In the over two decades since the Supreme Court issued its decision in Gottschalk v. Benson questioning the patentability of computer software, the courts have tried to articulate when a software-based invention is proper subject matter under section 101 of the patent statute. But because the courts can only judge the cases brought to them and interpret the existing intellectual property laws, the courts are ill-suited for determining both how software should be protected and how patent protection interrelates with copyright and other intellectual property protection for software.
Now, the Patent and Trademark Office ("PTO") has weighed in with its Examination Guidelines for Computer-Related Inventions. Unfortunately, the executive branch is even less capable of resolving how to properly protect computer software than the judicial branch. The guidelines, quite properly, only address the PTO's view on patentability of software-based inventions, and not how such patents fit into the general scheme of intellectual property protection for software. Resolving such questions are beyond the authority of the PTO.
It is time to act on the final advice given by Justice Douglas in Benson: with respect to the proper protection of software-based inventions, "considerable problems are raised which only committees of Congress can manage."
Congress is the only branch of government that properly can address how software should be protected by patents and that can coordinate patent protection with the other forms of protection. With three simple changes in the patent statute, Congress can clarify the patentability of software-based inventions, provide proper protection for truly innovative software advances, and preserve the important balance between user and developer rights provided by the copyright statutes.
Past commentators have questioned the need for, or advisability of, patents for software-based inventions. Some complaints express problematic situations with patents in general, such as when someone who independently develops a patented invention is forced to stop practicing the invention. If those complaints have merit, Congress should address them by reforming the patent system as a whole, not by excluding one area of technology.
Critics have also stated that the term of a patent, 17 or 20 years, is far too long for a fast-moving technology like computer software. Most people view software development as one of this country's most dynamic industries, and, in terms of growth and impact, these people are right. But much software development has not been particularly innovative. If much of the growth of the software industry is not due to innovation, what has caused its explosive growth? One answer is computer hardware, which continues to become more powerful and less expensive. Personal computers available at any appliance store have more power and better graphics than systems found only in a few research labs a decade ago.
Others who oppose software patents have pointed out that the software industry has flourished without patent protection. But a key question remains unresolved: would patents have retarded software development, or have caused that development to advance even more rapidly?
Software developers spend much of their efforts reinventing the wheel because there is not an adequate collection of how problems have been solved in the past. Sometimes they will consult a book on algorithms, but unlike other areas of technology, they will seldom conduct a literature review before starting their work because there is little to find. Until recently, patents were discouraged and commercial developers kept their work secret. Because most software products do not reveal their underlying technology, the user never learns the techniques used to get particular results. Thus, even the wide distribution of a software package does not add to the body of computer science knowledge. For example, when using a word processor or spreadsheet, the user does not learn the techniques used to get particular results. An active patent system for software-related technology would make these techniques known. Then others could improve upon a patent or "invent around" a patent by using new solutions.
In other fields, there is a strong tradition of seeking patent protection that both reveals technological advances and gives protection to those advances. Those who do not apply for patents may find that someone else has been granted the patent, and they will face expensive litigation to establish their rights to practice the invention. In this way, the patent system forces disclosure of what normally would be kept secret in exchange for protecting that disclosure for a limited time.
Moreover, the concern over patent infringement could have accelerated the trend in the computer system development industry towards using software modules purchased from other developers. Developers of electronic circuits only worry whether their overall designs infringe a patent, not whether components that are used in that design are infringing. This is because the suppliers of the components indemnify them for any patent infringement by the component. The use of purchased modules or libraries not only minimizes concerns about patent infringement, but can result in faster software development. Further, because the module developer can spend more effort producing a comprehensive program, greater capabilities or more efficient operations can result.
The latest programming trend is object-oriented programming using languages such as C++. Object-oriented programming is characterized by routines for manipulating data structures that can be incorporated easily into either a higher-level object or a final program. The patent system provides protection that is well suited for object-oriented programming. Truly innovative techniques for manipulating a data structure can be protected as processes. Purchasing an appropriate object for use in the system being developed not only saves programming time, but could provide indemnification against patent infringement for that portion of the system. Concern over patent infringement can encourage the use of object-oriented programming and purchased objects and libraries.
The past position of the courts and the PTO regarding whether software-related inventions were statutory subject matter under section 101 has created other problems. Enterprising patent practitioners tried to hide software-related inventions in obtuse language and unclear claims. Artificial hardware implementations of techniques were often given as alternatives for software implementations to make the invention seem more like a conventional machine. This has not only made it more difficult for people to understand the teachings of the patents, it has made accurate classification and searching more difficult.
