The John Marshall Journal
of Computer & Information Law
Volume 17,
Issue 1
Patentability of Computer Software Instruction as an "Article of Manufacture": Software as Such as the
Right Stuff Vincent Chiappetta*
17 J. Marshall J. Computer & Info. L. 89 (1998)
Note: The footnotes have been omitted from the web version of this article. The complete text of the article may be obtained in printed form or via Lexis, WestDoc, or UnCover.
I. Introduction
The last five years have witnessed a dramatic shift in the approach taken by the Court of Appeals for the Federal Circuit ("CAFC") and, under the CAFC's stern if somewhat incomplete guidance, the United States Patent and Trademark Office ("PTO") to the seemingly intractable problem of determining whether software inventions qualify as patentable subject matter under the United States patent laws. Beginning with a series of CAFC decisions in 1994 and culminating with the PTO's issuance of its Final Examination Guidelines for Computer-Related Inventions (the "Guidelines") in 1996, the paradigm shifted from a "mathematical algorithm" based analytic structure to an apparatus model driven by the presence or absence of related computer hardware in the claims.
Has this new approach finally brought the 30-year quest for a solution to a close? Certainly, abandoning the mathematical algorithm approach is a proper, and indeed vital, first step toward untangling the confusion surrounding the patentability of computer software. And from the applicant's and examiner's perspectives the new regime establishes a more well defined and objective set of requirements which when followed results in much greater certainty in the examination process. Unfortunately however, this promising alternative leaves the Gordian knot uncut. This article examines the current treatment of the increasingly important "article of manufacture" variant of the new hardware model and explains logical gaps that still remain. The results of this inquiry are used as the basis for a proposed new software as implementation versus software as language approach to resolving the software patentability conundrum.
I. Background
The starting point for any software as patentable subject matter analysis must be first principles: the Constitutional policy objective of promoting "progress . . . of . . . [the] useful arts" and the nature of software itself. In this framework the normative answer to the question of "is software patentable subject matter" is straightforward enough; only when the claimed software is in the useful arts. Properly drawing this line, however, requires more than a tautological statement of the obvious. The analytic structure must be able to accurately test the true nature of each particular claim to a software invention for compliance. Only by ensuring that exclusively deserving software wines are put into patent bottles can the patent system lay claim to "appellation controle."
Measured against this decision-making backdrop the administratively appealing hardware focused approach, particularly as articulated in the Guidelines, fails to meet the need. The talismanic reliance on the presence or absence of computer readable carriers results simultaneously in "over" and "under" inclusive application of the patent incentive to software inventions.
Under the current state of play, over-inclusion is most apparent. Because the Guidelines overstate the case for article of manufacture "hardware carrier" style software claims as technology, a significant opportunity exists to smuggle non-useful arts claims into the patent system under their guise. In addition, the resulting failure to clearly and properly define the actual nature of software inventions by applying the patentable subject matter analysis leads to inadequate identification of prior art and insufficiently stringent review for novelty and non- obviousness.
As a consequence, examination of software applications is inappropriately weakened resulting in a disruption of the economic cost-benefit balance underlying the U.S. patent system. Rather than providing appropriate incentives for innovation in the technological arts, the current system offers the possibility of obtaining patents covering inventions outside the targeted class. This results in an industry rush to build ever more impressively sized portfolios of at best marginal and at worst inappropriate patents couched in software terms. Large amounts of time and energy which could be otherwise invested are directed to obtaining, assessing, avoiding and litigating software patents which have come through an examination process that makes inadequate distinction between justified and unjustified exclusionary rights. One must take seriously the specter of a software industry dominated by patent "haves" holding each other at bay by brandishing large binders of patents as deterrent threats of mutual destruction while offering the "have nots" unattractive, or even impossible, "pay or perish" decisions. The inappropriate application of the patent law incentive becomes a source of economic inefficiency, imposing unjustified costs, impeding progress in the very fields it was designed to promote, distorting competition and reducing rather than enhancing the general social welfare.
We must be equally wary of the less apparent under-inclusion lurking in the current software as article of manufacture analytic framework. The Constitutional policy objective is not to restrict or eliminate patent protection for software related inventions. It is to ensure that the patent incentive is directed in ways that generate invention and progress in the technological arts, including, when appropriate, in the field of software development. The imposition of a hardware carrier requirement, which is unnecessary to proper application of the patent laws to software inventions, prevents direct protection of software innovation per se. The effects are not inconsequential. Inventors are forced to mischaracterize software innovations in hardware incarnations in a "real world" technological environment that is heading in precisely the opposite direction by moving computer system functionality from hardware to software. The system currently, therefore, inhibits the ability to describe and protect the true nature of software inventions under the patent laws. In addition, including the unnecessary carrier element in the claims provides a basis for irrelevant and unmerited non-infringement defenses completely unrelated to the underlying functionality (where the actual dispute resides) thereby unjustifiably increasing enforcement costs. The overall result is too little incentive to innovations in this important field of technological endeavor.
III. Analysis
This Article argues that the solution lies in distinguishing between two possible functional uses of software when making the "useful arts" policy inquiry. First, software can act as the specific technological vehicle for implementing new or existing activities as part of a computer system, the actual instructional program for the hardware. Second, software can serve as a language, which like any other language is a symbolic vehicle for communicating an underlying idea, in this instance the contained methodology.
The test for whether a software invention constitutes patentable subject matter should be tied expressly to drawing this implementation versus a language dichotomy based on the patent claims. If the claims only cover the use of the software functionality as instruction in a computer system the software serves solely in its implementation role, then such claims should be routinely treated as patentable subject matter. In contrast, if the claims are not so limited then software terminology is being used as a language to express an idea. This situation does not make the functionality claimed per se unpatentable. It does require, however, that the claims be assessed based solely and directly on the underlying functionality, apart from the fact that software language has been used as the means for its expression.
By eliminating the faulty reliance on the presence or absence of a hardware carrier, the implementation versus language test resolves both the over-inclusion and under-inclusion difficulties in the current article of manufacture regime. Over-inclusion disappears because claims, which reach beyond computer system implementation of the software functionality, cannot hide behind mere inclusion of computer readable media as a means to sneak through § 101. Under- inclusion is eliminated because the test permits direct treatment of "software as implementation" functionality as patentable subject matter without forcing such inventions into hardware formats or requiring surplus claim elements to avoid potential rejection under inappropriate process tests.
To be effective an appropriate patentable subject matter test must also permit and enable proper operation of the other patentability requirements. The proposed implementation versus language model does so. First, it clarifies how the progress of the useful arts objectives inherent in the § 102 novelty and § 103 nonobviousness reviews should apply to software claims. For software as implementation inventions the novelty and nonobviousness must be found in the technology which constitutes the substance of these claims: normally the computer technique or the fact of computerization (automation) itself. On the other hand, software as language inventions are not permitted to look to the computer software label or expression as a source of novelty. They must stand or fall solely on the progress or innovation found in the underlying expressed functionality. Additionally, clarifying which type of software invention is under review, computer technology or underlying functionality, facilitates the location of, and comparison with, appropriate prior art by applicants, examiners and the courts. Finally, the model avoids any distorted use of the § 112 definiteness and enablement inquiry as an ineffective tool for distinguishing between software as implementation and software as language inventions based on abstract idea notions.
The proposed model will result in a more appropriate application of the patent incentive to the field of software technology. By avoiding the over-inclusion problem inherent in the present analytical framework, the proposed model helps eliminate undesirable software applications and patents. The software as implementation limitation under § 101 and the increased efficacy of the § 102 and § 103 reviews will likely result in less frequent grants of software related patents. The patents which actually do issue will be more clearly defined and of more focused scope. The net result will be fewer but more appropriate and stronger software patents, precisely the types of patents current patent policy strives to promote industries internalizing as the cost of providing appropriate incentive to software inventors.
A transition to the new model can be effected without significant effort. Because much of the existing software patentability muddle is of judicial origin, the courts can equally quickly retrieve us. The CAFC can adopt the proposed implementation-language analytic framework simply by modest clarification of the rationale supporting their current article of manufacture line of analysis. With this accomplished, the PTO can modify the Guidelines and the system will be fully operational.
This article develops the case for the proposed implementation versus language analytic framework in several steps. Part A briefly sets out the origins and basic objectives of the United States patent system and explains the roles played thus far by the primary United States patent law requirements in the software patentability debate. Part B provides background and context for the article of manufacture discussion. Specifically, Part B provides a concise historical review of the development of the software patentability debate from the original appellate decisions through the various mathematical algorithm formulations to the CAFC's In re Alappat and State Street Bank & Trust v. Signature Financial Group "machine" decisions. Part C provides a discussion of the evolution of the current article of manufacture approach, including the relevant 1994 CAFC decisions and the PTO Guidelines. Parts D, E and F explain why the current article of manufacture approach set out in the Guidelines gives rise to the problems of over-inclusion and under-inclusion. Part G describes the implementation-language alternative system of analysis and explains why it consistently provides results congruent with the policy objectives of the United States patent system.
A. Origins of United States Patent Law and Relevant Patent Law Basics from whence They Came and Where They Seek to Go
The general origins of United States patent law are relatively clear. After a brief debate between James Madison and Thomas Jefferson concerning the general undesirability of monopolies, Jefferson relented in the case of incentives for certain types of inventive behavior. The United States Constitution was drafted to include an express congressional power "to promote the progress of science and useful arts, by securing for limited times to authors and inventors the exclusive right to their respective writings and discoveries." This provision not only authorized the creation of a federal patent law but, uniquely among the grants of congressional power, also expressly set out the objective and even some details concerning the means. An incentive in the form of a limited right to preclude competition could be offered to inventors to encourage the investment of their time and resources in inventive endeavors which would progress the useful arts. The hope was that this encouragement would result in the desired growth in the useful arts which, in turn, would result in overall economic growth and a general increase in prosperity.
The Constitutional provision does not require Congress to create such incentives, it merely empowers it to do so. At the founding of the nation pro-industrial sentiment was strong and Congress acted with alacrity, adopting the first United States Patent Act in 1790. The United States has had a patent law ever since. Although there have been numerous revisions, there has been virtually no change in the patent law's generalized pursuit of the Framers' desire to incentivize progress in the useful arts. This objective remains the driving force behind the patent system and, therefore, continues to provide important guidance when determining to what extent claims to software related inventions should receive patent protection.
