Journal of Law, Information and Science
|
|
Home
| Databases
| WorldLII
| Search
| Feedback
Journal of Law, Information and Science |
|
by
John V. Swinson[*]
As a result of decisions of the United States Patent Office and the Courts most software related inventions can now be protected by patent. This has been criticised by some who consider that future development in software will discontinue if the buidling blocks of computer programming and computer science are removed from the public domain. This article reasons that patent protection should be available to those who develop new and innovative softare. A patent system will encourage research and development in this area and the concerns of those opposed to software patents is overstated. The current law applicable to the patent protection of software is discussed as are the policy grounds for and against patent protection. The conclusion is reached that most criticisms are not specific to software patents but about the patent system generally and that not to allow patents for software inventions may be highly detrimental to science and industry.
___________________________
"... by the production of some useful effect, ... patent law has distinguished, so far as it has distinguished, between the discovery of a principle of science and the making of an invention." I.B.M. v Commissioner of Patents, (1991) FCR 218, 224 per Burchett J.
Patent law protects "anything under the sun that is made by man." Diamond v Chakrabarty, [1980] USSC 119; 447 U.S. 303, 309 (1980).
A computer program is not something that exists naturally in the universe, waiting for a person to discover it and then input it into a computer. New computer software techniques take time, effort and money to develop. One must be imaginative and skilful to create new computer software or develop a new software technique. A computer, operating in accordance with a computer program, is a useful device.
Until recently, however, the United States Patent Office routinely rejected many patent applications for software related inventions on the ground that software, or more correctly, the algorithm represented by the software, was like an abstract mathematical or scientific principle and preexisted in nature. The algorithm was merely discovered. Recent United States Patent Office and Court decisions have discredited this reasoning. As a result, most software related inventions can now be protected by patent in the U.S. Many patents for software related inventions have already issued. This change, and a corresponding decrease in the scope of copyright protection given to computer code by U.S. courts, has greatly increased the role of the patent system in protecting the investment in software.[1]
Many software developers are alarmed that patent protection can be obtained for software related inventions. They believe that the underlying principles of computer programming and computer science should not be removed from the public domain to become property of one person or one company. It is their belief that if the building blocks of computing are not available for all to use, then future development in the software field will grind to a halt.
It is the thesis of this article that software developers who invent new and novel software techniques should be rewarded with the grant of a patent, just like inventors of other new and novel methods and apparatus. Software companies, like companies in other innovative fields, should be encouraged to risk time and money on research and development. The incentives of the patent system provide such encouragement. The fear of those opposed to software related patents is largely unfounded - there is no reason to believe that the effects of the patent system on computer software development will be any different to its effects on development in other areas of commerce and industry. As experience has shown, the patent system, when operating optimally, results in benefits to society greater than any detriment that may result from the granting of exclusive rights to individual inventors.
The first part of this article will outline the current law applied in the U.S. to determine if a software related invention is the proper subject of a patent. Recent cases will be discussed to illustrate that the patentable subject matter test is not a bar to patenting most new and novel software applications and programming techniques. The second part of this article will consider whether granting patents for software related inventions is a positive development. On policy grounds, should governments grant patents for novel features of software products? Are patent rules are able to deal with the peculiarities of computer software? It will be shown that the complaints that the patent system is counter-productive to the growth of the software industry are directed to the patent system generally, and are not specific to software related patents.
Unlike copyright, the author of a new computer program cannot routinely apply and receive a patent to protect her new program. It is only a software related invention that is embodied in the program, that can be protected by patent. Typically, a software related invention is at a higher level of abstraction than that of computer code. For example, a typical software related invention may be a data structure used in an expert system program, an algorithm used to search a database or a method for allowing a user to interact with an operating system. Computer code is merely a representation of one embodiment of the software related invention. Copyright will protect the expression used to implement the software related invention in a particular computer language. Patents have a broader scope. Patents will protect any new and novel software related invention, regardless of how it is expressed in code or the computer language used to implement the invention. It is the high level concept, not the actual implementation of the concept, that is the proper subject of a patent.[2]
Therefore, when one looks at a patent for a software related invention, it is unusual to see any computer code in the patent at all. The patent will describe the software related invention in general terms, explaining its application and how it is an improvement over the prior art. Diagrams of relevant hardware utilized by the software, algorithms (usually in flow chart form) and data structures are explained. Sometimes, depending on the type of software related invention, typical screen displays are included. Importantly, the patent will have a number of claims that define the invention, in effect setting out its legal boundary. For a new and novel algorithm, method claims will describe, usually at a high level of abstraction, the steps of the method or process that is the algorithm. For a new and novel software structure,[3] apparatus claims will detail the components that comprise the structure. It is the claims of the patent, not computer code, that a court will examine to determine if the patent is infringed. As it is the high level description of the invention in the claims of the patent that define the invention, what is patented is generally wider in scope than the expression in computer code.
Most programmers, when creating a new computer program, will not invent anything that is patentable. Many clever programming tricks are likely to have been used previously by someone else writing a program in the same computer language. The programmer's coding techniques will rarely be general enough to have application when other programming languages are used. The code itself (compared with the underlying algorithm that is coded) is not the invention that is the subject of patent protection. Usually, it will be a team of research computer scientists, solving problems at a high level of generality, who will create software related inventions that are worthy of patent protection. Consider, for example, how often a programmer will develop a new application for software (such as a spreadsheet) or a new data structure (such as a B-tree) or a new algorithm (such as for data compression). These inventions do not generally result from the coding process.
When considering all the computer software ever created, only a small percentage will have elements protectable by patent.[4] But in absolute terms, because there are a large number of software companies undertaking constant research and development, many software related patent applications are filed each day. Copyright still has an important role in protecting the investment in programs. Copyright will protect most, if not all, computer programs immediately upon creation. But the scope of copyright does not extend to much beyond the code's literal expression. Software related patents protect high level concepts of general application, such as algorithms and data structures. Due to the requirement of novelty, patents are harder to obtain than copyright protection, however they have a much wider scope.
Until recently, the hurdle most software inventors faced in obtaining a patent was that a software invention was not considered to be the proper subject of a patent. The rules applied in the U.S. to determine if a computer software related "invention" is the proper subject of a patent are simple. The Patent Act[5] states that if what is claimed to be invented is "any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof" then it is patentable.[6] Excluded from patent protection are laws of nature, natural phenomena, and abstract ideas.[7]
The U.S. Patent Office, interpreting this statutory mandate and various court decisions,[8] applies a two-part test[9] to determine if a software method is statutory subject matter, or alternatively, if any patent over the method will preeempt something already existing in nature.[10] The two-part test is as follows:
It is first determined whether a mathematical algorithm is recited in the claims. If so, it is next determined whether the claimed invention as a whole is no more than the algorithm itself; that is, whether the claim is directed to a mathematical algorithm that is not applied to or limited by physical elements or process steps. Such claims are nonstatutory. However, when the mathematical algorithm is applied in one or more steps of an otherwise statutory process claim, or one or more elements of an otherwise statutory apparatus claim, the [statutory requirements] are met.[11]
The first part of this test determines whether the claim, directly or indirectly, recites a mathematical algorithm. Most claims for software methods will not recite a mathematical algorithm, and accordingly, will recite statutory subject matter. The second prong of the two-part test is rarely reached for software related inventions.
