On December 5, 2016 the USPTO will hold its second Patent Subject Matter Eligibility Roundtable to discuss issues in patent eligibility. The USPTO published a list of eighteen questions in anticipation of the event, dealing with various issues in eligibility theory, legislative changes and so forth, and requesting comments in response. Here are the answers I will be submitting to the Office.
Impact of Judicial Interpretation of Section 101
1. How has the Supreme Court’s interpretation of 35 U.S.C. 101 in the past several years affected the enforcement of patents and the development of subject-matter- eligibility law?
The Supreme Court’s current definition of a law of nature is scientifically incorrect. The analysis of this issue is quite complex, but I will attempt to summarize the issue. The Court’s definition in the Mayo case came down to: “the relation itself exists in principle apart from any human action. The relation is a consequence of the ways in which thiopurine compounds are metabolized by the body—entirely natural processes. And so a patent that simply describes that relation sets forth a natural law.”
This approach makes every natural relationship a “law of nature.” This is not how the scientific community would define a law of nature. A law of nature defines the basic, essential characteristics of reality, in a way that can be said to be 1) universally true under all circumstances, and 2) true by necessity, not merely by definition or mathematics. For example, consider the statement “All robin eggs are greenish-blue.” To be universally true, there would have to be no examples of robin eggs that are not greenish-blue. Without this universality, the statement would not be a law of nature. It would merely be an “accidental generalization,” a statement that happens to be true for a large body of observations. In essence, it would be no different from the statement that ‘all patent attorneys have registration numbers.”
Second, there must be something that makes it necessary that all robin eggs are blue. For some theories of laws of nature, necessity is what distinguishes a law-like statement from a (mere) universal truth. The obvious answer here is no, it is not necessary that all robins’ eggs are greenish-blue. We can easily imagine and likely find eggs of other colors, as a result of genetic variations.
The absence of the universality and necessity applies all biological relationships, including the relationship of genetic mutations to particular diseases, and thus prevents them from being laws of nature. This is because all biological relationships are the result of evolution. Different biological outcomes are inherent in the operation of evolution and have several sources: 1) random mutation, which is necessary for any adaptation, 2) variances in the environments that present selection pressure and 3) the existence of multiple different functionally equivalent adaptations. Stephen Gould, in Wonderful Life: The Burgess Shale and the Nature of History (1989), puts it vividly: “evolution is like a videotape that, if replayed over and over, would have a different ending every time.” Even the specific sequence of the human genome is argued to be the result of evolutionary force. Francis Crick, who discovered the structure of DNA, held essentially this view of the randomly evolved and thus arbitrary nature of the human genome. Crick's "frozen accident" hypothesis argues that the genetic code—the relationship between specific codons and the amino acids they code for—is an arbitrary relationship that resulted from all life having a common genetic ancestor. (While variants of the genetic code have been discovered, nonetheless a significant portion of the code is common across all life forms).
As a result of the Court's incorrect definition of laws of nature, the lower court's have found numerous patents to specific uses of biological relationships to be ineligible, such as:
- The effect of nutrition on the expression of least one gene in the genomic map of an animal. Hemopet v. Hill's Pet Nutrition, Inc. (affirmed by Fed. Cir.)
- Whether a dog is susceptible to "exercise induced collapse" based on the presence of a mutation at position 767 of the DNM1 gene. Genetic Veterinary Sciences v. Canine EIC Genetics.
- The production of interferon-gamma by CD4 T-cells and CD6 T-cells in response to the presence in vitro to synthetic peptides found in tuberculosis bacterium. Oxford Immunotec Ltd. v. Qiagen, Inc.
- The presence of certain mutations in the kinase domain of a patient’s EGFR gene substantially increases the sensitivity of epidermal growth factor receptor to tyrosine kinase inhibitor therapy, in the treatment of epithelial cell cancer. Esoterix Genetic Labs. LLC v. Qiagen Inc.
- The relationship between an internal body temperature to ambient (skin) temperature. Exergen Corp. v. Brooklands Inc. (affirmed by Fed. Cir.)
- The bioavailability of controlled released oxymorphone being affected by renal function. Endo Pharms. Inc. v. Actavis Inc.