1. The Basic Patentability Inquiry
The Patent Act and associated PTO regulations and guidelines contain a number of requirements which must be met to obtain a patent. The five key requirements for purposes of the present inquiry include the patentable subject matter restriction, novelty, nonobviousness, usefulness and adequate description (identification and enablement). Each of these requirements plays a particular role in ensuring that the patent incentive furthers the desired Constitutional objective of promoting progress in the useful arts for the general public's benefit.
The patentable subject matter requirement keeps the patent incentive aimed exclusively on endeavors within the Constitutionally targeted useful arts. Section 101 of the Patent Act lists the following classes of patentable subject matter: "any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof." The breadth and generality of this language when coupled with the two different functions of computer software described above has made actual application somewhat problematic. Much of the software as patentable subject matter debate reflects a concern that claims implicating unpatentable subject matter will be smuggled past the § 101 border guards by expressing them in software "technology" terms. The patentable subject matter requirement has been, and continues to be, a prime focus of the software patentability debate.
The remaining requirements play additional roles in the patent policy framework. The § 102 novelty and § 103 non-obviousness "inventiveness" requirements and the § 101 usefulness requirement all seek to ensure that the invention is truly an inventive step forward (progress) in the useful arts. Finally, to ensure the invention involves actual implementable technology, to give notice of the extent of the exclusive patent privilege and to make the invention available to society at the end of the patent term, § 112 requires a sufficiently detailed description to permit those skilled in the relevant field to clearly identify, understand and practice the invention.
The novelty and nonobviousness "inventiveness" requirements have had three important connections with the software patentability debate. First, they have been a matter of significant logistical concern to the PTO. Particularly at issue are whether adequate numbers of properly trained patent examiners can be employed and whether the technology that already exists (the "prior art") can be reliably identified, both being essential to properly determining if the applicant's invention is truly novel and nonobvious. These pragmatic institutional concerns have been one motivation for the PTO's long standing position that computer software was not patentable subject matter. Although institutional considerations should not be used indirectly to undermine patent system objectives, we must recognize that for the patent system to operate properly, appropriate resources are required. To date the issues have been managed, albeit with some difficulty. However, if it becomes clear proper resources do not exist or are not funded, then the policy objectives need to be re-thought in light of those realities.
Second, because the confused state of the patentable subject matter debate under § 101 frequently prevents a clear determination of the precise nature of the invention at issue, many software related inventions fail to have any significant applicable prior art cited against them. This is an inappropriate and unfortunate result. The patentable subject matter question is only one of the patent policy gatekeepers. If the patent system is to operate properly the § 101 review must adequately inform the novelty and nonobviousness examination to ensure those requirements are also satisfied. This is particularly true in the case of many computer software inventions, where the key question frequently is whether the claimed computer implementation is obvious given the current state of computer technology.
Third, the inventiveness requirements have in the past become entangled with the § 101 analysis, confusing both discussions. At a policy level the two tests embody separate objectives. The patentable subject matter requirement ensures that the incentive is being applied to activities in the proper field of human endeavor, those in the "useful arts." The inventiveness requirements address whether the particular results represent advances or progress in those arts. Finding that an invention involves patentable subject matter says nothing about the "inventiveness" aspects of novelty and non-obviousness or vice-versa. These separate issues should involve separate inquiries.
A pair of simple examples can demonstrate why this is. Dramatic works are not technology and do not come within the "useful arts" target class of inventions entitled to the patent incentive. As an entire category they simply are not patentable subject matter. This determination is entirely independent of questions of novelty. Non-patentable subject matter remains non-patentable subject matter no matter how novel it may be. It is irrelevant whether the particular play is new and nonobvious or a verbatim copy of a previous work.
Conversely, lack of novelty in no way negates the fact that a particular invention is § 101 patentable subject matter. For example, a patent application covering a computer dynamic random access memory chip clearly involves patentable subject matter under § 101. If the same invention has been previously patented, no patent will issue because the previous patent demonstrates lack of novelty. This finding of lack of novelty, however, does not make the memory chip any less patentable subject matter, any more than a finding of novelty makes the play § 101 subject matter.
Despite this lack of inter-relationship, the Supreme Court's Parker v. Flook decision in 1978 appeared to legitimize a "point of novelty" approach to the § 101 inquiry. The Court's approach involved separating the claimed invention into old and new aspects as a way of identifying where the invention (the "new") was located and then testing only these "inventive" aspects under § 101. This technique quickly became popular with the PTO in the early days of the software as mathematical algorithm test, as it permitted the examiner to parse the claims, identify the invention as the underlying unpatentable mathematical formula and thereby justify a § 101 rejection. A three-year debate followed over whether a search for the "point of novelty" in the invention was an appropriate part of the patentable subject matter inquiry.
The approach was fully and finally discredited by the Supreme Court in Diamond v. Diehr. The majority opinion in that case pointed out that under a "point of novelty" parsing exercise, any invention in which the novelty involved a combination of old elements or old and new elements would be improperly rejected. Therefore, the Court held that the "claims must be considered as a whole," without consideration of novelty, for purposes of § 101.
The description and enablement requirements under § 112 play a central role in any patentability inquiry. The most critical portion of any patent application is the specification, consisting of the detailed description and the claims, which together define and limit the actual invention for which the patent is sought. The specification is, therefore, the basis on which a software related invention should be characterized as an "article of manufacture," a "process" or something else. The claims, in particular, must be the focal point for any framework used to determine whether an application describes and is limited to § 101 patentable subject matter.
The last requirement, "usefulness" historically has been called into service to justify arcane results like the non-patentability of perpetual motion and gambling machines. The usefulness hurdle currently is extremely low, effectively inquiring whether the invention performs as claimed regardless of the purpose. To date, the requirement has not had a significant effect on the software patentability debate.
The Guidelines employ a form of "usefulness" test (termed "utility"). Although the terminology sounds similar, the Guidelines "usefulness" test is actually related to the patentable subject matter requirement: acting as an examiner's guide to whether an invention has a practical application with "real world" value in the technological arts as opposed to being a mere idea or concept. Although the Guidelines contain some serious implementation difficulties, this article argues that the PTO was certainly on the right road, or at least intersected with the right road, in moving in this direction. Focusing on "usefulness/utility" of a software invention in a "useful arts" sense contains the key to resolving the software as patentable subject matter conundrum.
B. A Short History of Software Patentability Up To Article of Manufacture Claims: How the World Started to Change With Alappat.
The first significant software case, In re Prater, was decided by the Court of Customs and Patent Appeals ("CCPA"), the predecessor to the CAFC, in 1968. In this and other early cases, the CCPA exhibited a generally robust view of the patentability of inventions arising out of the emerging computer software industry, eliminating the problematic "mental steps" doctrine and fairly consistently facing down a reluctant PTO.
This generally positive approach was abruptly cut off in 1972 by the Supreme Court's first word on the subject in Gottshalk v. Benson. In Benson the Court announced in fairly incoherent terms a § 101 test which precluded patent grants to any software invention which preempted the use of a mathematical algorithm. The opinion, which can at best be described as skeptical concerning the patentability of software, was interpreted by some (including most notably the PTO) to indicate that computer software inventions should be viewed as per se unpatentable subject matter. The decision refocused subsequent debate between the CCPA, which continued generally to favor patentability, and the PTO which did not, on the meaning and extent of the Court's mathematical algorithm preemption test.
The CCPA-PTO contest was further complicated by the Supreme Court's 1978 decision in Flook which reinforced the Benson mathematical algorithm preemption approach. The decision not only strengthened the view that the Court believed the entire class of computer software related inventions was potentially unpatentable subject matter, but added the confusion of the "point of novelty" approach to the analysis discussed above. The PTO embraced the decision, setting off yet another round of § 101 rejections and appeals to the CCPA.
The rising tide against the patentability of software related inventions finally started to recede with the Supreme Court's pivotal 1981 decision in Diehr. Following Benson, the CCPA had begun to develop what was to become the Freeman-Walter-Abele ("FWA") two-step test as a vehicle for controlling the confusion caused by the mathematical algorithm approach. Although the Court did not expressly overrule Benson or Flook, or confirm the FWA approach in Diehr, the opinion discredited the argument that software was per se unpatentable and lent considerable support to the substantial limitations that the CCPA had been building around those decisions through the FWA test.
In its final post-Diehr incarnation, the FWA test effectively narrowed the Benson/Flook mathematical algorithm § 101 rejection from virtually all software claims to software related inventions which involved a "mathematical algorithm" and no external physical connections or effects (read generously in favor of the applicant). With the patentable subject matter inquiry directed by the FWA test, relative peace reigned between the CCPA/CAFC and the PTO concerning software patents through the 1980s and early 1990s.
In 1994 the software as patentable subject matter dispute between the CAFC and the PTO re-erupted on a new front. The conflict arose over the proper analysis of new hardware-software combination claim formats which had been developed by applicants to address continuing difficulties under the FWA regime.
Because Diehr failed to expressly overrule either Benson or Flook, the FWA test remained predicated on, and limited by, the mathematical algorithm approach. Determining when the PTO or the CCPA might find claims involved a mathematical algorithm, meaning the application would have to run the FWA gauntlet, was something of a Delphic enterprise. In addition, claims to software applications in which a mathematical model was a prominent feature and there was no demonstrable, non-trivial pre- or post-solution external activity, were certain to attract undesirable attention under the test. Any attempts to extend claims toward purer software-only inventions, therefore, held great possibility for misadventure before the PTO and even the CAFC.
Software development companies also sought to resolve the enforcement difficulties inherent in the PTO's and CAFC's treatment of most software inventions as claims to the underlying processes. Process patents are directly infringed only by use of the process. This meant that although a competitor might be distributing the enabling means (the software for implementing the process), only the actual end-user of the software who actually was implementing the process could be attacked as a direct infringer. Even if a patentee could overcome the commercial difficulty that the end-user was the patentee's own customer or potential customer, the problems of making multiple demands and bringing multiple suits against widely dispersed and hard to identify defendants remained. The only alternative was a more problematic contributory infringement claim against the competing software distributor. A more efficient and less problematic enforcement alternative was clearly desirable.
Inclusion of associated hardware platforms (a computer) or computer readable media carriers (diskettes, ROMs or the like) provided a vehicle for resolving both these difficulties. The theory was that by adding the computer or carrier, claims covering a software implemented process could be transformed into a "machine" (consisting of the software loaded in the computer) or an "article of manufacture" (the software loaded on the computer readable carrier). Because machines and articles of manufacture are expressly listed in § 101 as patentable subject matter, this reclassification would eliminate the application of the FWA test and its attendant difficulties.