The Supreme Court defines a mathematical algorithm as a "procedure for solving a given type of mathematical problem."[12] The Board of Patent Appeals and Interferences has further explained the concept of a mathematical algorithm, stating that:
a claim should be considered as reciting a mathematical algorithm only if it essentially recites, directly or indirectly, a method of computing one or more numbers from a different set of numbers by performing a series of mathematical calculations.[13]
The emphasis is "on what the claimed steps do rather than how the steps are performed."[14] If the claimed process is "essentially directed to detecting the occurrence of events",[15] rather than "propos[ing] a solution to a mathematical problem", then the claimed process is statutory. Therefore only if the claimed method takes pure numbers[16] as input, performs mathematical calculations, and outputs pure numbers, is the method not the proper subject of a patent. Where the input or output are not pure numbers, or the invention is directed to performing a real-life function unrelated to number crunching, then the method does not solve a mathematical problem and passes the statutory subject matter test. In other words, if the object of the method is to accomplish a task which is not simply computation of a mathematical formula, the method, whether carried out on a computer or not, is statutory. As most computer programs do not solve mathematical problems, most software related inventions will be patentable.
For example, (i) a method (implemented in software) that detects the commencement of inspiration intervals by determining when a time-varying respiration signal crosses an adjustable trigger level,[17] (ii) a method (implemented in software) for use with an expert system that diagnoses and provides remedies for problems incurred in the operation of a manufacturing process[18] and (iii) a method for translating between natural languages (implemented in software) using a programmable computer where the computer carries out "unthinking, abstract mathematical operations on abstract values stored in the memory of the computer"[19] all do not solve a mathematical problem and are therefore the proper subject of a patent. In contrast, a method for solving a complex vector equation or for solving the Pythagorean theorem solve mathematical problems and cannot therefore be patented.[20]
A computer program is an algorithm (or method) that describes the steps a processor is to perform to accomplish a task. Rarely is the task mathematical, and therefore, rarely is a computer program a mathematical algorithm under the Supreme Court's definition. Thus, most algorithms embodied in software are the proper subject of a patent.
The above two-part test only applies where the claims are drafted in method format. Many software related inventions can be claimed as an apparatus. This is particularly easy to do, for example, if the software is modular in design or if the invention is a unique data structure. Claims "truly directed to an apparatus as a 'machine' or 'manufacture' ... do not fall within the judicially determined mathematical algorithm exception..."[21] An apparatus is simply not a mathematical algorithm. Further, all software operates in conjunction with a processor and is usually stored in a memory. Therefore, the software related invention can often be claimed as a system comprising a computer program, a memory and a processor. If the claim "contains recital of specific elements which go well beyond any steps of an algorithm,"[22] such as a memory[23] or a processor, then the mathematical algorithm rule does not apply.
In summary, after applying only the first prong of the two-part statutory subject matter test, a software related invention will rarely be nonstatutory subject matter.[24]
At least judge sitting on the U.S. Court of Appeals for the Federal Circuit,[25] Rader J., has abandoned the two-part test for statutory subject matter. Rader J.'s broader test focuses on the terms of the Patent Act, "rather than on non-statutory, vague classifications."[26] Doubting if even a mathematical algorithm is nonstatutory, Rader J. concentrates on whether the claim as a whole discloses a patentable process in the broad sense. Only laws of nature, natural phenomena and abstract ideas cannot be patented under Rader J.'s analysis. Further, Rader J. concludes that a method or process performed by a computer is not "like" a law of nature. Under Rader J's thoughtful analysis, it would be difficult to imagine that a software related invention could ever be nonstatutory subject matter.
To conclude this section, and to convince those who are sceptical about the wide scope of software related inventions, the following are examples[27] of recently issued "software patents":
1. "Method For Prioritizing Data In Financial Information System", U.S. Patent No. 5,095,429 issued 10 March, 1992. (The patent includes three flow charts, a screen display (shown in Fig. 1) and five pages of code in a Pascal like language. The patent is for a method that allows a user to lock a cell on a spreadsheet to prioritize the data in that cell.)
2. "Graphic File Directory And Spreadsheet", U.S. Patent No. 5,093,907 issued 3 March, 1992. (A system that associates selected cells of a matrix (e.g. a spreadsheet) with graphic data files.)
3. "Real Time Event Driven Database Management System", U.S. Patent No. 5,093,782 issued 3 March, 1992, assigned to Texas Instruments.
4. "Method And Apparatus For Concurrent Modification Of An Index Tree In A Transaction Processing System Utilizing Selective Indication Of Structural Modification Operations", U.S. Patent No. 5,123,104, issued 16 June, 1992, to IBM Corp.
5. "Data Processing System For Hub And Spoke Financial Services Configuration", U.S. Patent No. 5, 193,056, issued 9 March, 1993.
6. "Apparatus And Protocol For Local Area Network", U.S. Patent No. 5,123,089, issued 16 June, 1992.
7. "Compressing And Decompressing Text Files", U.S. Patent No. 4,955,066, issued 4 September, 1990, to Microsoft Corp.[28]
8. U.S. Patent No. 4,555,775 (a process for the use of backing store to allow hidden windows to be retrieved quickly).
9. U.S. Patent No. 4,308,582 (a menu system for a word processing program).
There are many advocates who predict doom for the software industry if patents for software related inventions continue to issue. The loudest lobby group is the League For Programming Freedom, "a grass-roots organization of programmers and users opposed to software patents".[29] A number of recent government reports have considered the desirability of patents for software.[30] This section will discuss the most common objections to software related patents.
It is argued that if you give a computer-related problem to a group of computer scientists, it is likely that some of them will independently "discover" the same solution to the problem. Moreover, computer scientists, when compared with other scientists and researchers, are more likely to independently and simultaneously solve problems in the same way. Where independent reinvention is common, one rationale of the patent system, the encouragement of the dissemination of knowledge through publication, loses force: anyone who considers the problem is likely, without much effort, to arrive at one of a limited number of solutions. Further, as a patent will prevent anyone (even another independent inventor) from making, using or selling the product or method claimed in the patent, programmers who come up with the patented solution on their own (as is likely) will be prevented from utilising it. Due to the constraints a computer language imposes on the expression of the algorithm, the physical constraints of the computer itself, and the large number of programmers solving problems daily, the suggestion is that there is more repetition of "invention" in the software field than occurs elsewhere in science and business.[31]
The argument as stated above contains a number of fallacies. First, it assumes the existence of the problem to be solved. In many instances, ascertaining the problem, not the solution, is the more inventive step. Problems do not just exist, sitting around waiting to be solved. Inventing a new use for a computer processor, compared with creating the program that instructs the processor, is often the more difficult and inventive step. Second, for many problems, the solution is not easily found. For example, successful data compression techniques for video pictures and algorithms that translate between natural languages have not been flowing forth. Third, simply because some solutions to a problem are easily discoverable does not mean that the best solution will be continually "reinvented" - it may not even be known.
The proponents of the "independent reinvention" argument focus narrowly on two situations: (i) that it is common for programmers, because of the constraints of a particular programming language, to independently write code using similar programming techniques that other programmers use, and (ii) that because of the fast pace of the software industry, even if the first person to invent a major new (high level) technique was never born, then someone else would have invented the same technique at about the same time that it would have been invented anyway.