Section 101 practices specific to software-related inventions have also resulted in increased pendency of applications and a lower level of scrutiny regarding novelty and nonobviousness. The time spent by an examiner on an application is limited. The hours spent rejecting a section 101 claim and then reviewing the response from the applicant to the rejection may mean that the examiner is left with less time to address the important section 102 and 103 questions.
But the past PTO position on software-related inventions has distorted more than just patent law. When patent protection, the traditional method of protecting a process or machine, was questionable for software and not widely used, other forms of protection for the underlying techniques and implementation structures of software were found. The result, in decisions like Whelan v. Jaslow and the final district court decision in Lotus v. Borland, was an attempt to extend copyright protection to cover the "non-literal" aspects of a program. This can result in protection beyond that provided by the patent law, without disclosure or examination, and for a much longer period of time. But a working patent system for software-related inventions can remove much of the need for protecting software by copyright beyond literal or near-literal copying, resulting in less stretching of copyright protection.
With the issuance of the Examination Guidelines for Computer-Related Inventions, the PTO has accepted the patentability of most software-based inventions. The guidelines effectively and properly shift the emphasis during examinations of software-based inventions from consideration of statutory subject matter to the determination of whether the invention is truly novel and nonobvious. But the guidelines also permit patents covering computer software that is stored in conventional memory devices, creating uncertainties regarding the infringement of such a patent whenever a user copies the software while normally using the software.
It has long been recognized that a book unique only with regard to its content is not patentable. Protecting content is the province of copyright law even though there is no specific prohibition against such protection in the patent statute. Rather, a court-made "printed matter" doctrine denies patentability to material whose only novelty is in its expression. However, In re Gulack notes that the printed matter doctrine "stands on a questionable legal and logical footing" in light of the requirement of section 103 that a claim must be considered "as a whole." In re Lowry refused to extend the printed matter doctrine to data structures stored in conventional memory devices that could only be "processed by a machine."
In In re Beauregard, IBM claimed an already-patented computer process stored in conventional memory devices (e.g., floppy disks) as an article of manufacture made novel by the stored computer process. The Board of Patent Appeals and Interferences had upheld the examiner's rejection of the claims under the printed matter doctrine. IBM appealed the Board's decision to the Federal Circuit. On April 26, 1995, the PTO altered its position and moved for the dismissal of Beauregard, agreeing with IBM that the claimed subject matter was patentable. The PTO stated that they were "preparing guidelines for the patent examining corps, advising the corps the computer programs embodied in a tangible medium, such as a floppy diskette, are patentable subject matter."
The Proposed Guidelines stated that "a computer-readable memory that can be used to direct a computer to function in a particular manner when used by the computer is a statutory 'article of manufacture.'" However, because it is a man-made device, any memory device is an article of manufacture, whether its contents can direct the functioning of a computer or not. Under the Proposed Guidelines, you would have had to look at the contents of a device, rather than just observe that it is man-made and therefore an article of manufacture.
The final guidelines take another approach: to be statutory subject matter, an invention must not only be a "machine, manufacture, composition of matter, or process," but also must be "useful." An invention is "useful" if it has "a practical application." Usefulness or utility previously has been used to limit the patentability of impossible inventions (e.g., perpetual motion machines), ideas not reduced to practice, or chemical compounds with no known use other than further research.
Such a distinction works well for software-based inventions claimed either as a process or as a machine running the software. Implementations of technological applications are patentable, but abstract functions are not. However, the distinction becomes more complicated for memory devices containing information. In this case, the guidelines require that the information be "functional" for it to be given any weight in determining whether the invention is novel and nonobvious: "'functional descriptive material' consists of data structures and computer programs which impart functionality when encoded on a computer-readable medium."
The Examination Guidelines are problematic because they do not recognize that there are two distinct types of computer-readable memories: executable memory that directly controls a digital computer and repository memory from which a program must be transferred before it can control a computer.
Executable memory is the random-access memory ("RAM") or read-only memory ("ROM") that is directly connected to the digital computer's instruction fetching and decoding circuitry. This memory provides the instructions that indicate the sequencing of the computer and the processing of data also stored in the executable memory. Programs in executable memory configure the digital computer, transforming it into a new, special-purpose machine that operates in a specific and predetermined manner. Executable memory differs from a program repository because it implicitly includes the digital computer processor that it directly controls. Functionality is imparted to the controlled digital computer only from information stored in executable memory.
In contrast, programs in a repository memory, such as a hard disk or floppy disk, are merely data for another computer program which loads them into the executable memory. The programs do not configure the computer until they have been loaded into executable memory. In most cases, a loader program must transform the programs into the form that can be executed by the computer (e.g., linking to libraries, relocating addresses, and allocating data areas). There is little difference between the digital data representing programs stored in a repository memory and the digital data representing the audio stored on a compact disc. Both are data that are read and processed digitally to bring about the desired results. Information stored in repository memory does not directly impart any functionality to a computer system.
