software patents that the PTO was allowing too many overly broad patents.³⁵

At the same time, the scope of the problem seems to be worse this time. Partly, this is a simple matter of overall volume: the PTO has experienced a very rapid increase in the number of patent applications filed in the past few years.³⁶ For reasons that will be explained later,³⁷ there are numerous incentives inside the PTO to issue rather than reject patent applications. As a consequence, the number of patents issued has also grown sharply in the past few years.³⁸

The concerns about quality, especially in light of the data on overall volume, point to one conclusion: the patent system is in crisis. Therefore, this is an opportune moment to take a step back and ask an important preliminary question: how thorough should patent examinations be? A thorough analysis of these fundamental issues will help immensely in deciding whether the reported crisis is genuine.

B. Sketching an Ideal Patent Office

In this section, I discuss how we would design a patent system if we were starting from scratch today. It seems to me there are four important subissues here: (1) How much time and effort should the PTO spend on each patent application, and is there any way for the PTO to sort patent applications by anticipated economic value? (2) Inasmuch as the competitors of a company that receives a patent will potentially bear the costs of an improperly granted patent, is there any way to harness their self interest and their intimate knowledge of the technology to bear on the patent application process? (3) What is the proper "division of labor" between the PTO which issues patents and the courts which later review them-in essence, what is the ideal standard of review for the validity of an issued patent? And (4) what are the optimal remedies and punishments for acquiring an invalid patent or asserting it against competitors?

I will address the first two issues in this article, leaving the others for later analysis. Of course, it should be understood that changes in one area of the system may have important consequences for other areas. But we must start somewhere, and the PTO-as the government agency that serves as the first and important line of defense against socially wasteful patents-is as good a place as any.

1. The function of a patent office

It is curious that in all the vast economic literature on patents, virtually nothing has been written about the functioning of a patent office. When patent granting authorities are mentioned, it is usually as a "black box" bureaucracy out of which patents emerge.³⁹ There is scant literature on auctioning research projects.⁴⁰ This literature describes a proto-patent office that auctions off the right to investigate and develop a discrete and identifiable technological "opportunity." But it seems self-evident that this is farfetched enough to be disregarded. Technology is rarely so readily identifiable, and its future prospects rarely if ever well enough understood, to induce reasonable bidding. In addition, the potential developers of a technological prospect are unlikely to be so readily identifiable as permit them to assemble in a single auction.

Because of the dearth of antecedents in this area, we will have to proceed on first principles. In that spirit, I offer the following description of the goals of an ideal patent system. This is then tempered with some real-world considerations. But first, some first principles.

2. The goals of a patent office

On one level, of course, it is easy to describe the goal of the PTO. It should follow its statutory mandate closely, issuing only patents that its enabling legislation permits or deems desirable. According to this view, no patent which lacks statutory novelty, which is obvious in light of the prior art, or which includes claims that are not enabled under the terms of the statute, ought ever be issued.

This general statement cannot be faulted on one level: the issuance of an invalid patent results in some social costs that could have been avoided. (All patents, even those that are in fact completely valid, involve social costs; the only issue here is whether those costs could have been avoided by more thoroughly searching the prior art to find invalidating references.)⁴¹ Of course, the costs of invalid patents include the direct costs of filing and prosecution. There is also a myriad of indirect costs, including: unnecessary licensing fees; foregone research opportunities, abandoned or avoided by the patentee's competitors who fear infringement liability; and the activities of rent-seekers⁴² who may respond to the combination of lax patent standards and robust rewards to patentees by diverting excessive resources out of productive activities and into the "patent game."

The fundamental assumption behind public expenditures on a patent office in the first place is that, as a society, we do not want to bear the costs of a significant number of invalid patents. Indeed, as described below, historically the current system of a professional corps of patent examiners grew out of our disastrous experience with a patent registration system run amuck. The social costs of large numbers of invalid patents were considered high enough to justify the significant expense of setting up a real patent office.

But does this necessarily translate into a goal of zero invalid patents? If this is the goal, then the issuance of a single invalid patent-one that is in fact anticipated, or obvious, but that the patent office has erroneously issued-means the office has failed.

This is not only unrealistic, as we will see below; it is also inconsistent with certain signals we receive from our patent statute. If no invalid patents are supposed to be issued, then why have the independent court review of patent validity called for by our statute? Why have a mere "presumption" of validity rather than a "conclusive presumption," i.e., an unreviewable determination that patents, once issued, are valid for all time? Perhaps one reason is that, sometimes, prior art does not mature or come to light until after patents are issued. But this could be addressed by phasing in a conclusive presumption after some period of time, much as a trademark can in some cases become "incontestable" five years after it is first registered.⁴³ An argument against instituting this kind of delayed presumption is that, given the high social cost of an invalid patent, even prior art discovered very late in a patent's term should be brought to light.⁴⁴ The contrast with trademarks is obvious; at least traditionally, these are much weaker rights. Further, because there are many words and logos that might serve as adequate substitutes for a trademarked term, the social cost may be small enough that incontestability is a valid protection.