It may be true that programmers will write computer programs using similar techniques to those used by others. In writing computer programs, where constraints of the programming language inhibit expression, the likelihood of independent "reinvention" of sections of computer code may indeed be great. Once an algorithm has been designed to solve a problem, the likelihood that two independent programmers using the same algorithm and programming language will write the exact code is quite high.[32] The constraints of the programming language, if a well-defined language, do not allow much room for creative thought process or variations in expression. But this assumes that the invention is the computer code.[33] In such a case, the programmer is merely expressing an algorithm at the level the computer can understand, and is not in fact inventing anything. Granting a patent on the way an algorithm is expressed in the form of computer code may indeed be deterimental to the software industry, but this is not the focus of a software related patent.
The inventive steps in software design almost always take place at the level of algorithmic creation, not in coding the algorithm. It is the algorithm, and not the computer code, that the computer scientist wishes to patent. As stated above, the computer code is protected by copyright. Copyright protects expression, where the chances of two people independently producing the same expression is very small. Patent, on the other hand, "operates at the level of generality at which there is at least a plausible possibility of independent creation of the same invention."[34] The more abstract the interest for which protection is given, the more likely are the odds that two people will independently create the same thing.[35] For this reason, proponents of the "independent reinvention" argument claim that at the more abstract level of algorithm design, because of the fast pace of the software industry, two computer scientists are likely to independently invent the same algorithm.
There are three responses to this argument. First, algorithm design by computer scientists is merely problem solving where a computer is the intended processor. Why is the invention of an algorithm that is to be executed by computer any different from the invention of an algorithm where the intended processor is not a computer? Is it because of the physical constraints of the computer? The solution to many problems, not just those involving computers, are subject to physical constraints.[36] Having a computer as an algorithm's intended processor does not determine if independent reinvention of the algorithm is likely.
Second, patent law requires that before a patent issues, the Patent Office conduct a search to determine if the invention is novel and nonobvious. If the Patent Office believes that the invention is so obvious that any computer scientist, with knowledge of the relevant prior art, would be able to come up with the same algorithm as the patent applicant, then no patent will issue.[37] In other words, for a patent to issue, the invention can not be so obvious that independent reinvention is likely.
Third, the patent system throughout its long history has been confronted with boom industries, where the pace of invention has been hectic. In many fields of endeavor, not just that of the software industry, claims of independent simultaneous invention have been made. Although at present there may be more computer scientists solving more problems using more accessible technology resulting in greater progress and more inventions (and possibly the same solutions to those multitudes of problems) than in any other field, this phenomena of intensive creative effort in a developing technology is not new to patent law. At stages of history there is often a race to be first to make a new invention. As existing knowledge reaches a stage where the time is ripe for such invention, numerous people may independently discover the same solution.[38] The patent system has previously dealt with such occurrences: only one inventor will win the "patent lottery".
The criticism is simply a complaint by software developers, that due to the speed of development in the software industry, most software developers will miss out on being granted a patent, and not be able to sell their software product. However, in many instances, the race to invent is caused by the patent system. One benefit of the patent system awarding the patent to the first to invent is that it increases the pace of invention. Software related inventions should not be taken outside the patent system just because, at the present time, computer science is a growing field.
As the late Professor Newell, a computer scientist, writes, "Algorithms of immense generality and scope will continue to emerge so long as science endures."[39] They will continue to emerge for problems as yet unknown and provide better solutions for known problems. Excluding simple and trivial problems, it is unreasonable to assume that even for all known problems, all efficient solutions have been thought of and published. Further research will lead to more solutions and better computer implement algorithms in many fields. It is a goal of the patent system to encourage such research.
A second criticism by computer scientists of the patent system is that, as most software developments are not novel and are obvious, it is inefficient for a government body to determine which software related inventions are novel and nonobvious.[40] Moreover, it is claimed that because the U.S. Patent Office is unable to determine which patent applications describe novel inventions, most patents that issue for software related inventions are for methods that are known. It is argued that the U.S. Patent Office has granted patents in the software field too easily, and without proper knowledge of what is occurring in the industry.
Without carrying out a major research project in this area, it is impossible to determine, as a general matter, whether most software "developments" that result in the issuance of a patent are novel developments in the art or have been done before. It may be that the patent laws that determine if an invention is novel and not obvious do not work well with software related inventions - in other words, that the Patent Office is unable to determine what has been done before. Why would this be so? Although there is a distinction between the tests for novelty and nonobviousness,[41] the same attack is made on both - that these tests allow common software techniques to be patented, thus hindering legitimate developments in the software industry.
The rapid growth and nature of the software industry has resulted in the commercial success of "back yard" companies. Developments are ad-hoc and undocumented. Many new techniques are spread through the use of electronic bulletin boards, where they are not physically stored, and after a period of time may become irretrievable or be unlocatable.[42] Programmers are more concerned with completing contracts than publishing academic papers. "Sometimes it is possible to patent a technique that is not new precisely because it is obvious - so obvious that no one would have published a paper about it."[43]
Not only is it difficult to determine what the prior art is or was at any particular time, the problem is exacerbated because the U.S. Patent Office has had until recently, little experience dealing with computer software.[44] The U.S. Patent Office, until recently, did not have a special classification for software related inventions, and did not have a software library, making it difficult for Examiners to find existing prior art.[45] This resulted, and may still result, in patents being wrongly granted in borderline cases. Even the Director for the "Software" Examining Group in the U.S. Patent Office, Mr Gerald Goldberg, has publicly stated at seminars around the country that many software related patents that have issued are probably invalid.[46]
It is also argued that the "standard of obviousness developed in other fields is inappropriate for software"[47] because it is the nature of programming to apply the techniques used to solve one problem in the solution to a completely separate problem.[48] For example, sub-procedures, such as sorting routines, are often used over and over in a variety of programs as one tool to accomplish the ultimate goal of the program. As programmers are trained to generalize, it is obvious to them to use or adapt different techniques to different settings.[49]
Moreover, many software developments occur soon after a new hardware technological development.[50] These software developments would have occured sooner if a computer scientist had earlier access to the hardware. The new technique may only be novel because no one has had the opportunity to access such hardware, and it is likely, if history is any guide, that the software technique would be obvious to anyone given the chance to use the new hardware.
The answer to this criticism is not to abandon the patent system, but to improve it through reform. If prior art cannot now be easily found, then it must be catalogued properly. Train computer scientists as patent examiners. Have computer scientists teach the Patent Office about the use of similar software techniques to solve a variety of problems in different areas, so that patents do not issue for obvious software techniques. Some of these steps are already taking place in the U.S.[51] In any event, if it is true that most software developments are not novel and are obvious, then the Patent Office should exercise care, and conduct thorough searches before issuing a patent for a software related invention. The truly great inventor should not be prevented from having a patent issue merely because many patent applications by others are for old or obvious "developments".