Additionally, patents in these formats provided a means of direct enforcement against competitors. Because the claims expressly cover the software and diskette combination or the software pre-loaded into a computer, any competitor making or selling the software "bundled" in these forms would directly infringe. The article of manufacture format will generally be the preferred alternative. In many situations the competitor will distribute the software separately from the overall computer system. In these cases the only infringement of a machine format claim would be by the end-user customer when she loads the software into the computer. The article of manufacture format, however, permits direct attack against a competitor's media based manufacturing and distribution activity. More importantly, the article of manufacture format is the minimal claim format. In most cases the patent holder will be able to successfully argue that software on a disk or memory device incorporated in the computer infringes the article of manufacture claims, even though it is being sold as part of a complete system.
The "machine" version of these hardware-software combination formats was tested before the CAFC in In re Alappat. Alappat's claims were written in "means for" terms. Both the applicant and the PTO agreed the claims were broad enough to cover a general-purpose digital computer specially programmed to perform the specific task (line smoothing on a pixel matrixed oscilloscope screen). Based on this reading of the claims, an eight member PTO review panel assembled for the reconsideration hearing (the PTO Board), treated the application as de facto claiming of the underlying process. It then applied the FWA/mathematical algorithm test to the process and found it involved non-statutory subject matter. The CAFC, sitting en banc, rejected the PTO analysis. The court held that, properly interpreted in light of the specification's recitation of specific computing elements, the claims were limited to a "specific machine to produce a useful, concrete and tangible result" and as such, patentable subject matter.
The court's specific ruling concerning the claim coverage of a general purpose computer running special purpose software was, however, somewhat unclear. The court held that merely because claims read on a specially programmed general purpose digital computer does not automatically render them non-statutory: "a computer operating pursuant to software may represent patentable subject matter, provided of course, that the claimed subject matter meets all of the other requirements of [the Patent Act]". Concerning when such claims actually are statutory, however, the court only provided cryptic guidance. The opinion does quote earlier CCPA and CAFC holdings that "[w]e have held that such programming creates a new machine, because a general purpose computer in effect becomes a special purpose computer once it is programmed to perform particular functions pursuant to instructions from program software." Thus, in the court's view, at a minimum the introduction of the software creates a new machine. The opinion, however, left it ambiguous whether this newly created machine is itself statutory subject matter (i.e., with the "other requirements" referring only to novelty, nonobviousness, definiteness and usefulness), or if something additional is required under § 101 and, if so, what that something more might be.
In 1998, the CAFC had a second opportunity to address the "machine" format in State Street Bank & Trust v. Signature Financial Group. The case involved claims to a computer system for making allocations of assets, income, gains and expenses in a multi-member mutual fund partnership. The United States District Court for the District Massachusetts had originally heard the case as a declaratory judgment action. That court held the patent invalid on patentable subject matter grounds. The court found that although the claims where written in "means for" apparatus format, they de facto covered the underlying process. It went on to hold that process involved a mathematical algorithm which did not satisfy the FWA test.
The CAFC reversed. Applying the same approach used in Alappat, the court read the supporting structure contained in the specification into the respective means stated in the claim and held that the patent "properly construed, claims a machine" and "[a] ‘machine' is proper statutory subject matter under § 101." Addressing the lower court's mathematical algorithm analysis, the CAFC found that the mathematical algorithm doctrine actually raised abstract idea concerns and should be reviewed accordingly. As a result, the critical issue was whether the algorithm/abstract idea had been "applied in a ‘useful' way," that is to produce "a useful, concrete and tangible result." The court went on to hold that as the transformations performed by the claimed computer system did produce such a result, the patent involved patentable subject matter. The court noted that in this analytic framework "the Freeman-Walter-Abele test has little, if any, applicability to determining the presence of statutory subject matter" and, in particular, should not have been applied in this case.
The State Street Bank decision has two important ramifications for the hardware claims approach. First, the opinion appears to resolve the Alappat issue concerning the status of a specially programmed computer under § 101 by finding that such systems are patentable subject matter only when they produce a "useful, concrete and tangible result." Second, although the "output" analysis imposes some problematic and unjustified restrictions on use of the format, the decision resoundingly signals the success of the machine claim format in avoiding the uncertainties of the mathematical algorithm and FWA test.
C. "Article of Manufacture" Claims in the CAFC and the PTO
The CAFC's willing abandonment of the FWA/mathematical algorithm analysis in the machine claim format context set the tone and direction for its approach to the second hardware-software combination claim format: software as article of manufacture. The article of manufacture approach defines the invention as a "software program product" consisting of a computer readable medium, called out as a diskette, ROM, DRAM or comparable piece of hardware in the detailed description, which "contains" a software program (set of instructions) designed to perform a particular task. As noted earlier, if this format holds up, it is preferable to providers of software products. It offers the same escape from the mathematical algorithm/FWA difficulties and provides better enforcement options than the machine format. The CAFC has had three opportunities to deal with the article of manufacture approach, in various degrees of directness, in In re Warmerdam, In re Lowry and In re Beauregard.
In re Warmerdam involved "a further refinement of prior art bubble systems" used as part of a collision avoidance mechanism. The basic concept involved using "nested" hierarchical bubbles enclosing the object to be avoided, with each succeeding internal level of bubbles more closely defining the exact contours of the object. Warmerdam's application included six claims with four of the claims being process claims. None of the process claims were in software article of manufacture format. The process claims set out methods for constructing the bubble hierarchy, provided a claim to a bubble system "data structure" generated by any of the methods set out in the four process claims and disclosed a claim to "a machine having a memory which contains data representing a bubble hierarchy generated by any of the four process claims."
On appeal the Board had sustained the examiner's rejections, finding as follows: the process claims "recited no more than a mathematical algorithm in the abstract and thus failed to [claim statutory subject matter]," the claimed data structure "is not within one of the categories of patentable subject matter," and the machine claim was indefinite under § 112, paragraph 2 (which requires a claim to clearly state and specify the invention) because "it left ‘unclear and unexplained how a memory is made or produced by the steps [in the method claims].'"
Addressing the method claims first, the CAFC panel noted that applying the mathematical algorithm approach raised a variety of problematic definitional issues, not the least of which is determining what precisely constitutes a "mathematical algorithm." The panel found it unnecessary to deal with the problem, instead holding that the methods were unpatentable abstract ideas: "[The methods] describe nothing more than the manipulation of basic mathematical constructs, the paradigmatic ‘abstract idea.' As the Supreme Court has made clear ‘[a]n idea of itself is not patentable,' Rubber-Tip Pencil Co. v. Howard, 87 U.S. (20 Wall.) 498, 507 (1874); taking several abstract ideas and manipulating them together adds nothing to the basic equation."
The court also upheld the rejection of the "data structure" claim, holding that "[s]ince the ‘data structure' . . . is nothing more than another way of describing the manipulation of ideas contained in the [method] claims, it suffers from the same fatal defect they do." The court specifically distinguished Warmerdam's data structure from the claims in In re Bradley on the grounds that in this case the claims did not involve an "arrangement of hardware."
Regarding the "machine" claim the court started its analysis by noting in dicta that the claim "is for a machine, and is clearly patentable subject matter." The court then reversed the Board's § 112 indefiniteness rejection. The court found "[t]he legal standard for definiteness is whether a claim reasonably apprises those of skill in the art of its scope." Applying this test to the claim, the court held that:
[t]he ideas in [the method claims] are well known mathematical constructs, and lend themselves to manipulation through known computer technology. There has been no showing that one skilled in the art would have any particular difficulty in determining whether a machine having a memory containing data representing a bubble hierarchy is or is not within the scope of [the claim].
Because the applicant did not use a software as article of manufacture claim structure, Warmerdam does not provide much direct indication of the CAFC's view of that particular format. It does, however, provide interesting insights into this panel's thought processes on two relevant issues. First, the different view of the machine claim and the data structure claim reinforces the use of hardware "environments" in software claims as an effective way to clear the § 101 requirement. Second, the definiteness discussion notes that software claims should be examined under § 112 based on the same "skill in the art" standard applied to other categories of invention.
The second case, In re Lowry, also involved a "data structure." The invention was based on an "attributive data objects" ("ADO") model which permits information about data to be extracted and described in terms of both its characteristics and its relationship with other data. Unlike Warmerdam's general data structure claim, however, Lowry's key claims were couched in the following terms: "a memory for storing data for access by an application program being executed on a data processing system" with the data arranged according to the ADO model.
The examiner rejected the claims as non-statutory subject matter and for obviousness. On appeal the PTO Board reversed the non-statutory subject matter claim, finding instead that the claims "directed to a memory containing stored information, as a whole, recited an article of manufacture" and was statutory subject matter. On the obviousness rejection, the PTO Board analogized the data contained in the memory device to printed matter and found that when "printed matter is not functionally related to the substrate, the printed matter will not distinguish the invention from the prior art in terms of patentability." The PTO Board, therefore, refused to consider the content of the memory, in particular the special organization and structure of the data according to the ADO model, and upheld the obviousness rejection over prior art which essentially disclosed "a CPU using a memory and containing stored data in a data structure."
On appeal, the CAFC panel found the PTO Board's printed matter analogy inapplicable on two grounds. First, "[t]he printed matter cases have no factual relevance where ‘the invention as defined by the claims requires that the information be processed not by the mind but by a machine, the computer'". Second, the court found that the claims were not to the specific information content of the data structure, but to "the specific electronic structural elements which impart a physical organization on the information stored in memory." The court's holding on this point is worth citing in full:
More than mere abstractions, the data structures are specific electrical or magnetic structural elements in memory. According to Lowry, the data structures provide tangible benefits: data stored in accordance with the claimed data structures are more easily accessed, stored and erased. . . . In short, Lowry's data structures are physical entities that provide increased efficiency in computer operation. They are not printed matter. The Board is not at liberty to ignore such limitations.
Having eliminated the "printed matter" underpinnings for the PTO Board's rejection, the court easily found that the specific organization of the data according to the claimed data structure was distinguishable over the cited prior art and reversed the PTO Board's obviousness finding.