3. Pros and cons of a simple registration system

That Congress has chosen not to rely exclusively on administrative determinations of validity tells us something about the proper role of our patent system, something that we can build on in thinking about how to reform that system. But before we go on, we might want to consider the opposite extreme: why not revert to a registration system, similar to the one that was in effect between 1793 and 1836?⁴⁵ Why not, in other words, shift all the burden to the private sector, by registering any patent that comes along and letting the parties sort things out in litigation?

The argument in favor of registration is easy to make; it is what justifies the current copyright registration system. There are many copyrighted works that have either low intrinsic value (a brief trade press article, a schlocky picture), or have many close substitutes (most songs, many "genre" novels such as romances or mysteries), or both. To spend governmental resources sorting the good from the bad would be a waste of time. Instead, the copyright system lets private parties choose which copyrighted works are valuable enough to examine in detail. Then, in the course of litigation, the parties who deem it worthwhile will spend money describing why the copyrighted work is or is not protectable. Private sorting is more efficient.

This system was tried and rejected for patents, largely because of the high social cost. Private industry and Congress both concluded that the high cost of registering invalid patents was not worth whatever benefits were provided by this low "entry barrier" to inventors.⁴⁶ Even in the early nineteenth century patent litigation was complex, and therefore expensive.⁴⁷ Also, because courts then (and now) are not necessarily well trained in technology issues, the risk of error at trial was significant. Hence, there was even more expense, in the form of appeals to have an erroneous trial results reversed. Because of the cost and the potential for error, the threat value of even an invalid patent was substantial.⁴⁸ There were assertions that patents were being used to hold up bona fide manufacturers.⁴⁹ As patents became associated with rent-seeking rather than innovation, the net result was to undermine the integrity of the patent system as a whole, and thereby (presumably) reduce the incentive to innovate that patents are supposed to represent. Innovation was replaced with rent-seeking, as unscrupulous people and firms played the game of patent extortion.⁵⁰ Thus, patent examination-an increased public expenditure on patent quality-was instituted on a formal, regularized basis.

An economist reading this history would conclude that it is a classic illustration of government intervening to overcome externalities.⁵¹ Private parties, responding only to market signals, produced too many invalid patents; patent litigation mushroomed; and (again, presumably) there was an overall negative effect on innovation.⁵² In stepped the government, after having determined that the expenditure on patent examination would increase the net benefits of the patent system by reducing the social cost of an excessive number of invalid patents.

So, from the history and structure of our current system we learn these lessons: neither pure public (administrative) proceedings, nor pure private (registration system) proceedings are efficient. Our patent system envisions a mixture of public and private expenditures to determine the validity of patents. Indeed, it is part of a larger theme in patent law: the division of labor between the public and private sectors in the issuance and enforcement of these property rights.⁵³

C. Optimal Public Expenditures on Patents

Before we can determine the ideal mix of public and private expenditures on patents, we must address the public side of the ledger in isolation. The question of interest here is how to determine the correct magnitude of public expenditure on patent quality control.

In theory, the answer is simple. Following conventional principles, the public expenditure should increase until it is not worth increasing it any more-until the marginal cost and benefit are equal. To determine this, we only need to know: (1) the cost to the patent office of each additional unit of validity information; and (2) the estimated social cost of each patent, expressed as a function of the volume of validity information processed for that patent. This second element is needed to reflect some sense of reliability: if we know that an invalid patent on average costs society $X, and we know that each additional unit of search effort reduces the probability that the patent office will issue such an invalid patent by Y%, then we can determine the expected savings to society resulting from a more thorough and careful search of the prior art.

1. The benefits of sorting applications

It follows from the preceding that patent applications should be subject to differing levels of scrutiny depending on how much social cost they entail. Applications for patents that would be very costly to society-because they are very broad, for example, or because there are no good substitutes for the patented technology-ought to be examined more closely than those for minor improvements, gadgets, or novelties.

If we express this in simple qualitative terms, it means that, ideally, we would sort patents according to their prospective social cost. We could then allocate the available search resources so as to spend more resources on the patent applications likely to mature into patents with a high social cost, and less on the on the applications likely to produce patents with little social cost.