In 1980, when the law in this area was more uncertain, and it was generally believed that there was no patent protection for computer programs, a U.S. lawyer wrote "the industry is growing in leaps and bounds without [copyright or patent protection]."[52] At that time, it was further suggested that patent protection would stifle competition, whereas continuation of the status quo would encourage software developers to constantly improve their products.[53] A rationale for the patent system is to provide incentives to inventors and encourage innovation.[54] If additional incentives are unnecessary, a patent will give nothing of value to society, but will increase costs by preventing others from making and using a useful product or process. By 1993, the industry has developed and patents have been granted to software related processes. No study has been conducted to determine whether the perceived change in the law has encouraged or hindered innovation. This section will merely outline the arguments that the patent system impedes innovation in software development.
First, it is claimed that there are many other incentives to invent new software apart from the chance of being granted a patent.[55] When there was no patent protection the industry grew rapidly.[56] Many successful developers never attempted to apply for patents, but still produced software for commercial gain or intellectual satisfaction.[57] Innovation, it is claimed, often occurs by accident in solving problems where invention was not the goal of the programmer.[58]
Second, it is the market that determines whether an innovation will be successful, not the granting of a patent. Developments occur so swiftly that patented programs are of little commercial value in themselves,[59] and any patent will impede possible improvements that could be made to the program by other developers. The lead time a software developer has in the market place gives that developer enough commercial advantage so that the additional incentive of a monopoly over the invention is not needed.[60]
However, software developments can be quickly and cheaply copied and distributed, reducing this lead time advantage. Even when literal copying is not involved, software is different from most other products. It is inexpensive to design and easy to manufacture, as compared with a hardware system with the same number of components.[61] As software is not designed using real components that have to be physically assembled and tested, and not manufactured in large plants that have to be equipped and tooled, but built from well defined mathematical objects, the reverse engineering of software will take less time than for other products.[62]
Third, software is often designed by universities as part of research or through government subsidized programs. The resulting inventions are not due to the incentives of the patent system. On the other hand, universities often patent and develop the results of research, and the investment by private enterprise in such ventures would be unlikely without the knowledge that the product or process was patentable.[63]
Without further study and economic analysis of the patent system, no final conclusions can be reached here. But it should be noted that many of the above arguements are not specific to software related inventions. The issues that computer software innovation present, due to the infancy of the industry and the different methods of production used, may put a gloss on any general findings about the efficiency of patent system in encouraging innovation. In the long term, there is no reason to believe that the ratio of innovation in the software industry simulated by the patent system to innovation in deterred by the patent system will be different from that that would occur in other industries.
At present, the expression in software is protected by copyright. Whether patent should provide protection to software in addition or as an alternative to copyright is examined elsewhere.[64] However, due to the fact that software can be copied easily, one may assume that the incentive to write programs would be less if no intellectual property protection at all was provided. Further, as copyright does not generally protect the high level idea and algorithms of a program, and allows reverse engineering of the program to uncover the program's idea, some protection in addition to copyright seems desirable.[65]
There are a number of points commonly raised as to why licensing of computer software inventions is harmful to the software industry and the community. Although these may be valid complaints, this section will only show that they are not unique to software patents. It will be left for others to prove (if it is possible) that the patent system does not fulfill its claim of promoting efficiency by enabling others to license existing inventions.
The first is that patents for software related inventions are hard to find, and if found, are impossible to understand. The U.S. Patent Office had, until recently, no classification for software related inventions. Software patents were filed everywhere and anywhere, "most frequently classified by end results ... but many patents cover algorithms whose use in a program is entirely independent of the purpose of the program."[66] Even a diligent inventor who did not wish to infringe anothers patent may be unable to find out whether a patent exists over a certain process. For patents issued after 1991, when the classification system was in operation, this argument carries little weight.
When a patent is found, the argument continues, even though it is meant to disclose to the world the new apparatus or process, the patent is often difficult or impossible to understand. Another computer scientist reading a patent, would have a problem establishing if the patent covered his or her product, and as the patent owner is likely to assert that the patent is more comprehensive that it actually is,[67] would be hesitant to proceed into dangerous waters. Such complaints are not new.[68] However, in rebuttal, a corporation uncovering a possibly "dangerous" patent would be wise to seek legal counsel's advice as to the scope of the patent.[69] Patent attorneys, who are generally technically trained and can understand complex legal language, are often able to find reasons to invalidate a patent, to determine ways to design around a patent and can provide advice on whether a planned product infringes a patent.
It is also argued that, as most commercial programs are large and were developed using many software techniques and algorithms, a creator of new software will be unable to search for patents for every process used in the new software. Since each patent search conducted at the Patent Office can cost hundreds dollars and the new software may have thousands of "danger points", a complete search would be far too expensive. This ignores the fact that all patents issued in the past 20 years can be easily and cheaply search via electronic online databases. It would also be rare for a computer program to have a large number of "danger points". What is of more concern is that a U.S. patent application remains secret if and until the U.S. patent issues. As many software related patents take well over two years from filing the application to issuance, even the most comprehensive search will not find the most recent inventions.
If valid patents are discovered that cover a software product, and it is decided to take a license from each patent owner, it is argued that marketing the software may then be unprofitable. For example, if a program infringed twenty patented inventions,[70] each licensed at a rate of one percent, the second software developer would be at a large commercial disadvantage breaking into the market.
This leads to another complaint - that existing firms can stifle competition, and therefore innovation, by obtaining licenses over many different software related inventions, and keep rivals out of the market by refusing to license, by excessive license fees, or by forcing rivals to waste resources inventing inferior processes that accomplish similar results in less efficient ways.[71] The problem is worse for small time programmers, who cannot afford to test in court the validity of a patent.[72] These programmers do not have the money or legal resources to challenge the big software firms.[73]
Moreover, the holder of a patent may have no incentive or desire to license its software patent rights to rivals. Patent protection allows a software developer to introduce its patented program into the market without having to license it and expand the network of rivals. A firm with brand recognition thus would reap increased rewards by preventing a rival reducing the well recognized firm's market share. A compulsory license may be the solution to this problem, especially considering that society benefits from the expanded network of suppliers.[74] The purpose of such a compulsory license is said to be to reduce the extent to which patent ownership conveys monopoly power.[75]
These "abuses" of patents, if the strict enforcement of government granted patent rights can be regarded as an abuse, are not new or solely software related. For example, the United Shoe Machinery Corporation, in the 1950's, had 3,915 patents which, to some extent, blocked potential competition in the shoe making business.[76] Where patents are abused to create monopolies or to restrain trade, the trade practices and antitrust laws may provide the desired remedy.[77]
On the positive side, the patent system allows anyone, even small firms to enter the market place without having its intellectual investment in its product stolen.[78] A small firm, like any other firm, may be granted a patent for its inventions and so have the ability to enter a concentrated market successfully and compete where it otherwise would not be able to do so.
Large software projects are often built from the components of other programs and use techniques developed for other applications but modified or adapted to fulfil the new project's goals.[79] Successful programs are often developed from similar less successful programs, for example, to add a more congenial user interface, or to add new features,[80] or to run on more popular machines or with more accessible operating systems, or to run faster[81] or to use less memory.
If a patent is granted for the original invention, in many cases the later and improved product would infringe that patent. Advances, which often make the program commercially successful, would not be permitted to occur without the permission of the patent owner.[82] Computer scientists may be deterred from further research by the realization that their improvements could not be made or sold without infringing the original patent. This may lead to doubly wasteful results: the original invention may, due to lack of additional desirable features, be ignored in the marketplace and the later software developer, to compete, will have to waste resources inventing around the patent.