- The presence of two copies of the 577R allele being positively associated with "elite athletic performance." Genetic Techs. Ltd. v. Lab. Corp. of Am. Holdings.
None of these are laws of nature properly understood, and the invalidation of the underlying patents in these cases has been in error.
As to abstract ideas, again no definition has been provided, and the Supreme Court and Federal Circuit have essentially decided that they cannot define the term. The Federal Circuit in Amdocs recently stated:
However, a search for a single test or definition in the decided cases concerning § 101 from this court, and indeed from the Supreme Court, reveals that at present there is no such single, succinct, usable definition or test. The problem with articulating a single, universal definition of ‘abstract idea’ is that it is difficult to fashion a workable definition to be applied to as-yet-unknown cases with as-yet-unknown inventions. That is not for want of trying; to the extent the efforts so far have been unsuccessful it is because they often end up using alternative but equally abstract terms or are overly narrow.
The problem is not that articulating a definition is difficult. Rather, the problem is that the courts are entirely unaware of a long tradition in philosophy addressing this very issue and that there are a number of approaches to the problem.
First, the courts have confused “abstract ideas” with “abstractions.” There is a difference between an abstract idea and an abstraction, and it is this difference that is overlooked in almost every discussion of patent-eligibility. An abstraction is a generalization; it is a term or definition that identifies the principle aspects or features of the concept that are relevant to a given context, while removing features that are not important.
One use of abstraction is classification. Some concepts reference a particular species, and some concepts reference the genus—the genus may be considered an abstraction of the species. Consider the following concrete concepts: cup, glass, tumbler, stein, pitcher, demitasse, high ball, champagne flute. These are all familiar types of beverage containers. Thus, container is an abstraction over these various species. What makes container an abstraction is that it connotes the features that are common to these objects—generally hollow, capable of holding some substance—while eliminating the particulars that differentiate them—whether their size, shape, whether they have handles, and whether they are used for coffee, beer, or spirits. But a container is clearly not an abstract idea.
Abstraction is also a process, which when applied to a given concept in a given domain for a given purpose, yields the desired generalization. It is an artifact of the English language, and not something inherent in the word abstraction, that the same word is used for both the process and the result; similarly explanation is used to refer to both the act of explaining, and the thing provided. (Though linguists and strict grammarians would prefer the Latin explanans). Thus, labeling a claim as an abstraction does not in any way answer the questions as to whether the claim is for an abstract concept or a concrete concept.
Finally, all patent claims necessarily make use of abstractions, but that does not make them recite abstract ideas. That patent claims use abstractions—as opposed to abstract ideas—is easily demonstrated. Consider the following claim:
A method of preparing a dessert, comprising:
freezing a volume of water into ice;
granulating the ice;
compacting the granulated ice; and
applying a sweet liquid to the granulated ice.
No one would argue that this claim for making the familiar sno-cone is ineligible subject matter. The claim can be broadened by applying the process of abstraction as follows:
A method of preparing a dessert, comprising:
freezing a volume of liquid into a frozen portion;
granulating the frozen portion;
compacting the frozen portion; and
applying a sweetener to the frozen portion.
Here, the specific concrete nouns, “water,” “ice” and “sweet liquid” have been replaced with broader abstractions, “volume of liquid,” “frozen portion” and “sweetener.” The result here is not a claim on the “abstract idea” of making dessert, but simply a broader claim covering different types of treats, including ice cream. Of course, the verbs here could have been abstracted as well: "freezing" could be replaced by "cooling," "granulating" by "fragmenting," etc., again altering the scope of coverage, but always resulting in a clearly patent-eligible claim on a real physical process. The same is true of any claim in software, hardware, chemistry, etc.
The process of claim drafting by abstracting from specific nouns and verbs to more general ones is not some artifice or exercise in “clever drafting.” Rather it is the fundamental way in which claims are written. Contract attorneys use particular clauses for granting a license or making an offer. So too patent attorneys select particular terms, whether narrow or broad ones, as necessary to distinguish the claim over the prior art. “The choice of which word (gerund) to use for introducing a method step is similar to the choice of which word to use for describing elements in apparatus claims. Generally, one chooses the broadest word that the prior art will allow.” Farber, Landis on Mechanics of Claim Drafting, 3rd Edition, Practicing Law Institute (New York 1990), p. 102. Indeed, the parallel between contract drafting and claim drafting is particularly apropos because patent claims form the essential part of the bargain between the grant of the patent right in exchange for public disclosure of the invention.