Although Lowry does not specifically involve a software program claimed on computer readable media as an article of manufacture, several key points are relevant to the article of manufacture approach. First, the CAFC panel does not object to, in dicta, footnote or otherwise, the PTO Board's characterization of the claimed invention as an article of manufacture rather than a process. The case did not arise in a vacuum of first impression. Certainly the panel was aware of the effect under § 101 of the PTO Board's classification. Even though the issue was arguably not technically before the court, if the panel had been concerned about the issue it had ample opportunity to so indicate. Quite to the contrary, the panel's analysis, resting on the physical changes in the medium caused by the data structure, is entirely consistent with classification of the claimed invention as an article of manufacture. Thus, Lowry reinforces the general efficacy of including computer hardware in the claims as a method for avoiding FWA/mathematical algorithm § 101 review.
Second, the Lowry data structure receives much different treatment than the ill-fated data structure claim in Warmerdam. Citing In re Bernhart, the Lowry panel expressly indicated its willingness to find that data structures are physical entities based on the electronic or magnetic arrangement in the computer memory, even though this physical structure is "invisible to the eye." This makes it appear that the difficulty with Warmerdam's data structure claim may have been Warmerdam's failure to put it on an electronic medium. So modified, at least under the "physical structure" logic of Lowry and the dicta concerning Warmerdam's machine claim, Warmerdam's data structure claim could be transformed from the mere manipulation of "abstract ideas" into an apparatus consisting of the specific physical electronic structure of the memory reflecting the incorporated bubble structure.
Although it is not the holding in either Warmerdam or Lowry, these cases provide reasonable support for the following article of manufacture claim argument: if the claimed implementation causes changes to the electronic structure of the included computer readable medium, then the claim contains physical structure. If the claim contains physical structure, then it involves patentable subject matter, either a machine or an article of manufacture, under § 101. There is, therefore, no need to apply the mathematical algorithm based FWA test.
The final case, In re Beauregard, presented the software as article of manufacture format directly by claiming an article "composed of a computer usable medium in which a program code is embodied." The software program implemented a method for filling-in complex shapes on a computer screen. The case posed the "next step" question, if under Lowry a data structure on computer readable media is § 101 patentable subject matter as an article of manufacture, why not a software program on computer readable media?
The PTO Board of Appeals had rejected by a 4-3 vote all the claims using a variant of the printed matter analysis used in Lowry. The CAFC never heard the case. Faced with the decision in Lowry, the PTO conceded "that computer software programs embodied in a tangible medium such as floppy diskettes, are patentable subject matter under [§ 101]." The CAFC, noting the parties agreed this concession eliminated any present case or controversy, vacated the PTO Board's rejection and "remanded for further proceedings in accordance with the Commissioner's concessions."
D. The PTO Guidelines
Faced with the decisions in Alappat, Warmerdam and Lowry, the PTO had capitulated. In February 1996 the PTO issued its final Examination Guidelines for Computer-Related Inventions in which it stated it was adopting the CAFC's approach to software inventions as the guiding principles for the examination.
The Guidelines do not deal specifically with software as article of manufacture claims. They set out the following generally applicable six-step approach to examination of all software-related claims:
1. Determine what the applicant has invented and is seeking to patent.
2. Conduct a thorough search of the prior art.
3. Determine whether the invention claims patentable subject matter (§ 101 review).
4. Evaluate the application for compliance with § 112 requirements.
5. Determine whether the claimed invention is novel (§ 102) and non-obvious (§ 103).
6. Clearly communicate the findings, conclusions and their bases to the applicant.
To understand the PTO's position concerning software as article of manufacture claims requires review of steps 1, 3 and 4 of the software examination process.
Step 1 expressly states that the examiner "will no longer begin examination" by determining if a claim recites a "mathematical algorithm." Clearly, that analytical structure has been demoted under the new regime. Instead, the examiner is to "review the complete specification, including the detailed description of the invention . . . the claims and any specific utilities that have been asserted for the invention." This Step's objective is to understand the claimed invention as a whole, including applicable limitations on the claims, so the subsequent examination will be properly focused. The step clearly rejects any "point of novelty" parsing or breaking down of the claim, focusing the examiner on the invention's full scope and limitations.
The § 101 inquiry as laid out in Step 3 seeks to classify the invention either into one of the four listed § 101 statutory categories (processes, machines, manufactures or compositions of matter) or one of three non-statutory categories (natural phenomena, abstract ideas or laws of nature). The process for accomplishing this classification, reflecting the PTO's Beauregard concession on program code, contains a distinctly positive disposition toward software when claimed in article of manufacture format, generally classifying them as statutory subject matter.
To ensure that the favorable treatment afforded computer programs does not spill over to permit patentability of computer data, which quite properly remains extremely suspect under § 101, the PTO needed to distinguish between the two. The Guidelines address the problem by creating a classification scheme based on a general definitional rubric of "descriptive material." Descriptive material is divided into "functional" and "non-functional" categories. Functional descriptive material consists of "data structures and computer programs which impart functionality when encoded on a computer-readable medium." Non-functional descriptive material essentially consists of "everything else," including data.
Each type of descriptive material is assessed under the requirements of § 101. Because non- functional descriptive material (including computer readable data) is never able to impart functionality, it is always treated as non-statutory subject matter. Consistent with the concession in Beauregard, the Guidelines classify functional descriptive material, software and data structures which can impart functionality, as statutory or non-statutory depending on whether the claims include computer readable media:
[C]omputer programs claimed as computer listings per se . . . are not physical ‘things,' . . . as they are not ‘acts' being performed. Such claimed computer programs do not define any structural and functional interrelationships between the computer program and other claimed aspects of the invention which permit the computer program's functionality to be realized. In contrast, a claimed computer-readable medium encoded with a computer program defines structural and functional interrelationships between the computer program and the medium which permit the computer program's functionality to be realized, and is thus statutory.
The key is permitting implementation of the software functionality. Because claims to a computer program standing alone (a listing) do not permit the functionality to be realized, such claims are non-patentable subject matter. However, claiming the same software on computer readable media, thereby permitting its incorporated functionality to be read by the computer and implemented, makes the combination statutory.
Although the Guidelines expressly confirm that software presented in an article of manufacture form is generally statutory, the last step in the § 101 analysis revisits the "smuggling in" concerns and carves out process claims masquerading as apparatus claims. Only if the claims detail a specific machine or manufacture should they be viewed as true apparatus claims and expressly statutory. If the claims are written to cover "any and every" machine or manufacture they should instead be examined as de facto claims to the underlying process steps requiring further analysis to determine whether that process involves statutory subject matter.
Despite considerable effort at explaining the difference between claims to a "specific machine or manufacture," and claims to "any and every machine or manufacture" the distinction is not entirely clear. The basic requirement for finding that the former is present appears to be that the claims define the "physical structure of the machine or manufacture in terms of its hardware or hardware and ‘specific software.'" Because an article of manufacture claim will normally include every form of computer readable media, the primary question becomes whether the claims provide enough specifics concerning the operation of the "carried" software. Based on the example provided in the Guidelines, claims which include sufficient detail (either in actual program code or logic circuits) fall on the "specific manufacture" side and those which leave implementation primarily to "those of ordinary skill in the art" on the "process" side. The contest between applicant and examiner will come down to determining the minimum amount of specific detail required to satisfy the "specific manufacture" requirement.
Finally, step 4 of the Guidelines takes a fairly traditional approach to the § 112 requirements and will generally be easy to satisfy. Of interest are two references to the discussion which emphasize that any software relied upon to meet the § 112 requirements should also conform to the § 101 article of manufacture requirements. First, in the case of "means for" style claims if "the applicant discloses only the functions to be performed and provides no express, implied or inherent disclosure of hardware or a combination of hardware and software that performs the functions," a rejection under § 112, paragraph 2 is appropriate for failure to particularly point out and distinctly claim the invention. Reflecting the § 101 analysis, the Guidelines indicate that if software is to serve as the disclosed means, the software must be in article of manufacture/computer readable media form. In addition, to avoid such a rejection, there must be sufficient detail regarding the software to permit one of skill in the art to determine the scope of the invention claimed.
Second, for purposes of enablement, the examiner is instructed to "ensure that along with the functional block diagram, the disclosure provides information that adequately describes each ‘element' in hardware or hardware and its associated software and how such elements are interrelated." The "hardware and its associated software" again clearly emphasizes the article of manufacture claim structure. Regarding the description of the incorporated software, the applicant can elect either to expressly provide more detail or fall back on the "person of ordinary skill in the art."
The practical experience to date under the Guidelines indicates the PTO meant what it said. Examiners appear to be taking a literal and substantially more liberal view of the patentability of software related inventions claimed in article of manufacture form. Not surprisingly, the effect has been to shift applicants to use of this more favorable and less problematic format for claiming software-related inventions.
E. What's Wrong With Current Article of Manufacture Practice? What We Are Trying to Achieve With the Patent System.
The new hardware-focused regime under the Guidelines provides a number of administrative benefits compared to the prior mathematical algorithm approach, even in its Freeman-Walter-Abele incarnation. From the applicant's perspective, the framework provides significant improvements in terms of certainty and related increased likelihood of a patent grant. The traditional jousting with the examiner concerning the scope of coverage, particularly regarding the degree of precision required in detailing specific software elements will, of course, continue. However, by following the formula set out in the Guidelines, an experienced patent practitioner ultimately should have little difficulty in obtaining a patent with acceptable claims for a computer program using the computer readable media/article of manufacture format. From the examiner's perspective, the system's simplicity and quasi-objectiveness makes case examination more focused and, consequently, a more efficient, faster process. Finally, patents issued in the hardware formats, particularly the article of manufacture variant, permit direct enforcement against competing software distributors. Both software patentee difficulties under the former FWA/process approach appear to have been resolved.
Although efficient administration and enforcement are important, the primary objectives of the patent laws are not achieved merely by ensuring ease and speed of prosecution, predictability of results or maximizing enforcement options. To take the true measure of current practice requires an exploration of its consistency with the policy reasons for having a patent law at all.
As discussed in Part A above, the United States patent laws rest on the premise, embodied in the Constitution, that providing an economic incentive to inventors in the form of a right to exclude others for a limited term will lead to more inventions which, in turn, will lead to improvement in the general social condition. Accepting for purposes of this discussion that the economic approach actually operates to achieve the desired objective of increased invention, then the primary social engineering issue is to determine which inventive activities, if any, should be targeted by the patent incentive.