Consumers invest large amounts of money in particular software systems, both in purchase costs and more particularly, in staff training costs. One reason for this investment is that they are promised that the software will be upgraded and enhanced by the manufacturer. The consumer is told he or she will always have the best software without having to repurchase new systems or retrain staff each time a technological improvement is made. If the firm producing the software is either found to be infringing an existing patent or unable to upgrade its product due to a patent on the "enhanced" function, the consumer who has already locked in by purchasing the original software, will lose. The consumer must choose to remain with the outdated software or retrain its staff on new software.
The policy question that arises is whether society will benefit more by providing a wider scope for protection for the original inventor, and thus encouraging that inventor to risk more at the outset, or by encouraging subsequent inventors to improve existing software without fear of infringing any patent on the existing product. This question can be phrased as "whether allocating the incremental value of the new technological use to the original or to the subsequent entrepreneur will lead to more creation and marketing of technological advances."[83] It is a question that can be asked for all derivative works and is examined in detail elsewhere.[84] It may be, however, that the nature of computer development and the portability of program modules and algorithms requires that the term or scope of software related patents be limited.
The criticisms presented above do not present a clear solution to the question of whether, on policy grounds, patent protection should be denied to software related inventions. Most criticisms, when examined closely, are mostly complaints about the patent system generally and are not specific to software related patents. Other criticisms raise policy questions where there are sensible arguments on both sides.
One suggested solution is that, as the pre-patent software industry had "no problem that was solved by patents" there should be a complete elimination of software patents.[85] As the answers to the questions presented above are not clear, the former simple position - granting no software patents - should be adopted in case the negative aspects of protection of software developments by patents predominate.
A second solution, which is the one that has, it seems, been adopted by U.S. courts, is to allow patents for software related inventions pending a detailed review of the patent system as a whole. If a process is otherwise patentable, it should be irrelevant that a computer is the intended processor. If an algorithm implemented in hardware is patentable, why should it not be patentable if it is implemented in software? With advances in computer technology and the increase in the use of the digital representation of information, more and more products and processes will be controlled by software. Not to allow patents for software related inventions may be highly detrimental to science and industry.[86]
Whatever solution is adopted, the software industry in the U.S. has realized the present importance of patent protection, and software related patents are issuing by the hundreds ever year. Software related patents have already become one of the major assets and weapons of software companies in the U.S. It is certain that we will see many large court cases to enforce these patents.
[*] Associate, Kenyon & Kenyon, New York City; Barrister-at-law, Queensland;
B.A. (Computer Science), University of Queensland; LL.B (Hons), University of Queensland; LL.M, Harvard Law School. Fulbright Fellow 1990. The author expresses thanks to Victor F. Souto, Esq. and Dr. Natasha H. Christa for their comments on an earlier draft of this article.
[1] See e.g., Atari Games Corp. v Nintendo of America [1992] USCAFED 794; 975 F.2d 832, 839 (Fed. Cir. 1992), a copyright case, holding that the creation of copies of a program by reverse engineering to discover the program's ideas, methods and processes could be fair use and not amount to copyright infringement. The court recgnized that the U.S. Patent Act "provides protection for the method or process performed by a computer in accordance with a program."
[2] Accordingly, copyright will remain the weapon of choice to quickly and efficiently defeat those who make unauthorized copies of software. Patent, on the other hand, can protect basic innovations of enduring value against misappropriation by competitors.
[3] E.g., a data structure, or the program itself if modular in design.
[4] IBM Australia, whose U.S. parent holds many software patents, submitted to the Australian Copyright Law Review Committee that "patent law has only a minor role to play in software protection - as most software is insufficiently novel and inventive..." The submissions are held by the Australian Attorney General's Department and also in the collection of the Harvard Law School library, catalogued as "Collection of letters on copyright protection of software."
[5] 35 USC §101.
[6] "According to this language, 'any' invention or discovery within the four broad categories of 'process, machine, manufacture, or composition of matter' is eligible for patent protection. 'Any' is an expansive modifier which broadens the sweep of categories." Arrhythmia Research Technology Inc. v Corazonix Corp. [1992] USCAFED 298; 958 F.2d 1053, 22 USPQ 2d 1033, 1040 (Fed. Cir. 1992) (hereafter "Arrhythmia"). See also The Economist Policing Thoughts August 22, 1992 ("The Arrhythmia case has alarmed managers in America's booming software industry.")
[7] Le Roy v Tatham 55 U.S. (14 How.) 156, 175 (1852) ("A principle in the abstract ... cannot be patented."); Parker v Flook [1978] USSC 129; 437 U.S. 584 (1978); Diamond v Diehr [1981] USSC 40; 450 U.S. 175 (1981).
[8] In re Freeman 573 F.2d 1237 (CCPA 1978); In re Walter 618 F.2d 758 (CCPA 1980); In re Abele 684 F.2d 902 (CCPA 1982). The two-part test is often called the Freeman-Walter-Abele test for statutory subject matter.
[9] "[The two-part test] is not the only test for statutory subject matter ... and ... failure to meet that test may not always defeat the claim..." Arrhythmia at 1037.
[10] One rationale for the two-part test is said to be to prevent preemption of something that already exists in nature, such as the relationship F=ma.
[11] Arrhythmia at 1037.
[12] Diamond v Diehr [1981] USSC 40; 450 U.S. 175, 186 (1981). This definition was applied by the Patent Office in Ex parte Akamatsu 22 USPQ 2d 1915, 1917 (BPAI, 1992). See also Arrhythmia at 1043 ("Thus after Diehr, only a mathematical procedure for solution of a specific mathematical problem is suspect subject matter." )
[13] Ex parte Logan 20 USPQ 2d 1465, 1468 (BPAI, 1991) (emphasis in original).
[14] Arrhythmia at 1037 (citing Logan at 1468).
[15] Logan at 1467-68.
[16] A digital input, where the numbers input represent something physical, for example, a sound wave signal in digital form, is not regarded as the input of pure numbers. "The view that 'there is nothing necessarily physical about "signals" ' is incorrect." Arrhythmia at 1038.
[17] Ex parte Logan.
[18] Ex parte Kaemmerer, unreported, Appeal No. 92-1717 (BPAI, June 29, 1992).
[19] Application of Toma 575 F.2d 872, 874 (CCPA 1982). See also In re Pardo 214 USPQ 673, 677 (method directed to controlling the internal operations of a computer that operated "on any program and any formula which may be input, regardless of mathematical content. That the computer controlled according to the invention is capable of handling mathematics is irrelevant to the question of whether a mathematical algorithm is recited by the claims.")
[20] See Matter of Application of Bradley 600 F.2d 807, 877, 202 USPQ 480 (CCPA 1979) affirmed by an equally divided court, Diamond v Bradley [1981] USSC 49; 450 U.S. 381 (1981).
[21] Ex parte Alappat 23 USPQ 2d 1340, 1341 (BPAI, 1992); Ex parte Veldhuis, unreported, Appeal No. 92-1102 (BPAI, July 21, 1992) (Apparatus claims "are a fortiori patentable."); Arrhythmia at 1036 ("[A]pparatus functions that entail computer-performed calculations, whether the calculations are described in symbols or in words, do not of themselves render a claim nonstatutory.").