In view of the foregoing, it is manifest that there is a difference between a broad process claim that uses various abstractions and a claim that itself is for an abstract intellectual idea. One can have the former, without necessarily having the latter. A broad claim is acceptable, and may be patent-eligible and patentable (novel and non-obvious) as long as it is definite and supported by the patent specification.
Statutory Categories for Inventions
2. Should the patent statute be amended to further define the statutory categories of invention, i.e., process, machine, manufacture, and composition of matter? If so, please identify possible legislative changes, including which sections of title 35 should be amended, e.g., sections 100 or 101.
The statutory categories are generally based on a 19th mechanical view of the kinds of things that humans invent. After all, categories were defined before advent of computer software, biotechnology or other recent developments. The most advanced technologies in that day were highly complicated machines, including early automatons. As such, current technologies must be shoe-horned into these categories by claim drafting approaches. This makes it very difficult to claim inventions in software, such as application programming interfaces, user interfaces, virtual reality, and network-based inventions, just to new a few.
Section 101 doctrine now discriminates against inventions made using a “generic computer” when it is used in the “normal, expected manner.” This is illogical, and inconsistent with how the law treats any other type of technology. A court would not hold an invention that made use of a motor and battery to simply energize a mechanical device to be ineligible because the motor and battery operate in the normal expected manner. Indeed, it is bizarre to suggest that there would be an “abnormal” or “unexpected” manner of using a computer at all, and that that should be the basis of eligibility.
Section 101 in its own terms focuses on utility as the hallmark of eligibility, and a new use of a computer created by programming should be eligible by default. Section 100(b)—"the term 'process' means process, art or method, and includes a new use of a known ..., machine,"—sanctions the use of known machines—motors as well as computers—in new processes. However, in Electric Power, the Federal Circuit has essentially abrogated Section 100(b) by holding that using "readily available" "off-the-shelf, conventional computer, network, and display technology" is “insufficient to pass the test of an inventive concept". While we may find that a patent claim that recites the use of an off-the-shelf motor to motorize a mechanical process is obvious, we would never say it is ineligible because it merely uses the motor as a "tool." And yet the Federal Circuit's rationale, if applied generally—as any principle of patent law should be—would result in exactly that outcome.
No theoretical explanation of the Federal Circuit's expansion of Alice in this way is given, nor can there be because it is entirely contrary to the nature of both the patent law (at least Section 100(b)) and human invention. Humans frequently invent by taking conventional, off-the-shelf, readily available components and combining them into new structures and functions, all the while making use of the normal and expected operation of the components. In this sense, using a computer as a tool is no different from using any other physical device as a tool for some purpose, some other function. All inventions make use of existing components and use them as "tools" in their ordinary way to achieve a desired function; sometimes the function is known, sometimes it is new, but it is (almost) always useful. Any complex machine is made of components—whether nuts and bolts, motors and relays, integrated circuits, batteries, etc., etc.—that function precisely as they are intended: if the parts did not function as intended, we could not build anything. Similarly, we program conventional computers, using conventional languages and methods, to result in new functions—just like we create new machines by combining conventional parts using known engineering approaches.
We would not hold that an invention made of existing off-the shelf parts, used entirely for their intended purposes in the “normal, expected manner,” was not patent-eligible simply because the function performed by the parts was known, or because the parts operated in the expected manner. We would not say that the function was an abstract idea, and the application of that idea using conventional parts failed to provide an inventive contribution sufficient to be eligible. Any such analysis would clearly be seen as an obviousness analysis masquerading as eligibility. Yet, this is precisely what the Federal Circuit is guiding the lower courts, and the USPTO, to do.
Possible statutory amendments to address these issues would be:
100(b)(1) The term “process” means process, art, method, or procedure, and includes any [new] use of a known process, machine, system, computer, manufacture, composition of matter, or material, regardless of physical embodiment or means of implementation.
100(b)(2) The term “machine” means a system or apparatus, made by human agency, regardless of physical embodiment or means of implementation.