This determination, in turn, requires identification of those situations where granting patent protection ultimately provides social prosperity returns in excess of costs arising from the resulting restrictions on competition. This was the very issue addressed by Jefferson and Madison when debating whether to include the Patent Clause in the Constitution in the first instance. Thomas Jefferson eloquently captured the essence of the inquiry: "[D]rawing a line between things which are worth to the public the embarrassment of an exclusive patent and those which are not." The present inquiry into the patentability of software as article of manufacture, therefore, requires a determination of whether this claim format falls on the "worth it" side when tested under Jefferson's line-drawing exercise.
The Framers' decision concerning the incentive-embarrassment tension provides a good starting point. Their resolution is clearly reflected in the Patent Clause which expressly limits congressional use of the patent power "to promote the progress of . . . [the] useful arts." This language establishes the basic policy (and legal) outer boundary condition concerning the software as article of manufacture question. If under the Constitutional regime the Congressional patent power only extends to promotion of progress in the useful arts, the answer to when (if ever) software constitutes patentable subject matter is tied to when software claims should be viewed as falling within the targeted useful arts. That determination requires not only analysis of the reach of term standing alone, but because the exercise of the patent power is discretionary with Congress, whether that body has imposed additional limitations.
1. The Useful Arts
The term "useful arts" is not defined anywhere in the Constitution. However, the extensive literature contrasting the "sciences" with the "useful arts" for purposes of determining the respective scope of the copyright and patent laws under the Constitutional provision sheds some light. That discussion indicates that protection of expression ("writings") of general knowledge or learning ("science" in 18th century terms) was placed exclusively under the copyright laws, and protection of applied science ("useful arts") exclusively under the patent laws. Under this reading, the useful arts, and therefore the maximum Constitutional reach of United States patent law, are limited to the practical application of knowledge and learning to the industrial arts, that is to the creation of specific tools, products and procedures for performing activities such as the creation of new things, arrangements of things or relationships between things. The courts have consistently applied this interpretation, with the CAFC expressly equating the Constitutional "useful arts" with the present day term "technological arts."
Although this basic definition provides general guidance, it requires refinement for specific application. Much has been made in the patentable subject matter cases over the past 15 years of the basic Congressional premise cited by the Supreme Court in Diamond v Chakrabarty that the Patent Act is meant to "include anything under the sun that is made by man." If this were truly the case, the task would be a simple one: everything humans create are "useful arts" and should be patentable subject matter. Such a sweeping statement is clearly at odds with the above interpretation of the Patent Clause itself, with only the more limited class of technological activities falling within the useful arts. In fact, although the Court in Chakrabarty clearly instructed the lower courts to adhere to the guiding spirit of this pronouncement, it did go on to explain that there are a variety of human activities that do not merit the embarrassment of a patent on policy grounds. As always, to include some things is to exclude others. The difficulty comes in determining what is in and what is out. The core normative task in the software as patentable subject matter debate is determining on which side of the divide each particular software invention falls.
Judicial attempts to refine the "useful arts means technology" definition for application generally reflect one of two approaches: exclusion of entire classes of non-technological activities and exclusion of inventions deemed too inchoate to qualify as technology. Both have been largely unsuccessful in the software arena, but the reasons for each failure provide important guidance in making additional efforts.
The first approach seeks to identify and exclude software inventions when they implicate activities which fall outside the technological arts. The approach lead to the development of doctrines such as the unpatentability of "methods of doing business" and "mental steps." Although at the conceptual level such an approach makes sense, it has been unworkable in practice. The fundamental difficulty is that even when the target application for a software invention is clearly a non- technological activity, it does not follow that the computer tool itself falls outside the technological arts. Determining when a particular software claim is to the unpatentable activity or to the computer tool, using labels such as "method of doing business" or "mental steps," without more, has proven problematic at best. Even if the specific results have been less than satisfactory, keeping the important "activity versus tool" difference in mind is helpful to full resolution of the questions surrounding the patentability of computer software.
The second approach focuses on claims which are too inchoate to pass muster as technological application. They include the well-known doctrines of the unpatentability of laws of nature, natural phenomena and abstract ideas. This approach gave rise to the failed "mathematical algorithm" test in the software arena.
All three legs of the underlying framework pre-date the computer software patentability debate by many years. The first two legs, denying patents for claims to laws of nature and physical phenomena standing alone, arose in connection with much earlier technology inventive forays, including the telegraph, the telephone, radio antennas and pre-recombinant DNA "biological combination" efforts. The rationale for these exclusions has been varyingly stated as: these matters are "part of the storehouse of knowledge of all men," they are discovered not invented and "they are the basic building tools of scientific and technological work." These articulations reflect an underlying policy concern that permitting preemption of naturally occurring forces or relationships through patent law, thereby prohibiting their use in the entire range of potential technological applications, will severely constrain progress in the useful arts, rather than promote it. Consequently, these claims do not merit the "embarrassment of an exclusive patent" and the subject matter is left "free to all men and reserved exclusively to none."
The third leg of the exclusion, abstract ideas, also has venerable pre-computer lineage. It arose, however, in connection with a low-technology case involving the attachment of an eraser to the end of a pencil. In Rubber-Tip Pencil v. Howard, the Supreme Court found the claims sought merely to preempt the general concept without any "inventive" application. Stating that "[a]n idea of itself is not patentable" the Court held the patent invalid. In essence, a "great thought," even if its ultimate application falls within the technological arts once implementation issues are resolved, does not sufficiently further the specific patent law objective of technological advance. Quite the contrary, prematurely granting a patent in such cases severely limits inventive behavior aimed at achieving practical technological implementation of the idea. Abstract ideas, without more, are not yet useful enough arts to merit a patent.
Not surprisingly, the touchstone for patentability of inventions which raise concerns in all three of these areas ties back to the Constitutional useful arts requirement. Although laws of nature and natural phenomena are not themselves patentable, "a novel and useful structure created with the aid of knowledge of scientific truth may be." Likewise, in the realm of abstract ideas, although "[a]n idea is not patentable, . . . a new device by which it may be made practically useful is." These tests reflect the patent law's focus on technology. Entitlement to a patent requires that the details for one or more specific implementations of the law of nature, natural phenomena or general idea be sufficiently identified and fleshed out before it addresses patentable subject matter. The patent is then issued on and restricted to that specific implementation.
Unfortunately, as experience with the mathematical algorithm permutation of the "law of nature/abstract idea" test has demonstrated, something other than the mathematical algorithm test is required in the software context. The rationale giving rise to the concepts, however, provides useful policy guidance. The same underlying concerns driving a desire to avoid patents on general statements of laws of nature and abstract ideas apply equally in the context of software claims. A workable test must consistently separate software claims to matters falling within the too technologically inchoate classes of laws of nature, natural phenomena and abstract ideas from those which involve their application, the useful arts.
2. Congressional Action
As noted above, because the grant of the patent power is not mandatory, merely recognizing that the permissible scope of congressional action is limited to the technological arts (poorly defined as that may be) does not end the policy inquiry. Congress retains the power to include or exclude any or all specific "useful arts" inventions based on its case-by-case evaluation. It is, therefore, important to determine if Congress has made specific determinations which might affect the assessment of software as useful arts patentable subject matter.
The Congressional view on the question is expressed in § 101. The Supreme Court had the opportunity to interpret this provision and, in particular, the term "manufacture" which is especially relevant to the present inquiry, in Chakrabarty. Noting the now frequently quoted language from the Congressional Report accompanying the 1952 Patent Act indicating that statutory subject matter should "include anything under the sun that is made by man," the Court found that "[t]he subject matter provisions of the patent law have been cast in broad terms to fulfill the constitutional and statutory goal of promoting ‘the Progress of Science and the useful Arts' with all that means for the social and economic benefits envisioned by Jefferson." The Court went on to admonish against the addition of any limitations not expressly stated under § 101; that is beyond requiring the invention be a machine, article of manufacture, process or composition of matter.
Of additional relevance are the Court's observations concerning inventions "in areas not contemplated by Congress when the patent laws were enacted." Addressing the concern over the potentially adverse consequences of recombinant-DNA technologies to society, the Court noted that "[a] rule that unanticipated inventions are without protection would conflict with the core concept of the patent law that anticipation undermines patentability" and "that Congress employed broad general language in drafting § 101 precisely because such inventions are often unforeseeable."
In Chakrabarty the Court found that beyond the express language of § 101 itself (which should be broadly read even in the face of the unanticipated) Congress has done almost nothing on the exclusionary front. What the Court has to say about the congressional intent underlying the Patent Act is of binding effect. If Congress wishes to correct the Court's interpretation, it is required to act. Congress has infrequently responded to arguments in favor of excluding particular technologies from coverage under the patent laws, choosing instead to regulate risks on application. In the case of patent coverage of computer software Congress has consistently taken no action, leaving the matter entirely to evolving court interpretation in light of the Constitutional useful arts mandate.
The above demonstrates that the primary policy objective, as well as the primary limitation guiding the patentable subject matter inquiry, remains the Constitutional purpose of progress of useful arts as that concept has been illuminated by previous efforts of the courts and the commentators. What remains is to determine whether current software as an article of manufacture practice is consistent with this policy objective and if it is not, as this article argues is the case, why not and how appropriate adjustments can be made.
3. Can the Current "Software as Article of Manufacture" Practice Get Us There?
a. Bounding the Inquiry
To profitably pursue the question of whether software claimed in article of manufacture form should be viewed as within "technological arts," it is useful first to bound the target of the inquiry. This requires a brief discussion of the scope of the article of manufacture form of claim under consideration and the nature of software.
b. Scope of the Claim Format
First, clearly "pure" article of manufacture claims covering only inventions relating to the hardware carrier itself, e.g., the structure or composition of the diskette, ROM, CD or the like, are not controversial. There can be little debate that content free claims to the computer readable media alone constitute § 101 articles of manufacture. These claims are of interest, however, as they provide an intuitive, although incorrect, basis for the software as article of manufacture argument, because the hardware carrier standing alone is an article of manufacture, perhaps such a carrier plus its software content should also be considered an integrated article of manufacture.
Second, article of manufacture claims involving data content are outside the discussion. Although both data and software program content are (or can be) machine readable, they serve very different functions and raise very different patentability issues. Computer software consists of a set of instructions that cause a computer system to perform a particular sequence of operations directed toward accomplishing the particular task defined by the program. Data is the informational raw material on which the programmed computer system operates. Although the data affects the specific outcome of the computer operation in terms of result and possibly the order in which particular operations are performed, it does not create the operational framework for the implementing computer system. Although ambiguous "data content" article of manufacture claims can raise § 101 issues, they are readily resolvable against their patentability during the examination process. This article, therefore, focuses only on the more problematic software content article of manufacture claims.
c. Understanding Software: Technology, Language or Both?