[22] Ex parte Veldhuis at 13-14.
[23] In re Iwahashi [1989] USCAFED 628; 12 USPQ 2d 1908 (Fed. Cir. 1989).
[24] "... because most software falls in this category of non-mathematical algorithm, a significant percentage of software potentially can be patented." Lastova & Hoffman Patents: Underutilized Leverage for Protecting and Licensing Software 6 Computer Law. 7, 8 (1989). Mr Lastova is a Primary Examiner handling computer software applications at the U.S. Patent Office. See also Maier, Software Protection - Integrating Patent Copyright and Trade Secret Law 69 J. Pat. & Trademark Office Soc. 151, 165 (1987) ("patent protection is presently available for virtually all software inventions"), Jack E. Brown The Current Status of Copyright and Patent Protection for Computer Software 12 Computer L. Rep. 406, 407 (1990)("Provided it is not expressed as a pure mathematical algorithm, software that qualifies as nonobvious invention also is protected by patent"), David Bender The Case for Software Patents 6 Computer Law. 2 (1989) ("'software patents' are often available on a cost effective basis and may be quite valuable"); Property Rights Committee, Computer Law Section, State Bar of Michigan, A Survey of US Software Patents issued from July 1987 through December 1987 (Many of the patents granted by the Patent Office "are 'pure' software patents which indicates the Patent Office is now willing to grant patents for novel and nonobvious computer programs operating on conventional off-the-shelf computer hardware").
[25] The Federal Circuit was created in 1982 and is the court that hears appeals in all patent cases.
[26] Arrhythmia at 1044. ("The '459 patent claims a 'process' within the broad meaning of section 101.") Rader J. stated that despite numerous judicial decisions, the term "mathematical algorithm" remained vague. "Without a statutory anchor, this term was buffered by every judicial wind until its course was indiscernible."
[27] See also the list of software patents in Chisum The Patentability of Algorithms 47 U. Pitt. L. Rev. 959, 1021 (1986). The computer company EDS has recently compiled a list of software patents and found that over 2000 have issued in the U.S., 700 in 1989 alone.
[28] Microsoft has obtained at least ten patents. See R. Raysman and P. Brown Effective Establishment of a Software Patent Claim New York Law Journal, August 29, 1991. Microsoft is also a defendant in a current patent infringement suit (Stac Electronics v Microsoft Corp) involving data compression software.
[29] See Against Software Patents, a paper by The League for Programming Freedom, February 28, 1991 and also October 24, 1990, at 8. (Page numbers refer to the 1990 version). The League is based at MIT in Boston and includes as members successful entrepreneurs, executives, independent consultants and programmers, such as Richard P Gabriel, John McCarthy, Marvin Minsky, Robert Boyer and Patrick Winston. It is lead by Richard Stallman, a vocal opponent of software related patents. See, e.g., the debate between Stallman and McCabe in Computerworld Software Patents: Boon or Bane? June 1, 1992 at 33 The Leagues' papers can be obtained via internet mail, contact "league@prep.ai.mit.edu".
[30] See, e.g., The Advisory Commission on Patent Law Reform's Report To The Secretary Of Commerce (August 1992); U.S. Congress, Office of Technology Assessment, Finding a Balance: Computer Software, Intellectual Propery and the Challenge of Technological Change, OTA-TCT-527 (May 1992).
[31] The League for Programming Freedom, a main proponent of this argument, states generally that a programmer "solves many problems in developing each program. These solutions are likely to be reinvented frequently as other programmers tackle similar problems." Against Software Patents at 8.
[32] This does not mean, however, that coding an algorithm is mechanical or that all codings of an algorithm will be identical. The point is that the constraints of the language will limit the number of possible ways of expressing a particular algorithm. Compare encoding an algorithm with writing a screenplay from a novel. Although the plot will be the same, the number of possible screenplays will be extremely large.
[33] Cf Brief of Amicus Curiae, ADAPSO, in Whelan Assoc. v. Jaslow Dental Lab. Inc. [1986] USCA3 976; 797 F.2d 1222 (3d Cir. 1986): "It is not at the coding phase where the primary creativity is expressed in the writing of a computer program."
[34] Thomas Hemnes, Novelty, Scope and the Shared Geometry of Patent and Copyright Protection, at 22, paper delivered to the Computer Law Association at Boston, November 5, 1990. Compare John S. Wiley Copyright at the School of Patent 58 U Chi L Rev 119, 182 (1991), who claims that a partial reason why patent law requires novelty and nonobviousness and copyright does not is because patent innovation is incremental but copyright authorship need not "begin with library research." Professor Wiley's excellent article has also been published in 24 Intellectual Property L. Rev. 333 (1992).
[35] For example, there is dispute over whether Newton invented the system of calculus, as some claim Gottfried Leibniz came up with the idea first. It is likely, regardless of whether Newton or Leibniz had ever studied mathematics, that someone by now would have discovered and written about calculus. However, it is highly unlikely that anyone would have written a book identical to Sir Isaac Newton's Philosophiae Naturalis Principia Mathematica (1687).
[36] For example, consider the problem of determining a flight path between Hobart and Brisbane. There are many possible routes but, due to physical constraints, only few sensible flight paths and stopover airports. Thus, there exists only a finite number of reasonably efficient solutions to this problem.
[37] "It was never the object of [the patent] laws to grant a monopoly for every trifling device, every shadow of a shade of an idea, which would naturally and spontaneously occur to any skilled mechanic or operator in the ordinary progress of manufactures." Atlantic Works v. Brady [1883] USSC 53; 107 U.S. 192, 200 (1882).
[38] Note, for example, the ongoing historical dispute over who invented the airplane, the 20 year dispute in the U.S. Patent Office over who invented the silicon chip, and even the simultaneous discoveries in the glass bottle manufacturing industries as described in Unites States v Hartford-Empire Co. 46 F.Supp 541 (N.D. Ohio 1942) modified 323 U.S. 386 (1945).
[39] Newell Response: The Models Are Broken, The Models Are Broken! 47 U.Pitt. L.Rev. 1023, 1028 (1986).
[40] Patent law requires that the prior art be examined by the Patent Office to determine whether the claimed invention is novel and nonobvious before a patent issues.
[41] Simply, novelty looks at what is known and nonobviousness looks at what the next step may be and if that step has been anticipated.
[42] Prior art is only relevant if it is exists "in a manner accessible to the public ... and open to the people of this country ... on reasonable inquiry." Galyer v Wilder 51 U.S. 476, 496 (1850). The Patent Office must search with reasonable diligence for materials that could be located by a person ordinarily skilled in the subject matter. Unpublished uncataloged materials do not fall within the category of prior art. In re Hall [1986] USCAFED 46; 781 F.2d 897, 900 (Fed. Cir. 1986). For computer programs, it is likely that most of the existing prior art will not be able to be found "upon reasonable inquiry" and the Patent Office will grant patents over pre-existing processes. See also Wiley supra at 141-44.