It is important that law be anchored in reality, and distinctions be drawn only where they actually exist. The "mathematical algorithm" distinction did not work well because there is no similar distinction in computer science. In contrast, the concept that a general purpose computer when programmed becomes a new machine works well because the idea matches almost exactly the concept of virtual machines discussed in operating systems classes.
The Examination Guidelines make a distinction not anchored in computer science. Rather than recognize that the only memory where information is "functional" is executable memory, such as RAM or ROM, the guidelines try to include repository memory as being able to hold functional information. However, the information stored in a repository memory does not impart functionality to a digital computer until it has been loaded into executable memory. Until then, it is simply data, no different than "non-functional descriptive material" such as music, literary works, or a compilation of factual data stored in the same medium.
It is not surprising that the guidelines try to include computer programs when stored in repository memory as functional and therefore patentable, since the PTO had already told the Federal Circuit that the guidelines would be drafted in such a manner. The PTO was prevented from drawing the line between functional and non-functional where computer science dictates: between programs in executable and repository memories.
Although consideration should be given to the effect on copyright of patents for articles of manufacture that are programs stored in repository memories, such consideration is clearly outside the permissible scope of the PTO guidelines. Because programs are works of authorship stored in a tangible medium of expression, they are already protected by copyright. Further, copyright law not only provides protection to the author, but also provides rights to users.
It is an infringement of a patent to make the patented invention. Whenever the program is copied, a new instance of the program stored in a memory is made. If a memory storing a particular program is patented, any copying results in a new instance of the program stored in memory that infringes the patent. For most inventions, a replicator capable of taking one instance of a machine or manufacture and directly producing a new instance would be something out of science fiction. For a program stored in memory, the copy command on the computer system is such a replicator.
For example, if the program were originally stored on a floppy disk, as would be the case when it is received from the software developer, another instance of the patented stored program is made when the program is copied from the floppy disk onto a hard disk. Another instance is made whenever an archive copy of the hard disk is made. Finally, each time the program is run it must be copied from the hard disk into the RAM of the computer, making yet another instance of the patented program stored in memory.
The patent laws do not contain an equivalent to section 117 of the Copyright Act, which permits copying of a computer program if it is a necessary step in its use. Anyone copying a program that includes a method that is the subject of a "program stored in memory" patent would infringe that patent, no matter how minor the method is to the entire computer program. This would be similar to the situation where an automobile containing a patented screw infringes the patent on that screw, even if the screw is an inconsequential part of the automobile.
In addition, patent law does not provide an equitable balance between the legitimate rights of the patent owner and the rights of users such as provided by "fair use" under section 107 of the Copyright Act. Most analogous to "fair use" is the patent exhaustion doctrine, which allows the rightful owner of a patented device to use or sell that device.
It might be argued that the patent exhaustion doctrine permits the making of new instances of patented stored programs if such copies are necessary for the use of the software. Nevertheless, the patent exhaustion doctrine has never allowed the owner of one instance of a patented article of manufacture to make another instance without the permission of the patent owner. On the contrary, there are many cases that try to resolve whether an apparatus is being repaired, not making the invention, or rebuilt, making the invention and infringing the patent. It is uncertain whether a court would extend the patent exclusion doctrine to include the making of new instances of stored programs.
Patents on computer programs residing in memory devices could preempt many of the user rights provided under the Copyright Act, particularly section 117. By their nature, such patents will be infringed as part of the normal and expected usage of the programs. As a result, such patents distort both the patent and copyright systems simply to provide an avenue for suing the producer, instead of the end user, of a program that infringes a patented method. A far better solution would be to amend the patent laws to make the distribution of a computer program implementing a patented method either a direct infringement or an inducement of infringement.
By making three simple changes to the patent statute, Congress can clarify the patentability of software-based inventions and enhance their protection.
The definition of process given in section 100(c) of the patent statute should be amended to read as follows:
The term "process" means process, art or method, and includes a process implemented in whole or in part by a computer program or a new use of a known process, machine, manufacture, composition of matter, or material.
The definition of a computer program found in the Copyright Act is also satisfactory here: "A 'computer program' is a set of statements or instructions to be used directly or indirectly in a computer in order to bring about a certain result."
The proposed language supplants all the confusing case law on software patentability, from Gottschalk v. Benson to the present. In particular, it does not require that the computer program change the state of some physical thing to be patentable, nor does it differentiate between "mathematical algorithms" and other methods or algorithms for achieving a desired result.
Simply implementing a previously-known process by a computer program would not result in a patentable invention. While the implemented process would be statutory subject matter, it would not be patentable because an implementation of a known technique is obvious to a skilled programmer. Patentability would be determined by whether the invention is really a novel and nonobvious advance, not how the invention can be classified.