2. The problems (and politics) of sorting

While it is theoretically possible for the PTO to perform such a sort, there are currently significant barriers to doing so. One is informational: it is difficult at the time of filing to determine which applications may mature into high social cost patents. Patents are usually filed early in the development phase, and the inventor often has little idea whether or not the technology will "pan out." Hence it makes sense to delay the sorting for as long as possible. (In Japan, and to a very limited extent in the United States, applicants themselves can in some cases approximate this: they can either "activate" a pending application through filings with the patent office, or leave it dormant.⁵⁴ Early activation might be taken in this context as a proxy for higher expected private value, and hence higher social cost.)

But there is another barrier to sorting. The history and culture of our patent system reflects a broad egalitarian streak. In the patent system, by custom "all patents are created equal."⁵⁵ Any mechanism for separating patent applications would necessarily buck this tradition. It could of course be argued that proportional rationing of scarce examination resources still meets the test of equal treatment (in the sense that similar applications would be treated similarly). And perhaps this would prove persuasive. But there is still the possibility that any effort to segregate patents into various classes would be perceived with hostility by patent traditionalists. In this setting, it may prove very difficult to obtain approval for any effective sorting mechanism, which would by definition deviate from strict equal treatment.

3. Second-best solutions

If sorting is impossible for political reasons, what else might be done? Two things: (1) raise the standard of patentability and/or the filing fees, in order to induce applicants to sort out the least potentially valuable investments on their own; and (2) make a rational guesstimate regarding a reasonable average expenditure on examination, and set the overall patent budget accordingly.

The first proposal raises the cost of applying for a patent. In marginal cases, where the probability of receiving a patent is low, the value of the invention low, and the cost of applying for the patent high, prospective applicants will choose not to file. The filing fee might make the most sense as a screen; it could potentially raise revenue, and a fee increase is much easier to implement than increasing the standard of patentability. The easiest way to raise standards, conceptually, is to tighten the nonobviousness requirement of section 103. However, this is a notoriously subjective standard, and it may prove difficult, not only to draft a tightened requirement, but also to make it stick.

The second proposal is perhaps more workable: all inventors would presumably benefit from a rationally derived PTO budget.⁵⁶ In theory, the approach would simply be to set the PTO budget equal to the total social cost of all invalid patents. Then, assuming equal expenditure on each patent application, the PTO would spend an amount equal to the average cost of an invalid patent.

Note that, while valid patents would survive the examination process, so too would a certain number of invalid patents. These would be those patents that cannot be cost-effectively eliminated at the examination stage. Such invalid patents have a close corollary in the economic literature on tort law: accidents that cannot be avoided at reasonable cost.⁵⁷ As with these accidents, invalid patents that are too expensive to weed out must be tolerated. By definition, the money that would be spent to eliminate them is better spent elsewhere.

a) Why not penalize holders of invalid patents?

The tort/accident analogy suggests an interesting question: why does the legal system fail to require a patentee whose patent is invalidated to compensate an alleged infringer, all competitors, or even society in general (via a fine)? In tort law, legal damages are the negative incentive that induces precaution on the part of a potential tortfeasor. In our discussion so far, we have described the PTO as the relevant agent to determine the appropriate level of "precaution" against invalid patents. Why not shift some, or all, of this cost to the applicant?

The answer must be that we are concerned that such a rule would deter too many patent applications, and hence too much valuable inventive activity. Consider that, even though a good deal of the prior art that can invalidate a patent is publicly available, much is not. Internal developments at a competitor firm can manifest themselves in a number of types of prior art, and there is usually no way for a patent applicant to find out about this activity until after-sometimes, well after-a patent application is filed. If no amount of pre-filing search could have turned up this evidence, it is harsh and inefficient to punish a patent applicant when it comes to light.

On the other hand, where an applicant did know about a piece of relevant prior art, and failed to call it to the attention of the examiner, Rule 56 of the PTO practices results in the invalidation of the patent.⁵⁸ In addition, in extreme cases a patent applicant can be liable for up to treble damages in antitrust if he or she knowingly prosecuted an invalid patent application with an eye toward monopolizing a product market.⁵⁹ This is rare, however.

b) Who is the cheapest cost-avoider?

To complete the patent/tort analogy, it is appropriate to ask who is the cheapest cost avoider. In tort law, this consideration answers questions such as who, between two parties, ought to bear liability if there is an accident; and how should that liability be apportioned, if at all?⁶⁰

By analogy, we might ask: who is in the best position to avoid the social costs of an invalid patent? One possible choice for the cheapest cost-avoider is a public patent authority. The reasons for a public patent authority-both rational and political-have been sketched in the above sections. They boil down to these:

Up to a certain point, there may be economies of scale in doing "commodity" prior art searches such as searches of widely-available scientific and technical articles and prior patents;

There is value in a public examination function which guarantees some minimum quality level to patents, in part to prevent the most egregious patent "strike suits" or extortion attempts that depend for success on the high cost of patent litigation; and