The above amendments first clarify that Section 100 is about the kinds or types of processes—which is a separate question as to whether such processes are “new,” which is what Section 102 handles. Further, the first amendment clarifies that processes include uses of a computer (e.g., computer programs, user interfaces, etc.). The second amendment clarifies that a “machine” can be a system (such as a complex arrangement of distributed computers or other machines). Both amendments allow for any physical embodiment or implementation.
Exceptions to Patentable Subject Matter
3. Do you think there should be exceptions to patentable subject matter?
There is no need for the judicial exceptions. There is a well-developed argument by Robert Armitage (See, Amicus Brief for Eli Lilly) that the existing sections (102, 103, 112), when properly implemented, do all the necessary “work” to exclude otherwise “ineligible” subject matter.
In a nutshell, the argument is this. As to laws of nature and natural phenomena, these are handled by Sections 102 and 103. By definition, a law of nature is something that existed before it was identified by the alleged inventor. What the inventor came up with is the description of the law of nature, but not the actual law. Consider that before Newton derived the laws of universal motion, objects, in fact, moved according to those laws. What Newton discovered was a way of formally describing that motion. A claim to any process describing the motion of an object according to the law would thus be anticipated. A claim to controlling or determining the motion or expected location, time, etc. of an object moving according to Newton’s laws would be an obvious application of the laws themselves.
The same goes for natural phenomena. The phenomenon must have existed in nature prior to its identification by the “inventor.” Thus, a claim to the phenomenon itself is anticipated. The appropriate question whether the application of the phenomenon to a useful purpose is claimed, and if so, then that claim is evaluated with respect to Sections 112 and 103.
As to abstract ideas, the problem is only complicated by the very assumption that there are in fact claims to abstract ideas at all. If a claim is so broad as to be “directed to” and cover the entire category of an abstract idea—say “performing intermediate settlement” (Alice) or “hedging” (Bilski) or “creating a contractual relationship (buySAFE) or any other generalized category, then it necessarily is anticipated under Section 102 (and possibly not described or enabled under Section 112). If there are minor specific details that limit the claim to a particular category (e.g., hedging as applied to quinoa instead of corn), such limitations would be evaluated under Section 103.
The key issue then becomes the treatment of “discover” in Section 101. On its face, it would appear that one who discovers a natural phenomenon or law of nature would be eligible for a patent. This problem, however, is addressed by the requirement that the discovered thing must be “useful.” A law or natural phenomenon by itself has no utility, and thus a claim to the law or phenomenon would fail under Section 112. However, when the claim recites a useful application, it recites eligible subject matter.
Amendments to Section 101 to Define the Judicial Exceptions
4. Should the patent statute be amended to define the judicial exceptions? If so, please suggest possible legislative changes, including which sections of title 35 should be amended, e.g., sections 100 or 101.
The following are possible definitions for the judicial exceptions:
101(b)(3) The terms “process,” “machine,” “art”, “method” excludes any law of nature, natural product, or abstract idea, as defined by this title.
100(k) A “law of nature” means an express statement of a physical, causal relationship governing the natural properties or behaviors of physical objects and events, and that is recognized as such by the relevant scientific community.
100(l) A “natural product” means a material, substance, or composition as it exists in nature without any purposeful change, modification, or alteration by human agency, and excludes any extracted, purified, simulated, copied, isolated, or replicated product.
100(m) An “abstract idea” means a purely mental concept or mental process as it occurs in the human mind, and that is incapable of any physical embodiment, and excludes any process performed by any machine, computer, or computer program, or any data or information stored or carried in a tangible medium.
Are Discoveries Inventions?
6. Do you think that title 35 should be amended to revise the definition for the term ‘‘invention’’ and/or provide a definition for the term ‘‘discovery’’ along with specific examples of subject matter that should not be treated as an invention and/or discovery?
No revision is necessary. A clarification in the legislative history should be made that a “discovery” means the discovery of a useful application of a law of nature or natural phenomenon, not the discovery of the law of nature or phenomenon itself. If an amendment is deemed necessary the following is suggested:
100(n) The term “discover” includes the creation of a useful application of a law of nature or natural phenomenon, and excludes the identification or description of the law of nature or natural phenomenon itself.