While some may question whether processes implemented by computer programs should be statutory subject matter for patents, this amendment recognizes that since Diamond v. Diehr the PTO has issued tens of thousands of patents on computer-implemented processes. Furthermore, under the GATT/TRIPs Agreement, the United States has agreed not to discriminate on the basis of a specific technology when determining patentability.
True mathematical laws such as the quadratic equation, as well as other laws of nature, remain unpatentable because they have no inventor. Examiners should be able to recognize such laws because they will be formulas which are always true, separate from a particular application, rather than inventions for obtaining a particular result. Processes based on mathematical laws or laws of nature that are implemented by computer programs are patentable if they are novel and nonobvious.
While it would still be possible to characterize software-based inventions as either machines or as articles of manufacture, this amendment makes it clear that the preferred claim is for a process or method for achieving a particular result. The claim is not for the computer program itself, but for the process or method implemented by the computer program or a portion thereof. It will no longer be necessary to disguise a software patent by claiming the invention as a hardware system.
The proposed language only allows claims for methods actually implemented on a computer, not performed by a person, although such processes would still be patentable to the extent they were patentable before the amendment.
Such a claim recognizes that the digital computer system is not the novel part of the invention, and allows the examiner to search for prior art that describes the particular steps of the method to determine if the invention is novel and nonobvious. It does not make an old method novel just because it is implemented for the first time on a computer.
The problems with the guidelines previously discussed come from the PTO's position in In re Beauregard and the disfavor for non-statutory restrictions on patentability expressed by the Federal Circuit. However, it is just such a restriction that could provide the solution to the difficulties detailed above. The "printed matter" doctrine says that patentable weight is not given to information stored in a medium unless there is some novel and nonobvious relationship between the information and the medium. As previously noted, the guidelines adopt a variation of the printed matter doctrine, giving no weight to non-functional descriptive material.
A stored information exclusion should be recognized by adding a new sentence to section 103's standards for nonobviousness:
Patentable weight shall not be given to information stored in a medium unless there is a novel and nonobvious relationship between the information and the medium.
At the same time, Congress can indicate in the legislative history that it views with disfavor the past non-statutory restrictions, such as "method of doing business" and "mental processes."
Section 103 requires that claims must be considered "as a whole" and not dissected into their component parts. While considering the "point of novelty" in a claim may help an examiner in searching for the best prior art, it has no place in determining the patentability of a claim. The stored information exclusion allows the examiner to consider separately whether the information stored in memory has a patentable relationship to the memory and if not, consider the memory as if it did not contain any particular information.
With a stored information exclusion, there would be no need to look at the content of a storage medium to determine if the medium is statutory subject matter. It always will be, since it is a man-made object. There would be no need to determine if what is stored in the medium is "functional" or "non-functional" in deciding whether an invention is statutory subject matter. However, in assessing the novelty and nonobviousness of an invention, whether stored information is related to the medium would be considered to determine whether it should be given any weight in assessing the invention.
For example, a computer program stored in any memory device would not be given patentable weight, since it has no relationship to the memory device beyond that of any other computer program. But a data encoding, used to increase the capacity or detect errors in a particular storage device, would be considered by the examiner in determining whether an invention is nonobvious. The text of a book would not be given patentable weight, although the book would be an article of manufacture. DNA sequences stored in a cell would be given patentable weight to the extent that the sequence defines the nature of the cell.
The stored information exclusion reinforces the complementary nature of patent and copyright, particularly for computer software. Copyright does not protect any "procedure, process, system, [or] method of operation"--the subject matter of patents. Under the stored information exclusion, patents would not be available for information or expression unless there was a special relationship to their storage medium. Computer programs per se would be protected by copyright; new methods used in computer programs would be protected by patent.
This statutory change would also mean that data structures would not be patentable as an article of manufacture when combined with an ordinary memory, overturning In re Lowry. But novel and nonobvious ways of organizing data would still be patentable as methods operating on a computer. It would be the techniques for accessing, creating, adding, and erasing the information within the data structures that would be patentable, not the data structures themselves. This corresponds well to the concept in computer science of object-oriented programming, where an object is not only a data structure (often hidden from the user), but a set of routines for creating, accessing, changing, or removing the data structure.
One justification for a Beauregard-type patent on a memory device, like a floppy disk, storing a novel and nonobvious computer program is that the conventional ways of claiming a software-based invention did not provide proper protection against the true infringer. If the invention is claimed as a method, the claim is infringed only when the method is actually used. If the invention is claimed as a machine running the program, the claim is infringed only when the program is actually loaded into the computer. In both cases, it is the end user of the program who is the actual infringer, and who may not be aware that use of the software infringes a patent.