[43] Against Software Patents supra at 2-3 ("[M]any commonly-used software techniques do not appear in the scientific literature of computer science. Some seemed too obvious to publish while others seemed insufficiently general; some were open secrets.") See also Hulbert Special Considerations for Obtaining and Litigating Software Patents, 4 Software L. J. 1, 13 (1990) who states that it may be difficult to mount a defense for lack of novelty or for obviousness as "so many of the previously created programs (prior art) may be undocumented, stored only on disk"; Bender & Barkume Patents for Software-Related Inventions 5 Software L. J. 279, 290 (1992) ("most of the better software is either used internally, or marketed under a confidentiality provision which transmits to the user only object code... In either event, any inventive concepts in the software do not find there way into any printed publications. Second, ... because software patents have begun to issue in large numbers only in the past few years, the database of patents in the computer program technology is scanty.")
[44] "The Patent Office refused until recently to hire Computer Science graduates as examiners, and in any case does not offer competitive salaries for the field." Against Software Patents at 2 At present, most Examiner's dealing with software related inventions are trainees.
[45] Antton & Feirshans Is the United States Automating a Patent Registration System for Software? 72 JPTOS 894 (1990)
[46] Note that issued patents are presummed valid, they can be proven invalid (in a court action) by clear and convincing evidence. Electronic bulletin boards exist in the U.S. to enable defendants in a patent infringement suit to communicate with computer scientists to uncover relevant prior art, not considered by the Patent Office, that may be helpful in having a software patent declared invalid.
[47] Against Software Patents at 3
[48] Cf Walter Defining the Scope of Software Protection: Copyright Protection for Maximum Public Benefit 14 Rutgers J Computers, Tech. & L. 1, 84 (1988) ("...software develops incrementally; subsequent generations of computer programs are usually obvious and seldom based on novel processes.")
[49] Hulbert, supra at 4, claims that under the current law, a patent "may reach far beyond the particular software patented and may relate to other methods that may be useful on far different types of software." It is argued by people opposed to patents for software that such a patent should not have been granted in the first place as in all likelihood it is not a new process but itself borrowed from elsewhere.
[50] For example, the use of a backing store to store the contents of hidden windows was first developed at MIT, on a Lisp Machine. "The Lisp Machine was the first computer to use this technique only because it had a larger memory than earlier machines that had window systems.": Against Software Patents at 4 (1989 edition).
[51] For example, the U.S. Patent Office has asked software developers to send to it prior art. The Software Patent Institute is presently establishing a "folklaw" database which it will make available to the U.S. Patent Office and other interested parties. For information via internet, contact "spi@iti.org". The Software Patent Institute also gives lectures at the U.S. Patent Office on various aspects of software history and current software technologies.
[52] Gemignami Legal Protection for Software 7 Rutgers J. Computers, Tech. & L. 269, 309 (1980).
[53] Id. at 311.
[54] It is not only difficult to quantify the extent to which the patent system encourages innovation that would not otherwise occur, it is a formidable challenge to distinguish the software industry from an unknown general position. Cf Priest What Economists Can Tell Lawyers About Intellectual Property 8 Res. L. & Econ 19 (1986).
[55] However, the holder of a patent may license the invention. This may be an incentive in itself to create software or to acquire patents on software that others have been induced to create.
[56] In 1972, the Supreme Court considered the U.S. Patent Office's claim that there was sufficient growth in the industry without the need for patent protection, see Gottschalk v Benson [1972] USSC 223; 409 U.S. 63, 72 (1972). See also the 1966 Report of President's Commission on the Patent System at 13 (quoted in Benson at 72) which recommended denying patentability to software, as without the patent system, the software industry was "doing well." Most recommendations of the Committee were ignored by Congress.
[57] The pace of new development made inventions obsolete quickly, and so it was thought any patent eventually granted would provide no economic return. However, the first software related patent was issued, it seems, in 1968. See U.S. Patent No. 3,380,029 entitled "Sorting System".
[58] "Much software innovation comes from programmers solving problems while developing software, not from projects whose specific purpose is to make inventions and obtain patents. In other words, these innovations are byproducts of software development.": Against Software Patents at 9 (1989 edition).
[59] "Companies no longer wait for patent authorities to award them unchallengeable first crack at a market. Competition decides who gets the lion's share of the market. Patents, as they arrive, are swapped for royalties or other patents. Instead of being the arbiter of competitive position, patents are becoming just another tradable commodity, like bonds or baseball cards. ... Given today's shortening product cycles, the ability to create a steady flow of unique, innovative products is far more profitable than trying to stake a claim to any single idea." The Economist The Point of Patents September 15, 1990 at 19-20 (U.S. ed). The article gives the example of Mr Gilbert Hyatt, a self-employed inventor, whose patent claim on the microprocessor took 20 years from invention to be granted. If a company waited on such a patent, it would be "irredeemably behind."
[60] Cf Breyer The Uneasy Case for Copyright: A Study of Copyright in Books, Photocopies, and Computer Programs 84 Harv. L. Rev. 281 (1970). Even if a software product is imitated and marketed speedily by rival firms without the research overheads, competition in that market may be weak so that prices are not driven below a level where development costs can not be recovered.
[61] Against Software Patents at 4.
[62] A computer program may still be complex and require a large amount of time testing to see if it performs correctly. An industrial product, with a similar number of components, it is suggested, will be far more complex to design and test. Id at 5.
[63] For example, the British Technology Group claims to be "the world's leading technology transfer organization, licensing and financing products worth over £600 million in annual sales." Universities, polytechnics and government research establishments in the UK are the BTG's most important inventive sources: see firm brochure entitled "The World's Leading Technology Transfer Organization", undated. The academic may invent regardless of incentives, but the financier will not risk the possibility of imitation by rival firms when deciding whether to invest in a new project.
Compare the view of Michiyuki Uenohara, executive adviser to NEC Corp., who said "If patent law suits continue and worsen, no one will invest in new technology. While the invention is important and needs to be rewarded, far greater investments are usually made to manufacture and distribute a product." New York Times, Japanese Fight Back On Patent Claims September 5, 1992.
[64] See e.g. Swinson Copyright or Patent or Both: An Algorithmic Approach to the Protection of Computer Software 5 Harv. J. L. & Tech. 145, 212 (Fall, 1991). Recently, however, copyright protection for computer software in the U.S. has contracted, making patent protection more important and decreasing the force of any argument that copyright provides sufficient protection for the investment in software. See McCabe & Tanenbaum Copyright Decisions Increase the Value of Patent Protection for Computer Software Journal of Proprietary Rights (February 1993).
[65] A recent survey of software developers concerning patent and copyright law found that 56% of respondents agreed that patents should be granted to new and innovative software programs, although only 4% of respondents owned patents. Samuels & Cao Survey of the Opinion of Software Development Companies Concerning Intellectual Property Protection 32 IDEA, J. L. & Tech. 343 (1991).
[66] Against Software Patents at 6. See also Gottschalk v Benson [1972] USSC 223; 409 U.S. 63, 72 (1972) (citing report of the President's Commission on the Patent System which stated that software should not be patented due to the lack a classification system for software). A detailed classification system for software (in U.S. Class 395) was implemented in 1992.
[67] Sueing on a patent known to be invalid or asserting that the patent is wider than it actually is may violate trade practices and antiitrust laws. See, Handguards v. Ethicon [1979] USCA9 789; 601 F.2d 986 (9th Cir. 1979), cert. denied 444 U.S. 1025 (1980).