As noted above, software consists of a set of instructions directing the operation of a computer system. When the software instructions are executed by the computer system, the system performs the specific task defined by the software program. It has been frequently and properly observed that software operates algorithmically. The program's instruction set is based on and reflects a particular methodology for accomplishing the defined task. A person can review the software code and understand this incorporated algorithm. Consequently, software extends beyond mere computer system instruction. It can also serve as a means to express and communicate the incorporated methodology.
The two functions of software produce very different outcomes to the software as patentable subject matter inquiry when passed through the patent policy test developed above. When used to communicate the contained algorithm to a person, software serves precisely the same function as any other language or means of expression. It involves the symbolic representation of the incorporated methodology in a form that permits its interpretation by human readers. There is nothing, therefore, which makes couching claims in software terms for this purpose technological or deserving of patent protection. In contrast, when software acts as computer system instruction the software serves as the actual means for implementing that same methodology, causing and directing the operation of the computer system's performance of the specific task. This function is technological and comes squarely within the useful arts targeted by the patent laws. The objective, therefore, is to find a reliable way to consistently separate the former use of software from the latter.
In "real world" implementations, software programs can be expressed in a variety of computer programming languages. As the phrase "software code" implies, these computer languages vary in their suitability as vehicles for human communication or for computer instruction. For example, a program written in the COBOL programming language has a readily ascertainable connection to human language expression. An assembly language version of the same computer program is substantially more cryptic and will communicate very little to many readers. And the program expressed in machine code, consisting of the string of 1s and 0s commonly associated with computing, will communicate virtually nothing to anyone but a computer expert (and then only with considerable effort). Similarly, a computer system will require varying degrees of "translation" of each of these languages to effect the actual electronic operation of the hardware, more for higher level languages like COBOL, less for machine code. Generally stated, the less comprehensible software is to a person the more directly useable it is by a computer. Consequently, there is at least a superficial appearance of separation between the communicative and implementation uses of software based on the level of human readability versus computer usability of the particular software expression.
In the context of patent claims, however, even this glimmer of distinction frequently disappears and the computer instruction and human communication functionality of software become completely intertwined. Patent applicants frequently eschew direct claims to specific software program code as irrelevant to the invention. Instead they frame claims as a series of "means for" operational steps describing the incorporated methodology of the program instructions leaving any listing of particular program code exclusively to the detailed description in the specification. The result is an apparently complete merger of the two functions of software: the claims simultaneously serve as a means for expressing the contained algorithm to the reader in normal, if specialized patent claim, English language, and simultaneously obtain patent coverage for the actual technological means for instructing the computer system's operation.
The ability to properly determine when software inventions should be classified as useful arts technology depends on the ability to reconstruct a dividing line between software as a language (not technology) and software as computer implementation (technology) in the patent claims context. A proper test for patentability of software related inventions must clearly and consistently draw a line separating claims to software as the specific means for computer system implementation of the contained algorithms/processes (which are patentable subject matter) from those using a software context merely to express and communicate those algorithms/processes (which must be tested on their own merit independently of the software context to determine if they involve patentable subject matter).
F. Software Under the Guidelines
Having identified the dual functionality of software and the need to distinguish these functions in the context of patent claims, we can turn to an assessment of the viability of the current software as article of manufacture practice set out in the Guidelines. As will be demonstrated, it is not up to the task.
It is important to recall that the Guidelines are based on the Commissioner's concession described in Beauregard that computer software on computer readable media is patentable subject matter. The PTO Board majority opinion in that case provides important insight concerning the thinking behind the PTO's concession and the resulting positions taken in the Guidelines. The PTO Board majority determined the claims were to an article of manufacture consisting of the computer readable medium and the carried software program. Applying a printed matter test and finding no functional relationship between the printed matter (the computer program) and the substrate (the hardware carrier), the PTO Board sustained the examiner's rejection under § 101. Clearly prior to the CAFC dismissing the "printed matter" analysis in Lowry, the PTO Board majority did not believe that the presence of the computer readable medium added anything which made the claims to the software itself patentable subject matter. In fact, under the PTO Board majority's printed matter analysis, including the hardware actually clarified the case against patentability of the software as the applicant's use of the computer readable medium to obtain article of manufacture classification specifically prohibited any reliance on the software content.
Interestingly, the PTO Board majority opinion contains language reflecting, although not completely articulating, the need to distinguish between the implementation and language roles of software. Rejecting an argument that the software and diskette combination formed an article of manufacture machine component, the majority stated:
We do not agree that because appellant's article of manufacture is intended for use in a computer, or because its primary purpose is to control a machine, it is somehow transformed into a machine component. Similarly, we do not agree that since this program is intended to be read only by a computer, not a human, it is no longer printed matter but is somehow beyond printed matter. In the first place, appellants' list of instructions can be read by a human as fully discussed above [referring to the examiner's argument that "computer readable media" includes a sheet of paper]. Secondly, none of the claims at issue recites a machine control. To the contrary, the claims are directed to a list of instructions which have only the potential to control a machine. . . . . Under the present state of the law, printed matter is printed matter no matter in what language it is printed, and is not somehow transformed into something else because it can also be read and used by a computer.
And further that:
There is only a prospective functional relationship which will occur only when the printed substrate (the diskette with the program) is entered into the machine. This functional interaction is what the claims of the recently issued [process] patent define, but such a definition is not present in these [article of manufacture] claims.
The majority, therefore, all but expressly acknowledged the view expressed in Alappat that software, when introduced into a computer system, makes the general purpose hardware into a special purpose computer instructed by that software and that this functionality is technology. Their quite proper concern was that, as claimed by Beauregard, the realization of that useful arts software functionality was only "potential" rather than required. As a result the structure of Beauregard's claims failed to move beyond the mere expression of the same algorithm already covered by the corresponding process patents. The majority had walked up to, but failed to specifically recognize, that they were addressing the important division between software as language and software as implementation.
The Commissioner's concession "that computer software programs embodied in a tangible medium such as floppy diskettes, are patentable subject matter under [§ 101]," however, completely failed to address the fundamental language versus implementation concerns raised by the PTO Board majority in Beauregard. Instead it placed full reliance on the presence of the "tangible medium" as the indicator of patentable subject matter.
In doing so, the Commissioner's concession over-reads Warmerdam and Lowry. As discussed above, the two cases superficially appear to reflect the need for physical media in the claims, precisely as called for in the Commissioner's concession. What actually explains the divergent outcomes, however, is a more fundamental difference in the two claim formats when analyzed against the language-implementation dichotomy. In Lowry the "potential" for machine control is transformed exclusively into a software as implementation claim, by the claim preamble stating the invention is "for access by an application program being executed on a data processing system." In contrast, Warmerdam's claim to "a data structure generated by the method of Claims 1 through 4" leaves any possible machine control merely as "potential," there being no requirement that the invention actually be used as in a computer implementation. The different outcomes in the cases, therefore, actually reflect the fact that when analyzed as a whole the particular claims fall on different sides of the language versus implementation dichotomy, not a requirement that tangible media be present in the claims as reflected in the Commissioner's concession.
As a result of this over-reading, the PTO position reflected in the Guidelines calls for examination of software claims based on the following two inter-related and faulty syllogisms premised on the importance of the presence or absence of computer readable media:
1. Syllogism One
To be a § 101 article of manufacture (patentable subject matter) the functionality of the software must be electronically implementable by a computer. Claims to software on computer readable media permit implementation of the software functionality. Therefore, claims to software on media are patentable subject matter as articles of manufacture.
2. Syllogism Two
To be a § 101 article of manufacture (patentable subject matter) the functionality of the software must be electronically implementable by a computer. Claims to software that do not include computer readable media do not permit implementation of the software functionality. Therefore, software claims which do not include computer readable media are not patentable subject matter as articles of manufacture.
Mercifully, the Guidelines clearly shift focus away from the use of mathematical algorithm forms of analysis, including the related Freeman-Walter-Abele two-step test. The computer readable medium test, however, fails to draw an appropriate line between software used as language and software claimed as implementation. As a result, the Guideline syllogisms are at once over- inclusive (the first syllogism) and under-inclusive (the second) from a policy perspective. An analysis of each syllogism demonstrates the difficulties.
3. Syllogism One (Over-inclusion)
The vice of over-inclusion is the possible application of patent incentive to inventions which as a policy matter we do not wish to encourage. Under the present Constitutional regime this includes all fields of human endeavor that do not involve technology: subject matter falling within the fine arts, business, and the like, as well as the exclusion of laws of nature and abstract ideas as technologically inchoate. If using software as language can serve as a vehicle for expressing these concepts and at the same time serve as the basis for obtaining § 101 approval, there exists a serious potential for preemption beyond the intended scope of the patent laws.
The Guidelines' reliance on the inclusion of computer readable medium is an unreliable talisman that the claims contain the necessary technology focus. The physicality of claims to software on a computer readable media is appealing. The combination has the look and feel of an article of manufacture. It can be inserted physically into and read by a general purpose digital computer. However, to rest the logic of the software as article of manufacture approach on the physicality or computer readability imparted by the carrier medium leads to improper results.
There can be no disagreement that merely putting software on a diskette does not magically transform the software expressed functionality into technological apparatus covered by § 101. This has no more effect than writing an idea in a book. We would never confuse the patentability of a book, as a book, with the patentability of the ideas expressed if the language used were English. In the same way writing an idea into a book does not make the expressed ideas an article of manufacture, writing software onto computer readable media does not transform the combination of the media and the contained ideas into a unitary article of manufacture.
The Guidelines do not, however, predicate the § 101 determination solely on the physical writing of the software on the tangible medium. Addressing the PTO Board majority concern in Beauregard that software, as such, only has "the potential to control a machine," the Guidelines state the presence of the computer readable medium "permit[s] the computer program's functionality to be realized." The Guidelines assert this makes the software claims statutory.
The essential requirement of distinguishing between software used as language and software claimed as implementation, however, is to ensure we treat only the latter, technological application of the software functionality as § 101 apparatus. The former software as language claims must be separately assessed to determine whether what is being expressed is itself statutory or non- statutory subject matter. The requirement that the software be written onto media which can be put into and read by a computer does not draw this line. This requirement only means that the software functionality can be entered into the machine. It says nothing at all about whether it must be. The requirement, therefore, precludes neither coverage of other uses the software in the claims nor the resulting preemption of the making, use or sale of the expressed functionality beyond the claimed computer system implementation.