7. Does the concept of preemption, either separately or in the context of the Mayo two-step framework, capture useful insight in guarding against the issuance of overly broad patents? If so, please suggest possible legislative changes to capture those insights.
It is not the function of Section 101 to guard against such patents. To the extent that there is a need to “guard against the issuance of overly broad patents” that need is satisfied by Sections 102, 103, and 112 as well as by the various doctrines that restrict the basis of infringement and the doctrines of claim construction. Indeed, the phrasing of the question itself here reveals the answer: what does it mean that a claim is ‘overly broad’? Claims can only be overly broad relative to the prior art, or with respect to the disclosure (i.e., lack of written description or enabling full scope of claim). In those case, the claims are evaluated properly under the existing sections 102, 103, and 112, without resort to Section 101.
Further, the entire preemption doctrine, which originated in Gottschalk v. Benson (1972) was based on a mistaken understanding of the Supreme Court. The actual claims at issue in Benson were very narrow and easily designed around—they preempted very little at all. The Court mistakenly believed that that claims were directed to the already known algorithm for converting binary coded decimal to binary, but that was incorrect. Benson expressly admitted that that BCD algorithm was known and that his claims were limited to a specific, low-level set of operations (indeed at the level of memory registers) as one way to implement the algorithm. “The translation of binary-coded decimal numbers to binary numbers was solved long ago, indeed, as soon as BCD numbers were thought of in the first instance. Rather, the problem solved by appellants' invention is a machine problem, the slow speed and large storage capacity required for prior art translators.” Appeal to Board of Appeals, February 27, 1967, 84-85, found in Appendix to Gottschalk v. Benson at 75. The file history in Benson explains: “More specifically, the present invention is directed toward simplifying such numerical conversions by simplifying the apparatus necessary to achieve this conversion and increasing the speed at which conversion can be accomplished.” Id. at 52. In short, Benson’s claims was not dictated by the axioms of mathematics as believed by the Court (based on the misleading arguments of the Solicitor General) but rather was a specific improvement on existing BCD to binary conversions, designed to take advantage of the speed differences of then-available shifting registers over memory circuits. Thus, Benson’s claim did not “preempt” the BCD to binary algorithm as believed by the Court.
Finally, it should be remembered that the Supreme Court itself once considered preemption not as a problem, but a sign of the importance of the invention:
It may be that electricity cannot be used at all for the transmission of speech except in the way Bell has discovered, and that therefore, practically, his patent gives him its exclusive use for that purpose; but that does not make his claim one for the use of electricity distinct from the particular process with which it is connected in his patent. It will, if true, show more clearly the great importance of his discovery, but it will not invalidate his patent.
The Telephone Cases, 126 U.S.1, 535 (1888).
Patentable Subject Matter in the Life Sciences
11. To what extent should a ‘‘diagnostic method’’ be eligible for a patent? Please provide specific examples.
Diagnostic methods should be generally eligible. First, as pointed out above, the Supreme Court’s definition of a law of nature is incorrect. Methods of diagnosis based on relationships between biomarkers and diseases are unequivocally not “laws of nature.” That a particular disease results more frequently when a given gene mutation is found is not a law of nature—it is the result of evolution. It could have certainly turned out that the particular mutation had no relationship to the disease at all. Indeed, that is already what is the case: in many situations having a particular gene mutation increases the likelihood of a getting the diseases, but it does not necessarily cause the disease itself. In other words, as discussed above, the relationship between the mutation and disease is not even universally true, let alone necessary.
The Mayo case is a perfect example. There, the defendant, Mayo Clinic, has come up with different thresholds than those claimed by Prometheus for the toxic and effective ranges of the metabolites. Using the Court’s logic that would mean there were multiple different and conflicting laws of nature here, Mayo's laws and Prometheus's laws. And by extension, it would mean that every toxic and effective dosage relationship on every drug (there are over 1450 drugs approved by the FDA) would be a law of nature. Clearly, this are preposterous outcomes, demonstrating again the falsity of the Court's definition.