But the true infringer of the patent on the software-based invention is the competitor who included the patented method in their software and sold it to the end user. The competitor could be accused of inducing or contributing to the infringement of the patent. However, such an accusation presents a number of problems. For example, to be a contributory infringer, the competitor must know that what they are producing will lead to an infringement; they must have actual knowledge of the patent.
Beauregard-type claims attempt to get around these problems by claiming any memory device containing the patented software method. But as discussed above, such claims cause problems because end users infringe the patent whenever they copy or use the program, and because such claims base the patentability of a memory device on the intended use of its contents.
These problems again point out the difficulty of trying to adapt the patent laws judicially or administratively, rather than legislatively. If it is desirable that people distributing software containing a patented process be infringers, there are a number of ways section 271 can be amended. For example, paragraph (a) can be expanded by adding a second sentence, such as the following:
Distributing a computer program implementing a patented process infringes the patent on the process.
Such language would be a far better approach than the Beauregard-type claim because it would extend to any patented process. Additional claims of memory devices containing programs that implement the process would not be necessary (and would not be permitted under the amended section 103, as discussed above). Moreover, the infringement is tied to distribution, not to production, use, or sale. As such, users will not infringe the patent if they simply copy the program to another memory device. "Distributing" could have its common meaning, or a definition could be added to section 100 indicating that it includes not only sales, but any transfers to another. Because distributing can include not only the transfer of physical storage media such as floppy disks, but also electronic transfer of computer programs over a data network, this approach addresses future software distribution techniques not involving physical objects. Just as producers and sellers of infringing software would be liable for infringement, so too would be the operator of a bulletin board system distributing pirated patented software.
Over two decades ago, in his Gottschalk v. Benson decision, Justice Douglas called on Congress to clarify the patentability of software-based invention. Decades of confusing court decisions show that the problem cannot be handled adequately by the judicial branch. Because the problems extend beyond the issuance of patents and concern patent enforcement and other forms of software protection, such as copyright, the PTO cannot adequately solve the problems through examination guidelines. Only legislation can adequately clarify the patent of software-based invention. Such legislation need not be complex. This article suggests three simple amendments to the current patent statute which clarify that computer-implemented processes are patentable, exclude information stored in obvious ways in determining novelty, and protect those holding patents on methods from those who distribute computer programs implementing those methods.
* Lee Hollaar is a Professor of Computer Science at the University of Utah, where he teaches computer intellectual property law and conducts research into information retrieval systems. During the 1996-97 academic year, Professor Hollaar is a scholar-in-residence at the Dean Dinwoodey Center for Intellectual Property Studies of the George Washington University. He received his Ph.D. in Computer Science in 1975 from the University of Illinois and a B.S. degree in Electrical Engineering from the Illinois Institute of Technology in 1969. He is also a registered patent agent, primarily practicing in the area of computer-related inventions. The views expressed are his own and not necessary those of the University of Utah or any other organization. Professor Hollaar would like to thank Mark Lemley, Bill Ellis, and John Ogilvie for their comments on the first draft of this article. An earlier version was distributed to the membership of the American Committee for Interoperable Systems (ACIS) at the February 2, 1996, membership meeting.
1. 409 U.S. 63, 175 U.S.P.Q. (BNA) 673 (1972).
2. 35 U.S.C. 101 (1994) ("Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.").
3. The guidelines became effective on February 28, 1996. Examination Guidelines for Computer-Related Inventions, 61 Fed. Reg. 7,478 (1996) [hereinafter Examination Guidelines]. Proposed guidelines were issued for public comment on June 2, 1995. Request for Comments on Proposed Examination Guidelines for Computer-Implemented Inventions, 60 Fed. Reg. 28,778 (1995) [hereinafter Proposed Guidelines].
4. Benson, 409 U.S. at 73, 175 U.S.P.Q. (BNA) at 677.
5. See, e.g., Pamela Samuelson, Benson Revisited: The Case Against Patent Protection for Algorithms and Other Computer Program-Related Inventions, 39 Emory L.J. 1025 (1990); Richard Stallman and Simson Garfinkle, Viewpoint; Against Software Patents; Software Should Be Copyrighted, Not Patented, 35 Comm. ACM. 17 (Jan. 1992); Hearings on Use of the Patent System to Protect Software-Related Inventions Before the United States Patent and Trademark Office (Jan. 26-27 & Feb. 10-11, 1994) (transcript and comments), reprinted in Gregory A. Stobbs, Software Patents 402, 505 (1995) [hereinafter Hearings].