Two examples demonstrate the problem. First, as the examiner and PTO Board majority in Beauregard noted, broadly drafted computer readable media claims cover software written on a piece of paper. There is no difference in technical terms between such a claim and software written on a diskette. They are both machine readable; the former with the aid of an appropriate scanner and related software, the later using a disk drive and related software. In either case, after a computer reads in the program instructions, to actually "execute" the program, it must be loaded into "active" memory. Thus, the diskette carried software is no more directly machine implementable than the paper carried version. The difficulty, of course, is that if a "paper inclusive" media claim contains no limitation to computer system implementation of the software functionality, it would be infringed by anyone making, using or selling a description of the software's functionality (its algorithm) written on paper. This would be true whether or not the paper version was intended for execution by a computer system.
Second, consider creation of a computer readable file describing the software functionality for the purpose of forwarding it electronically (by e-mail, for instance) to pass those ideas on to someone for their review and comment. Once again, a media claim to the software functionality would be infringed by the act of creation and use of this computer readable file even absent any actual use of the expressed functionality as software instruction in a computer system implementation.
Two counter arguments can be offered in an attempt to demonstrate that these examples do not pose real difficulties. First, the computer readable media formulation required by the Guidelines strongly implies the intended use of the software functionality is limited to computer systems. A reviewing court could simply invoke the "reverse equivalents" doctrine, draw its own inferential limitation to the claims, and hold that use of the software expressed functionality outside a computer application did not infringe. Second, because infringement only occurs by documenting the expressed functionality on "machine readable media," anyone wishing to use that functionality is free to do so provided they ensure it is never written down in tangible form.
Neither argument is persuasive. In the former situation a potential infringer must rely on the uncertainty of court interpretation: a slim reed to rely on in face of a potential infringement lawsuit. In the latter situation the usefulness of the undocumented approach in the real world is extremely questionable given the problems associated with the introduction, implementation, and particularly the on-going use of even fully documented procedures. Nor is it appropriate given the growing pervasiveness of electronic forms of communication. There seems little valid reason to force third parties to rely on potential court created exceptions or careful use restrictions to carve back grants of overly broad patent claims. Far better to draw an express limitation which properly focuses directly on what is to be accomplished: separating use of software as technology from use of software as language.
Unfortunately, the presence or absence of machine readable hardware permits claim coverage to go beyond the technological software as implementation to more general coverage of the underlying functionality expressed. The result is precisely what the Guidelines were trying to avoid. An idea or process, which may or may not be patentable subject matter, becomes classified as a statutory article of manufacture per se simply by using computer software language to describe it and putting that description on computer readable media.
The Guidelines, in fact, do recognize this possible "smuggling in" difficulty with the media format. The examiner is expressly instructed to examine claims which are not limited to specific hardware or hardware and software combinations (covering instead use on "any and every" hardware configuration) based on the underlying process (the functionality expressed). This desire to distinguish between process and apparatus claims is driven, not surprisingly, by the same motivation driving the need to distinguish software as language from software as implementation: ensuring that general claims to process (software as language expressing functionality) are not masquerading as apparatus (software as technological implementation). The distinction the Guidelines attempt to draw is appropriate. The difficulty, however, is that the "any and every" approach misses the mark because it is looking in the wrong place for the distinction. The determination does not rest on whether the claims cover implementation of the claimed software functionality on all (any and every) hardware platform or only on specific computer hardware as the Guidelines would posit it. It rests on whether the claims restrict the software functionality to implementation as part of a computer system, whether a specific system or generally. The result is that the "any and every" formulation overstates its case and may improperly preclude deserving software as implementation claims from receiving apparatus treatment under § 101.
4. Syllogism Two (Under-inclusion)
The Guidelines also require, as reflected in the second syllogism, that software claims include computer readable media to be classified as articles of manufacture. This position rests on the premise that because computer readable media is sufficient, or at least necessary, to permit the computer program's functionality to be realized (thereby demonstrating the useful arts nature of the claim), the absence of computer readable media should reliably identify claims which do not. The truth is otherwise. A brief look at the basics of the underlying computer technology demonstrates why.
In very basic computer science terms a special purpose computer system combines specific software instruction with appropriate hardware. Performing the combination and implementation of the software functionality requires a number of steps. Taking a simplified case as an example (there are many permutations), the following might occur. First, the software program must be loaded into appropriate hardware. This can be done by putting the software program on a computer readable medium. Second, the software is transferred from the medium by reading it (using appropriate hardware and software) into system storage, either a hard disk or directly into main memory. Finally, the program is "run" by moving the software instructions from storage into the operating registers (another form of computer memory) and executing them, all in accordance with the system's operating algorithm.
Viewed against this backdrop, the presence of computer readable media in the claims is neither sufficient nor necessary to computer system implementation of the software functionality. Putting the software on a diskette or other computer readable media can be a part of the overall implementation process. However, it is not sufficient as the entire process, not merely the entry step, is required to realize the software's functionality in the sense that the special purpose system actually performs the specific instructed task. Nor is there anything necessary about putting program code on a separate computer readable medium. The same computer instructions could be typed or read in orally one at a time or hardwired with the same effect.
The mere absence of computer readable media in a software claim tells us too little about whether the claim reaches beyond computer system implementation of the software's functionality. Syllogism two, therefore, draws a too restrictive, under-inclusive policy line. The result is a failure to apply the patent incentive to the full range of activities we wish to encourage.
The Guideline computer readable media requirement, therefore, misses the patentable subject matter target on all counts. As discussed in connection with syllogism one, its presence does nothing to ensure the claimed functionality is covered only in its technological role as instruction for a computer system implementation. As just discussed, its absence does not mean the claims go beyond such coverage.
Because the requirement fails to achieve any useful purpose, it only adds irrelevancies to the software claiming process. The most obvious technical effect is to create a groundless rejection under § 101. One could argue this is of no practical significance. By proper drafting, the computer readable media requirement can easily be met.
However, forcing software claims into this unjustified format has two less obvious but far more serious ramifications. First, the technological advance in software implementation inventions is in the functionality imparted to the related special purpose computer system. Protecting this functionality solely when carried on a computer readable medium unduly limits the scope of the protection provided and, consequently, the incentive afforded to these inventive efforts. In particular, the requirement limits infringement only to situations when the competitor makes, uses or sells the software in combination with a machine readable medium. This has undesirable consequences as variations occur in the way software is recorded and delivered. Even if an expansive reading of computer readable media is accorded the claims, the need to determine whether the competitor's medium is included provides an uncertainty to enforcement, uncertainty breeds expense and expense reduces incentive.
Although this may seem of only minor concern given current technology primarily deals with software on relatively standard forms of computer readable media, there are already indications that this will not continue to be the case. Consider the situation in which the only copy a customer receives is down-loaded electronically from an Internet server located outside the United States. Does the provider infringe a United States media limited patent claim or is the patentee's only recourse against the consumer who creates and uses the new computer readable media version on receipt? Or, as human speech interfaces become more prevalent, does a media limited claim protect the functionality of software programs entered orally by a user? In both situations given the provider/distributor's activity is likely to be significantly more damaging to the operation of the patent incentive, and providing a direct remedy is certainly more efficient as well as commercially attractive, it seems inappropriate to restrict enforcement alternatives by what is an otherwise irrelevant hardware carrier limitation.
Second, forcing software claims into a "hardware" based mold in order to clear § 101 operates against the evolution of the technology itself. Computer technology continues to move away from hardware implementations. A large portion of all new development is done exclusively in software. As improvements are made, new versions or replacement products substitute software implementations for existing hardware (including hard-wired) solutions. Additionally, the software solutions themselves are becoming increasingly independent of their hardware operating platforms. Many software implementations are designed expressly to be transportable and inter-operable with as large a range of hardware alternatives as possible. Basing software as article of manufacture approach on the presence of hardware (even computer readable medium), therefore, forces computer technology innovators to switch paradigms when it comes to obtaining patent protection for the results of their inventive efforts. The effect is to retard the advance of computer technology rather than to encourage it.
G. A New Test and a New Syllogism
The foregoing analysis demonstrates there are serious shortcomings with where the Guidelines take us with regard to software as article of manufacture claims. The presence of machine readable media is neither a necessary nor a sufficient condition to article of manufacture apparatus treatment of software inventions. In fact, the hardware requirement in the Guidelines adds nothing to the patentable subject matter analysis but irrelevancies and distractions. It should be abandoned and a new paradigm adopted.
This same discussion also indicates the overall patentable subject matter inquiry should be refocused. The question is not "if software is patentable," but whether particular uses of software should be considered patentable subject matter in light of the incentive to the useful arts policy objectives of the patent laws. The general answer to this question is straight- forward: when software functions as technology it should be, when it does not, it should not. To draw this line, the patentable subject matter test must separate claims to software expressed functionality as instructions to a computer system from claims which use software as a language to express the described functionality. Only the software implementation claims fall directly within the useful arts. The software as language claims may or may not involve patentable subject matter depending on the nature of the particular underlying, software independent, functionality expressed. What is required is a test which reliably differentiates between these two roles of software in specific claim contexts.
Such a test can be simply drawn. Placing the dividing line between claims which limit coverage of software functionality to use in a computer implementation and those that do not provides a workable basis for making the "useful arts" patentable subject matter determination. When claims specifically limit coverage of the software expressed functionality to use as the instructional component of a special purpose computer system they only cover software expressed functionality in its technological implementation role. If the claims are not limited to the computer implementation role they move beyond software as technology into the realm of software as communication device. In these situations it is not the method of expression which carries the useful arts inquiry, but the nature of the functionality being expressed.
That this dividing line holds up for normative purposes can be demonstrated by analyzing the results achieved by separating these two types of software claims. First, straight-forward analytic progression shows that claims covering software only as computer system implementation are limited to matter falling within the technological arts. Claims covering general purpose computer hardware are clearly machines within the useful arts and the § 101 patentable subject matter categories. Claiming a computer system consisting of a combination of such a general purpose computer hardware plus specific software is tantamount to claiming a special purpose computer. There simply is no logical reason for finding that the inclusion of software instruction in an otherwise patentable general purpose computer system to permit performance of a specific task makes the combination special purpose computer non-patentable subject matter. Such a special purpose computer is also a machine within the technological arts and § 101.