From this, any such diagnostic relationship does not necessarily identify the “true” causal mechanism for the disease. This follows from the well-known principle that correlation does not imply causation. It may be that the true cause of the disease is some other unidentified agent that also causes the identified mutation in the biomarker. Or it may be that the disease itself causes the mutation, not the other way around. In short, the statement that “the presence of mutation A is correlated with an the risk of disease B” is not a statement of a law of nature, even if it is a true statement of a statistical fact. Thus, a claim to detecting the presence of mutation A to diagnose disease B is not a claim to a law of nature.
Second, laws of nature are found empirically. To assume that all natural relationships are laws of nature without empirical evidence is simply scientifically unsound. Many statements of diagnostic markers for diseases turn out later to be incorrect, either because of poorly designed studies or the discovery of the actual underlying mechanism, or other factors. This is very common in the biological sciences. For example, some researchers suggested that Alzheimer’s disease was caused by high levels of exposure to aluminum or zinc, or the result of a virus. All of these statements have turned out to be false; and yet a patent claim that used the presence of trace aluminum or zinc or some viral agent to diagnose Alzheimer’s would be rejected as an ineligible law of nature. (And again, they could not all be true at the same time). These results would be absurd on their face. Thus, a claim here should not be denied as a law of nature; at best it would be denied as lacking utility. At worst, the claim is allowed and the patent granted on a method that would not be used at all because it is ineffective. In that event, there is no harm to the public since no one would use the “invention.” Judge Rich, the author of the 1952 Patent Act was keenly aware of this possible outcome. He wrote:
If it [the invention] is a total dud, how is the public injured by a patent on it? A monopoly on something nobody wants is pretty much of a nullity. That is one of the beauties of the patent system.
Rich, The Principles of Patentability, J. Patent Office Society, Feb. 1960, vol. 52, No. 2, p. 75-91, at 85.
Third, the use of that relationship to diagnose a disease is a useful application of that relationship. It must be remembered that the characterization of something as a “disease” is a human construction for human purposes: nature does not “know” of diseases at all, only organisms that exist subject to mutation and environmental conditions. That a certain complex of bodily states and function is considered a “disease” is only such because we, as humans, decided it to be one. Nature does not distinguish between “healthy” and “diseased” organisms or tissues. Mutations occur in the human bodily constantly, but only some of them are considered to be diseases. A mutation causing a cancer tumor is just as “natural” an event as a mutation causing a birthmark; only humans decided that cancer is a “disease” and birthmarks are not. It follows the decision to associate a particular biomarker with the presence of a biological state that we call a “disease,” is the human originated, useful and intentional application of the underlying biological facts. It thus meets the current statutory definition of invention.
Process Patents and the Machine or Transformation Test
14. Should patents be available for methods that do not involve a machine or a transformation? If so, please provide specific examples.
Yes. First, there is a definitional problem: What is a “transformation”? At a reductionist level, all processes, including information processing, transform the atoms from one state to another. Similarly, what counts as a “machine”? Is a collection of DNA that performs biological computations a “machine”? What about a pair of atoms that yield quantum information? Without suitable definitions of machine or transformation the test is no better than the Alice test.
Focusing on the presence of transformation excludes some types of useful inventions. One example that does not offer clear a physical “transformation” are process claims for user interfaces. User interfaces are a well-developed field of technology, part of the field of human factors and ergonomics. User interfaces and the operations that they provide are inventions no different in kind than other inventions: they enable or improve the use of a machine or the performance of a method. However, it may be difficult to consider the structure or operation of a user interface as either a machine or providing any transformation. A process claim for the operation of user interface may provide useful results without clearly “transforming” anything other than an arrangement of icons, the appearance of the user interface, or causing changes in underlying operations of a machine. Such user interfaces are useful, they should be eligible regardless of whether there is a physical transformation.
Patentability of Business Methods
16. To what extent should an invention that involves a business method be eligible for a patent? Please provide specific examples.
Business methods are no different in kind than any other kinds of inventions. That such methods involve financial relationships does not make them suspect: after all, humans invented the concepts and implementations of money, contracts, accounting, as well as banking, finance, insurance, and so forth. All of these developments were the product of human creativity to solve particular problems. Such inventions are useful and should be eligible subject matter.