[68] For example, a court in 1965 complained "we cannot read [the patents] intelligently....Moreover, we have great difficulty understanding, even in a general way, the technical testimony and the discussion of that testimony by counsel...": Nyyssonen v Bendix Corp [1965] USCA1 33; 342 F.2d 531, 533-534 (1965).
[69] If the corporation, with knowledge of anothers patent, made, used or sold infringing products, it may be liable for treble damages for willful infringement. A legal opinion by patent counsel, inter alia, can often be used as a defence to allegations of willful infringement.
[70] "If a future program infringes on twenty such patents - and this is not unlikely, given the complexity of computer programs and the broad applicability of many patents - the combined royalties could exceed 100% of the sales price. (In practice, just a few patents can make a program unprofitable.)" Against Software Patents. A similar problem arises for the creators of multimedia works with respect to the number of copyright licences needed before a multimedia work can be created and sold.
[71] A further method of reducing competition is the threat of high damages in patent suits, which some authorities have claimed "makes the patent system a 'tool of extortion'" with claims up to $3.3 billion. Fitzpatrick & Fischer Patent Damages, Electronic and Computer Patent Law 737, 760 (Practicing Law Inst. Course Handbook Series No. 292, 1990). With regard to determining damages, the authors state "software patent claims typically must include limitations and/or steps in addition to the algorithm in order to claim patentable subject matter ... it is possible to imagine a royalty base for a patented software program including a computer selling at tens or hundreds of times the price of the program." Id. at 759.
[72] During the 1960's, fewer than 40% of U.S. patents were upheld by the courts. See Kayton, The Crisis of Law in Patents, Table A-2 (1970). The U.S. Patent Office "often seems to resolve doubts about patentability in favor of issuance", especially in close cases so that the Examiner's decision is not appealed. Areeda & Kaplow Antitrust Analysis, Problems, Text, Cases 186(d) (1988). However, these criticisms apply to the patent system generally, and not just in relation to computer software patents. Moreover, a recent study of software patents found that, out of 31 lawsuits, the patents "were not found invalid in over 70% of the resolved lawsuits." Lennon & Kappel A Statistical Analysis of the Enforcement of United States Patents Relating to Computer Software 8 Comp. L. Assoc. Bulletin 3 (1993) or 8 Tolley's Computer Law & Practice 174 (1992).
[73] Worse still, the big firms may use the threat of the (possibly invalid) patent to close down or purchase a rivals business. See e.g. Hartford-Empire v. U.S. 323 U.S. 386 (1944), and discussion of the case in Areeda & Kaplow supra at 295-296. (However, the engineers that formed the Hartford Company did so only to profit from the patent system and were not interested in glass manufacture - an example of the patent system encouraging manufacture.) Cf U. S. v General Electric Co. [1926] USSC 211; 272 US 476 (1926) (Holder of a valid patent, subject to continuous legal challenges and infringements, decided the easiest way to control the industry was to license all infringers). See also Areeda & Kaplow supra at 428 n.2.
[74] However, some software companies prefer not to patent new techniques, as it hinders the development of beneficial open industry standards. See e.g., New York Times Compaq and Microsoft Alliance April 14, 1993 at D6 (The CEO of Compaq "said that in creating new standards, Compaq did not expect to reap huge rewards by accumulating patents or licenses that might discourage other companies from following its lead, and buying related Microsoft products.").
[75] Ginsburg Creation and Commercial Value: Copyright Protection of Works of Information 90 Colum. L. Rev. 1865, 1924-1927 (1990).
[76] United States v United Shoe Machinery Corp. 110 F. Supp 295, 333 (D. Mass. 1953) aff'd per curiam 347 U.S. 521 (1954). The court, however, in compelling the defendant to grant any applicant a nonexclusive license at a "reasonable royalty" under any patent now or subsequently acquired from a nonemployee, stated that the defendant was "not being punished for abusive practices respecting patents."
[77] See, e.g, Unites States v United States Gypsum [1951] USSC 3; 340 U.S. 76 (1950), Transparent-Wrap Mach. Corp. v Stokes & Smith Co. [1947] USSC 38; 329 U.S. 637 (1947), United States v G.E. 82 F.Supp 753 (D.N.J. 1949), but see SCM Corp. v Xerox Corp. [1981] USCA2 200; 645 F.2d 1195 (2d Cir. 1981) cert. denied 455 U.S. 1016 (1982); Kaplow The Patent-Antitrust Intersection: A Reappraisal 97 Harv. L. Rev. 1813 (1984).
[78] See e.g., Rutter The Great Patent Plague Forbes ASAP March 29, 1993. (But "some of the biggest proponents of intellectual property were some of the best rip-off artists of the past.")
[79] In fact, most new useful inventions are based on products produced from original inventions, and not the original invention itself. Newell Response: The Models Are Broken, The Models Are Broken! 47 U.Pitt. L.Rev. 1023, 1034 (1986).
[80] E.g., a spelling check program that is used in conjunction with a word processor that modifies the screen display of the word processor by adding another pull-down menu.
[81] Cf Artic Int'l Inc v Midway Mfg. Co [1983] USCA7 257; 704 F.2d 1009 (7th Cir. 1983) (involving a speed up kit for a video game).
[82] Compare Spakovsky et al The Limited Patenting of Computer Software: A Proposed Statutory Approach 16 Cumb. L. Rev. 27, 44-45 (1986) (The authors claim that patenting software would encourage mediocrity. The first new nonobvious program performing a particular function would be patentable, no matter how efficiently it ran.)
[83] Stern Copyright Infringement by Add-on Software, 31 Jursimetrics J. 205, 212 (1991). The author concludes that "society gains more from rewarding the subsequent entrepreneur than the original one." The original entrepreneur did not foresee or market the new technology and its possible existence was not an incentive to produce to that person. The subsequent entrepreneur will have little incentive to develop an improvement if the law could be used to shut him down. See also Landes and Posner An Economic Analysis of Copyright Law 18 J. Legal Stud. 325, 332 (1989)(more extensive copyright protection would raise cost of creating new works and reduce number of works created.)
[84] Commission of European Communities Green Paper on Copyright at 172, para 5.2.9, June 7, 1988 (the real profit for the software house is in adding value to the original software by adapting it for each user). See generally, Merges & Nelson On the Complex Economics of Patent Scope 90 Colum. L. Rev. 839 (1990).
[85] Against Software Patents at 9. The League For Programming Freedom submits that if "it is ever shown that software patents are beneficial in exceptional cases, the law can be changed again at that time - if it is important enough. There is no reason to continue the present catastrophic situation until that day." The League would only allow patents for "implementations in the form of hard-to-design hardware" but not implementations of patented processes in software. Simply, the distinction they propose is between algorithms implemented in hardware (patentable) and algorithms implemented in software (unpatentable).
[86] Newell supra at 1031, after stating that all inventions may occur by the invention of algorithms, concludes "If methods and processes over large technological domains become an exercise in algorithms, then it would be extraordinarily dangerous not to patent algorithms."
AustLII:
Copyright Policy
|
Disclaimers
|
Privacy Policy
|
Feedback
URL: http://www.austlii.edu.au/au/journals/JlLawInfoSci/1993/9.html