Once one accepts the special purpose computer system analysis, it is a short and unavoidable step to finding that claims to the software itself, when limited to its "instruction" role as a part of the special purpose computer system, constitute both technology and apparatus (machine or articles of manufacture). The patentable special purpose computer system can be viewed as comprising two separate components: the computer hardware and the software as the implementation "instructions" for operating that hardware. This is entirely consistent with our view of other machines consisting of various components. Just as we unhesitatingly treat each component of other types of machines as a separately patentable article of manufacture, there is no reason not to do so in the case of computer systems. Clearly each of the general purpose hardware "components" of such a system can stand alone as patentable subject matter. And when the software component of the system is claimed only in that component role (by expressly restricting coverage in the claims to use as computer system instruction), it too should be treated as separately patentable subject matter.
In contrast, claims which do not clearly limit coverage of software expressed functionality to its implementation function merely employ a software context as the vehicle for expression of the underlying functionality. The particular choice of software as the vehicle to convey that information to others adds nothing more to the statutory subject matter analysis than would expressing the same ideas in English, Italian or Japanese. Nor does stating in the detailed description that the functionality being expressed in the claims may, even in the preferred embodiment, be written in software code and used in a computer, exclude coverage of other non- computer system uses of that functionality by the claim. In these situations it is the expressed functionality itself which must be assessed for patentability, not the particular form of its expression.
The above approach can be captured by replacing the Guideline's hardware focus with a new software "as such" article of manufacture test based on claims which only cover the software functionality when used as a computer system component. This suggests the following new syllogism: machine components are patentable subject matter (articles of manufacture). Claims which cover software functionality only when used as a component of a computer system implementing the task instructed by the software, are claims to machine components. Therefore, such software claims are claims to articles of manufacture and are patentable subject matter as such.
This test does not preclude an applicant from making broader claims couched in software language terms. The objective of the test is only to ensure that it is clear how the software is actually being used in a particular claim. Specifically, the test seeks to avoid confusing broad claims to the underlying ideas expressed in software language terms with software as technology claims resulting in inappropriately superficial review of those broader claims under § 101. If an inventor believes she is entitled to coverage of the expressed functionality beyond the software implementation of her invention in a computer system, she should seek it. But in those instances, she will not be permitted to rely on the software expression of the functionality for purposes of the § 101 inquiry.
1. Does it work? Testing the New Syllogism for Consistency With Patent Law Objectives and Practicality
Having proposed a substantially "bright line" test for determining if particular software claims should be treated as patentable subject matter, it is necessary to evaluate its efficacy. This requires the test work on three levels. First and foremost, its application must only identify as patentable subject matter inventions which are consistent with the objective of the patent laws: incentivizing exclusively those activities which fall within the technological arts. Second, it should not impair, and if possible should enhance, application of the additional requirements for patentability, in particular the "progress" objectives embodied in the requirements of novelty and nonobviousness. Third, it must work in practical terms, permitting efficient and effective drafting and examination of patent applications and appropriately flexible enforcement options.
a. Patentable Subject Matter
The first criterion requires the test clearly identify and separate software claims which involve subject matter falling within the technological arts from those that do not. Put another way, it should draw a dividing line which avoids either over-inclusion or under-inclusion. To avoid over- inclusion, the test must prevent software patents from reaching outside the useful arts to cover innovations in the fine arts, in business and in the social and physical sciences (including claims to newly discovered laws of nature and natural phenomena) or abstract ideas (claims to concepts without adequate technological implementation).
The key is to realize that the problem of software claims reaching outside the technological arts arises in connection with claims to the non-patentable subject matter "as such," as opposed to claiming computer technology tools useful in those fields. Merely because a computer system is useful in automating tasks in a non-technological activity (be it translating, accounting, selling or analyzing social or physical science data) does not make the machine itself any less a part of the technological arts. The tool is technology and the resulting increases in speed, accuracy, resource efficiency and the like arising from its use are technological advances. Therefore, the objective is to separate claims to computer technology (apparatus) from claims which reflect innovations in the underlying field of application.
This is precisely the issue addressed and resolved by the express dictates of the proposed test. By requiring that the claim to software functionality be limited to its role as part of a computer system to obtain patentable subject matter status as such, the test eliminates the software as language claims which pose the over-inclusion difficulty. When the limitation is present the expressed functionality is preempted by the claims only to the extent of its use in implementing the related computer system. There is no risk or confusion that the functionality is covered in broad terms as such. If the limitation is absent then software is being used to describe and obtain coverage for the underlying functionality in general application. In these cases the expressed functionality should be assessed directly and independently of the software expression under § 101 in the same fashion as any other non-software claim.
Two examples demonstrate. First, consider software styled claims to a system for pro rated allocation of mutual fund assets and costs similar to those at issue in the State Street Bank case. The invention clearly concerns an activity connected with the operation of a mutual fund business, a non-technological art. In order to pass the proposed implementation test, the claims must only cover use of the methodology as software instruction for a related computer system. Provided they do so, such claims do not prevent others from using the underlying allocation methodology in any other context. Nor do such claims extend to the underlying operation of the mutual fund business or even to the cost allocation activity "as such." They only cover the particular computer tools for implementing the allocation methodology. In this form, the software claims should be considered both useful arts and a statutory article of manufacture.
If, however, the limitation is absent then the claims reach beyond the computer implementation of the software expressed methodology, pre-empting its use by others more generally. In this case any potential software implementation should be ignored and the underlying asset-cost allocation methodology itself assessed to determine if it falls within the useful arts and constitutes patentable subject matter.
Second, consider claims to software implementation of the formula "F=ma." The formula, of course, represents a law of nature and under current Constitutional policy should not be patentable subject matter even in favor of the discoverer. If the claims only cover computer software functionality when serving as a component of a computer system there is no general preemption of the F=ma formula beyond its operation as the software component of the computerized acceleration calculator. The world remains free to use the formula's functionality in any other application. In this form the claimed software functionality is technology and patentable subject matter. In the absence of the limitation, however, then the formula itself must be analyzed without consideration of any software form of expression. Clearly, the law of nature does not involve patentable subject matter.
These examples raise two concerns. First, if the allocation method (or, less likely, the acceleration calculator) can only practically be implemented using a computer system, granting the patent appears to de facto preempt the incorporated and unpatentable functionality. Even though others may be free in theory to use the underlying methodology, in practice they simply cannot. Second, in both cases there appears to be a dramatic imbalance between the simplicity of the software implementation and the scope of the patent preemption. All the work seems to be in the underlying functionality, not in the computer implementation.
Neither concern is properly a patentable subject matter issue. Regarding the first, unquestionably the claimed invention "as a whole" is technology and under current policy should be included within the basic reach of the patent incentive. The fact that the only currently available technology for implementation is covered by the patent does not mean others are precluded from using the underlying functionality. The patent only prevents use of the specific implementation claimed by the applicant; others remain free, and encouraged by the patent incentive, to develop alternative technological means for accomplishing the task. This quest for additional useful art advance, and reward for achieving it, is precisely the point of the patent laws. The patent system is, therefore, working exactly as designed. The remedy, if one is required, lies in seeking a narrowing of the patent incentive system itself.
All that is required to deal with the second concern is to note that finding the claims involve patentable subject matter does not end the patentability inquiry. By passing through the § 101 gate the invention is only ensured a full review concerning the other elements of patentability. Closer scrutiny reveals that the proper basis for this concern is not that the software implementation of an accounting system or acceleration calculator instruction falls outside the useful arts, but whether there is sufficient inventive technological behavior to justify issuance of a patent based on novelty and nonobviousness considerations. This, of course, substantially mitigates the first concern in both examples as well. Only true technological advances will be rewarded with a patent.
Over-inclusion possibilities related to abstract ideas present a slightly more complex analysis and deserve specific attention. In this arena the general policy objection is that the claimed invention has not been sufficiently "applied" to merit reward as progress in the technological arts. On close inspection the "abstract idea" concern implicates two different kinds of abstraction. The first involves true § 101 "abstract ideas" concerns, those claims which completely fail to put the covered idea, good as it may be, to any specified technological use. The second kind of abstraction arises from excessive vagueness concerning how (versus in what context) the idea should be implemented. In these latter cases, which technically deal with § 112 enablement, although one or more technological applications of the idea may be clearly identified, there is insufficient technical detail to actually make it work.
The policies behind both exclusions are related. In both cases the concern is giving a patent "too soon." The specific concerns are different, however. Granting a patent to inventions which are abstract in the § 101 sense risks impeding development of the entire range of specific technological applications of the idea. A patent covering an abstract idea in the § 112 sense impedes development of the specific technology required to actually effect the identified application.
Although related, these two sides of the abstraction coin have very different ramifications and should not be confused. The cases involving the first abstraction concept, the lack of a specified useful arts application, are successfully dealt with by the test in the same fashion as other non- patentable subject matter situations. The test readily separates software claims raising this abstraction difficulty from those that do not. If the claims cover the software only when used as part of a computer system implementation there is by definition a specific useful arts application: use as a component of a computer system. Consequently, there is no patentable subject matter abstract idea concern. If, however, the claims do not contain the limitation then the software is serving in its language role. In these cases the software context is not relevant and the claims must be tested directly for lack of a specified useful arts application in the same fashion as any other potentially unapplied concept.
Software as implementation claims, like all claims, must still be assessed under the second enablement abstraction concept. They are tested under § 112 to determine whether the specification is sufficient to permit a person of ordinary skill in the applicable art to create the appropriate software component and to use it to instruct the appropriate computer system to perform the claimed computerized task. Unless this is possible the invention is not sufficiently mature to permit it to act as a machine component. No patent will issue, not because there is no patentable subject matter, but because of lack of enablement.
That the proposed test not be "under-inclusive" is equally important. As discussed above, the key technological advance in software as implementation inventions lies in the functionality they impart to the related special purpose computer system. Because the test eliminates the irrelevant machine readable media limitation it permits software as implementation to be claimed "per se," eliminating the possible wrongful exclusion of such inventions when claimed as such. Permitting claims to the invention in its actual form (rather than counter to the trend in industry away from hardware implementations) maintains the full incentive, free of irrelevant validity disputes and enforceable directly against anyone making, using or selling the software for use in its implementation role.
One final statutory subject matter argument remains: that convenient as it may be to treat appropriately limited claims to software implementation directly as articles of manufacture, the Constitutional and statutory subject matter provisions prevent this direct coverage of software functionality "as such" because of their intangible form. There are three potential issues. First, as a general policy matter, is there a reason to impose such a limitation? The basic pate