Historically, the USPTO granted patents on business methods going back to the 1800s. There are hundreds of patents for recording abstract of real estate title, sales and accounting books, and ledgers for recording securities, loans and other financial arrangements, tickets and coupons, as well as mechanisms for advertising and gambling. In those days, the inventions were claimed by linking them to the manner in which the useful business information was recorded: paper. Claims were directed to books, ledgers, and other forms of paper with specific information that defined the business aspects of the invention, say the arrangement of columns or rows on a ledger sheet to record a sales transaction. Today, the mechanism of recording information is of course computer memory. If business methods using paper “memory” were historically eligible subject matter, then doing so using computer memory should be as well. Thus, there is historical precedent for claiming business methods that make use of computers.
Patentability of Software/Computer- Related Inventions
17. To what extent should an invention that involves computer software be eligible for a patent? Please provide specific examples.
Software based inventions are patent-eligible and should be. The arguments against software patents (e.g., software is abstract, a language, just math) typically misstate the nature of the technology or the facts. Software is not abstract by being a set of instructions to a machine, any more so than the physical configuration of a machine that performs a desired function (e.g., a circuit designed to perform the same function as the software) is abstract.
Software in terms of programming languages are not like normal human languages. You cannot use a programming language to make a promise or to lie to the computer, create a metaphor or a simile that the computer can understand, or tell a joke to the computer and expect it to laugh in return. While programming languages use what appears to be normal words (“print”, "get", "if...then", "do", etc.), these are just arbitrary tokens for our benefit; to the computer they have no meaning other than linking to a specific functional operation of the computer. And while we can program a computer to be used to output promises, or tell lies or jokes, the underlying programming language is not performing those acts, but rather the overall program under human control.
Nor is software just mathematics; this confuses the underlying binary operations that are performed to control the digital logic of the computer with the higher functions actually performed by the computer. This is like saying that running a mile is really just cells in your body processing carbohydrates and fats for energy to power your muscles. That may be true, but it fails to describe what it is actually happening.
Patents protect structures (machines, apparatus, and compositions of matter) and functions (process and methods). Software is a way of configuring machines (computers) into new structures to perform new functions. In that regards, software satisfies the requirements of the statute.
“Overly Broad” Patents?
18. What mechanisms, other than the judicial exceptions, can be used to prevent issuance of overly broad software or computer-related patents that cover wide swaths of economic activity? Do you think that other provisions of title 35 (enablement, written description, definiteness, novelty, nonobviousness) could be used more effectively to achieve this goal? If not, please explain why.
To the extent that there is a need to “issuance of overly broad software or computer-related patents that cover wide swaths of economic activity” that need is satisfied by Section 102, 103, and 112 as well as by the various doctrines that restrict the basis of infringement, and the doctrines of claim construction. It is not the function of Section 101 to guard against such patents. Indeed, the phrasing of the question itself here reveals the answer: what does it mean that a claim is ‘overly broad’? Claims can only be overly broad relative to the prior art, or with respect to the disclosure (i.e., lack of written description or enabling full scope of claim), and thus are evaluated properly under the existing sections 102, 103, and 112.
The more subtle problem is the suggestion that “cover[ing] wide swaths of economic activity” is a proper justification to deny a patent. A patent examiner working on a given software patent will have little if any sense of how “wide” a swath of economic activity a given patent claim will cover now, let alone in the future. Patent examiners can only deal with the facts before them, not speculation about how a given patent claim may impact economic activity in the future. A patent claim on what appears to be an or dinary invention at the time may turn out to be enormously important in the future. Or it may turn out, as Judge Rich acknowledges, to be a total dud. What evidence could the examiner have to make such a finding? And if a patent examiner can reliably tell the difference, she shouldn't be examining patents, she should be investing in the stock market.
Let's assume that some patent will cover a wide swath of an emerging industry. That does not justify the examiner denying the patent claim. The invention may be a pioneering invention in the field, as noted by the Supreme Court in the Telegraph Cases. Would the USPTO have denied the Wright Brothers or Thomas Edison their patents simply because the airplane or the light bulb become important economically? Would it have denied Rivest, Shimon and Adair their patents on public-key cryptography? Would an examiner at that times of any of these inventions have been able to predict their importance, for example, that public-key cryptography would become essential to modern communications and business? Not likely. In short, this so-called “need” cannot be used to justify an examiner denying issuance of a patent.