6. Until January 1, 1996, the term of a patent was seventeen years from its issue. 35 U.S.C. 154 (1988). As a result of the General Agreement on Tariffs and Trade, opened for signature Oct. 30, 1947, 61 Stat. A3, 55 U.N.T.S. 1987, the patent term is now twenty years from the first application for the patent. 35 U.S.C.A. 154 (West Supp. 1995). The term of a patent can be extended for certain delays during its pendency. 35 U.S.C. 154 (1994).
7. See Hearings, supra note 5.
9. See In re Alappat, 33 F.3d 1526, 1570-71, 31 U.S.P.Q.2d (BNA) 1545, 1580 (Fed. Cir. 1994) (Newman, J., concurring).
10. Even though the United States is one of the few countries with a "first to invent" patent system, rather than a "first to file" system, unless the first inventors have kept excellent records of their work from conception to reduction to practice, it is difficult for them to prevail in an interference action or in court. This is because, in an interference, there is a rebuttable presumption that the parties made their inventions in the order of their filing dates. The junior party can overcome this presumption with a preponderance of the evidence unless the junior application was filed after the issuance of a patent on the invention, in which case clear and convincing evidence is required.
11. Section 2-312 of the Uniform Commercial Code requires a merchant seller to warrant that goods do not infringe any intellectual property rights, unless there is an agreement to the contrary or the goods were built to the buyer's specifications. U.C.C. 2-312 (1995).
12. C++, developed in 1980 by Bjarne Stroustrup of Bell Telephone Laboratories, is an extension to the C programming language that facilitates object-oriented programming.
13. If an object were considered "goods" under the Uniform Commercial Code, such indemnification would be automatically provided. U.C.C. 2-312 (1995). Otherwise, it might be used as a selling point by the object provider.
14. 35 U.S.C. 101 (1994).
15. See Hearings, supra note 5.
16. Often, the computer process is transformed into an apparatus by reciting the "means for" performing each step of the process. 35 U.S.C. 112 (1994). This is permitted under section 112, id., and the Examination Guidelines, supra note 3.
Section 112 states that "means for" claim elements are to be construed to cover what is described in the specification for that element and its equivalent. See In re Donaldson Co., Inc., 16 F.3d 1189, 29 U.S.P.Q.2d (BNA) 1845 (Fed. Cir. 1994) (en banc). However, sometimes a specification contains little or nothing describing the "means for" element. See In re Trovato, 42 F.3d 1376, 1382, 33 U.S.P.Q.2d (BNA) 1194, 1199 (Fed. Cir. 1994), vacated, 60 F.3d 807, 35 U.S.P.Q.2d (BNA) 1570 (Fed. Cir. 1995) (en banc). The lack of descriptive material in the specification makes these claims a "black hole" enveloping everything within the claim. Such black hole claims should be rejected under section 112, since they are not "particularly pointing out and distinctly claiming" the invention. 35 U.S.C. 112.
17. Id. 102-03. This author has reviewed file histories for issued patents where most of the prosecution was spent finding claim language to satisfy the examiner's section 101 rejection. When such language is found (and it is not substantially different from the original claim language), the application is allowed with only the slightest consideration of prior art.
18. Whelan Assoc., Inc. v. Jaslow Dental Lab., Inc., 797 F.2d 1222, 230 U.S.P.Q. (BNA) 481 (3d Cir. 1986).
19. Lotus Dev. Corp. v. Borland Int'l., Inc., 831 F. Supp. 223, 30 U.S.P.Q.2d (BNA) 1081 (D. Mass. 1993), rev'd, 49 F.3d 807, 34 U.S.P.Q.2d (BNA) 1014 (1st Cir. 1995), aff'd, 116 S. Ct. 804 (1996).
20. For example, the First Circuit, in its decision in Lotus v. Borland, 49 F.3d at 815, 34 U.S.P.Q.2d (BNA) at 1021, held that the Lotus menu command hierarchy was a "method of operation" excluded from copyright protection under 17 U.S.C. 102(b) (1994) and, citing Baker v. Selden, 101 U.S. 99 (1879), found that such aspects of a computer program should be protected by a patent, if they are to be protected at all.
21. Examination Guidelines for Computer-Related Inventions, supra note 3.
22. Id. at 7,481.
23. Id. at 7,483.
24. See In re Miller, 418 F.2d 1392, 164 U.S.P.Q. (BNA) 46 (C.C.P.A. 1969).
25. 703 F.2d 1381, 217 U.S.P.Q. (BNA) 401 (Fed. Cir. 1983).
26. 35 U.S.C. 103 (1994).
27. In re Gullack, 703 F.2d at 1385 n.8, 217 U.S.P.Q. (BNA) at 403 n.8. Another form of the doctrine says that printed matter is not statutory subject matter. In re Chatfield, 545 F.2d 152, 191 U.S.P.Q. (BNA) 730, 735 (C.C.P.A. 1976). This is problematic, because something that is clearly man-made either is or is not an article of manufacture because of its content, not its physical structure.
28. 32 F.3d 1579, 32 U.S.P.Q.2d (BNA) 1031 (Fed. Cir. 1994).
29. Id. at 1583, 32 U.S.P.Q.2d (BNA) at 1034.
30. 53 F.3d 1583, 35 U.S.P.Q.2d (BNA) 1383 (Fed. Cir. 1995).
31. Id. at 1584, 35 U.S.P.Q.2d (BNA) at 1384.
32. Id. at 1583-84, 35 U.S.P.Q.2d (BNA) at 1383-84.
33. Software Embodied on Diskette is Patentable Subject Matter, PTO Declares, 50 Pat. Trademark & Copyright J. (BNA) 3 (May 4, 1995).
35. Supra note 3.
37. Presumably, a floppy disk containing both a computer program and a text file would be an article of manufacture when one is looking at the program, but not when looking at the text file. It is not surprising that the PTO looked for a different approach to determine if a floppy disk was an article of manufacture for their final guidelines.
38. Examination Guidelines, supra note 3, at 7,481.
39. Id. at 7,479.
40. See, e.g., In re Ferens, 417 F.2d 1072, 163 U.S.P.Q. (BNA) 609 (C.C.P.A. 1969).
41. See, e.g., Rubber-Tip Pencil Co. v. Howard, 87 U.S. 498 (1874).
42. See, e.g., Brenner v. Manson, 383 U.S. 519, 148 U.S.P.Q. (BNA) 689 (1966).
43. See Gottschalk v. Benson, 409 U.S. 63, 155 U.S.P.Q. 673 (1972).
44. Examination Guidelines, supra note 3, at 7,479.
45. In re Freeman, 573 F.2d 1237, 197 U.S.P.Q. 464 (C.C.P.A. 1978), attempts to make sense of the term "algorithm" as used in Gottschalk v. Benson, 409 U.S. 63, 65, 175 U.S.P.Q. (BNA) 673, 674 (1972). The court correctly recognizes that an algorithm is a "step-by-step procedure for solving a problem or accomplishing an end." Freeman, 573 F.2d at 1245, 197 U.S.P.Q. (BNA) at 471. But this could also be the definition for a process, something that is clearly statutory subject matter. The court observes that the "refusal to recognize that Benson was concerned only with mathematical algorithms leads to the absurd view that the court was reading the word 'process' out of the statute." Id. at 1246, 197 U.S.P.Q. (BNA) at 471.
46. See, e.g., In re Alappat, 33 F.3d 1526, 1544-45, 31 U.S.P.Q.2d (BNA) 1545, 1558 (Fed. Cir. 1994).
47. 35 U.S.C. 271(a) (1994) ("whoever without authority makes, uses or sells any patented invention, within the United States during the term of the patent therefor, infringes the patent.").
48. 17 U.S.C. 117 (1994).
50. Id. 107.
51. See United States v. General Elec. Co., 272 U.S. 476 (1926); Met-Coil Sys. Corp. v. Korners Unlimited., Inc., 803 F.2d 684, 231 U.S.P.Q.2d (BNA) 474 (Fed. Cir. 1986).
52. See, e.g., Lummus Indus., Inc. v. D. M. & E. Corp., 862 F.2d 267, 269, 8 U.S.P.Q.2d (BNA) 1983, 1986 (Fed. Cir. 1988) ("The law entitles the purchasers of a patented apparatus to repair and replace the worn or broken parts, but replacement that amounts to a 'second creation of the patented entity' is not permissible."); Dana Corp. v. American Precision Co., Inc., 827 F.2d 755, 3 U.S.P.Q.2d (BNA) 1852 (Fed. Cir. 1987); Met-Coil Sys. v. Korners Unlimited, Inc., 803 F.2d 684, 231 U.S.P.Q. (BNA) 474 (Fed. Cir. 1986).
53. 17 U.S.C. 117 (1994).
54. 35 U.S.C. 100(c) (1994).
55. 17 U.S.C. 101 (1994).
56. 409 U.S. 63, 175 U.S.P.Q. (BNA) 673 (1972).
57. The requirement for the subject matter of a process "to be transformed and reduced to a different state" comes from Cochrane v. Deener, 94 U.S. 780, 788 (1876). However, Cochrane was addressing a process not tied to a particular machine. This is noted in Benson, where Justice Douglas indicates that "[t]ransformation and reduction of an article 'to a different state thing' is the clue to the patentability of a process claim that does not include particular machines." 409 U.S. at 70, 175 U.S.P.Q. (BNA) at 676 (emphasis added).
This is merely a restatement of the principle that patents apply to the technical arts, because they address either particular machines or compositions of matter, or processes that produce a physical change. A computer program, as opposed to an abstract algorithm, is always in the technological arts, because it operates on a particular machine--the special-purpose computer formed by programming a general-purpose computer to perform the specified function. See In re Alappat, 33 F.3d 1526, 1545, 31 U.S.P.Q.2d (BNA) 1545, 1558 (Fed. Cir. 1994).
State Street Bank and Trust v. Signature Financial, 927 F. Supp. 502, 38 U.S.P.Q.2d (BNA) 1530 (D. Mass. 1996), turned the principle of Cochrane on its head. The patent in suit was claimed as an apparatus, specifically a digital computer having the means for performing a specified sequence of operations. The court first determined that, because the sequence of operations could run on any digital computer, the claimed apparatus should be viewed instead as a process. It then determined that the resulting process did not involve a physical transformation, and was therefore not statutory subject matter. But the claimed invention is a particular machine, and therefore clearly in the technological arts and statutory subject matter.
58. See, e.g., In re Freeman, 573 F.2d 1237, 197 U.S.P.Q. (BNA) 464 (C.C.P.A. 1978).
59. See, e.g., Richard H. Stern, Solving The Algorithm Conundrum: After 1994 In the Federal Circuit Patent Law Needs A Radical Algorithmectomy, 22 Aipla Q.J. 167 (1994). After removing "algorithms" from the present patent statutes, Stern suggests a new petty patent for algorithms. Id. at 170. Such a radical change will cast us on new uncharted waters just as things are becoming reasonable for the protection of software-based inventions.
60. 450 U.S. 175, 209 U.S.P.Q. (BNA) 1 (1981).
61. Agreement on Trade-Related Aspects of Intellectual Property Rights, Including Trade in Counterfeit Goods, opened for signature April 15, 1994, 33 I.L.M. 81.
62. Id. at 93-94. Exclusions can be made only to protect "morality" or the environment, medical procedures, or plants and animals if there is an alternative protection for plant varieties. Id. at 94.
63. The form for such a claim might be:
A method for doing [the objective of the method] on a digital computer,
said digital computer including [the components that are necessary to perform the method (such as disks for storing data, keyboards for operator entry, or displays for showing results) providing an antecedent basis for those components],
said method comprising:
A. [step one of the method];
B. [step two of the method];
[remaining steps of the method].
For those wishing to additionally claim the invention as a machine, a dependent claim such as "A digital computer programmed to perform the method of claim N" could be used.
64. Examination Guidelines, supra note 3.
65. 53 F.3d 1583, 35 U.S.P.Q.2d (BNA) 1383 (Fed. Cir. 1995).
66. Id. at 1548, 35 U.S.P.Q.2d (BNA) at 1384 (Fed. Cir. 1995).
67. 35 U.S.C. 103 (1994).
68. Judge Newman has correctly pointed out that "method of doing business" is a fuzzy and "unwarranted encumbrance to the definition of statutory subject matter" and advises "that it be discarded as error-prone, redundant, and obsolete. It merits retirement from the glossary of section 101." In re Schrader, 22 F.3d 290, 298, 30 U.S.P.Q.2d (BNA) 1455, 1461-62 (Fed. Cir. 1994). However, the method of doing business exception returned in its full glory in State Street Bank and Trust v. Signature Financial, 927 F. Supp. 502, 515-16, 38 U.S.P.Q.2d (BNA) 1531, 1542 (D. Mass. 1996). If the Court of Appeals for the Federal Circuit upholds this exception in the pending appeal, the validity of well over a thousand similar patents would be in question.
69. 35 U.S.C. 103.
70. See, e.g., In re Musgrave, 431 F.2d 882, 167 U.S.P.Q. (BNA) 280 (C.C.P.A. 1970).
71. 17 U.S.C. 102(b) (1994).
72. 32 F.3d 1579, 1584, 32 U.S.P.Q.2d (BNA) 1031, 1035 (Fed. Cir. 1994). The method claims in Lowry would remain patentable subject matter.
73. 35 U.S.C. 271(c) (1994) ("Whoever offers to sell or sells . . . a component of a patented machine . . . knowing the same to be especially made or especially adapted for use in an infringement of such patent . . . shall be liable as a contributory infringer." (emphasis added)).
74. Id. 271.
75. This added protection would be similar to the added protection for processes already given by section 271(g), which makes it an infringement to make, use, or sell a product made by a patented process. 35 U.S.C. 271(g) (1994).
76. Id. 103.
77. Id. 100.
78. 409 U.S. 63 , 175 U.S.P.Q. (BNA) 673 (1972)
79. Id. at 73, 175 U.S.P.Q. (BNA) at 677 (1972) ("considerable problems are raised which only committees of Congress can manage.").