Chapter 13: Sui Generis Protection of Software

Prefatory Note

Beginning in the late 1960s, Elmer Galbi of IBM urged development of a sui generis form of legal protection for computer software, based on modification of the patent system. See, e.g., Galbi, Proposal for New Legislation to Protect Computer Programming, 17 J. Copr. Soc'y 280 (1970). The World Intellectual Property Organization (WIPO) proposed a copyright-like model software protection law in the early 1970s, building on Galbi's ideas. See Model Provisions on the Protection of Computer Software (WIPO pub. 814(E), 1978). The WIPO model law was not implemented and WIPO eventually withdrew it in favor of urging expansion of copyright protection for software.

Also in the early 1970s, after the Supreme Court's decision in Benson, IBM (one of the amicus curiae in opposition to Benson/Bell Labs in the Benson case and, in this regard, then an ally of the Antitrust Division in the Algorithm Wars) proposed a sui generis software protection law, which Representative Fish introduced in the House. The Antitrust Division, however, asserted that the legislation had many technical flaws and opposed IBM's suggested compromise. Apparently, it saw no need for compromise while it was the beneficiary of a string of appellate victories in Benson, Johnston, and Flook. No further action on the Fish Bill occurred. For subsequent advocacy of sui generis software protection during this period, based on perceived inappropriateness of traditional forms of intellectual property protection, see R.H. Stern, The Case of the Purloined Object Code: Can It Be Solved? (Part II), BYTE, Oct. 1982 at 210, 222.

In the early 1980s, the Japanese Ministry of International Trade and Industry (MITI), concerned about very expansive interpretations of copyright law that some U.S. courts were adopting (so-called look and feel protection), conducted a study. The report of its study group recommended that Japan adopt a sui generis form of protection for computer software. IBM and other U.S. companies then favoring copyright protection of software strongly opposed this approach. They successfully urged the U.S. Commerce Department and Special Trade Representative to protest to the Japanese Government against the proposals of MITI and its study group. The Japanese Government then withdrew support for a sui generis law and shifted control of software regulation from MITI to the Cultural Affairs Agency, which had supported an extension of Japanese copyright law to cover software. See Dennis S. Karjala, Lessons From the Computer Software Protection Debate in Japan, 1984 Ariz. St. L.J. 53.

More recently, four commentators published a “manifesto” in which they argued that copyright and patent law are “ill–suited to protecting software innovation” and urge that an additional sui generis intellectual property paradigm be developed, based on the doctrine of misappropriation (see INS v. AP, ch. 1, above). P. Samuelson, R. Davis, M. Kapor, and J. Reichman, A Manifesto Concerning the Legal Protection of Computer Programs, 94 Colum. L. Rev. 2308 (1994).

They observe that patents cover only particular methods of achieving useful results rather than the results themselves. Yet, it is possible “to produce functionally indistinguishable program behaviors” by use of different methods. Hence, a patent on one method of generating certain results cannot prevent a competing rival's use of another method, and thus escaping the reach of the innovator's patent. This is said to be a defect in patents that makes them fail to “protect the primary entity of value in software,” which is behavior or function.

Copyright is “mismatched to software” because it does not focus on behavior, the principal source of value in a computer program. Protecting the expression of a computer program fails to keep second comers from copying the useful behavior of the computer program and thus taking a free ride on its creativity. This “sharply reduces incentives for innovation, and thus thwarts the policy behind legal protection.”

Accordingly, the Manifesto commentators propose development of a law covering misappropriation of software behavior, which would provide cumulative legal protection to supplement existing copyright and patent protection. The Manifesto commentators do not provide any draft legislation, however, to implement their ideas. It is therefore somewhat unclear just what they propose should be done and correspondingly difficult to evaluate their proposals. See Jane C. Ginsberg, Four Reasons and a Paradox: The Manifest Superiority of Copyright Over Sui Generis Protection of Computer Software, 94 Colum. L. Rev. 2559, 2556 (1994). See also Paul Goldstein, Comments on a Manifesto Concerning the Legal Protection of Computer Programs, 94 Colum. L. Rev. 2573 (1994)(arguing that Manifesto needs “more hard work” and suggesting that “[i]t would be a useful exercise to draft a fully integrated bill — specifying subject matter, rights, and remedies — that incorporate these trajectories, not as a proposal in itself, but as a way of showing how these principles would work in concrete form.”).

Chip Act Paradigm for Protection of New Technology
Robert W. Kastenmeier, [Then] Chairman, House Judiciary Committee’s Subcommittee on Intellectual Property
Foreword to R.H. Stern, Semiconductor Chip Protection (1986)

The enactment of the Semiconductor Chip Protection Act of 1984 was a signal event. ln point of historical fact, proprietary protection for semiconductor chip products is the first significant intellectual property right created in nearly 100 years. Passage of the Act also marked the first time that the Congress enacted a sui generis intellectual property law specifically directed to the characteristics of a new technology. The Act thus avoids “distortion by shoehorn” (to borrow John Hersey's phrase) that may result when new technologies are forced into 19th Century forms of intellectual property that were originally devised for artists or authors, completely foreign to the subject matter at hand. The Semiconductor Chip Protection Act is our first break with that past tradition, an important precedent for Congressional consideration of the industrial property needs of new technologies.

One of the most important lessons of the Act is that Congress can constitutionally rise to the challenge of designing laws to meet technological changes. ln enacting a public law, Congress crafts a solution with words. As in any exercise with words, there is an inevitable degree of uncertainty in the resulting product; there is always some subjectivity in interpreting language.

We have a common law system in the United States. We do not pretend to the completeness of civil law countries in their codes. There will always be debates on the meaning of our laws, therefore, by commentators, lawyers, and judges. The interstices will have to be filled. Indeed the House Report expressly states that the Congress intends that the courts will “develop a new body of law specifically applicable to semiconductor chip infringement.” The courts will borrow from copyright and patent precedents where appropriate and they will fashion new law when neither of those bodies of traditional intellectual property law is appropriate to the needs of the subject matter. But in doing so, the courts must be faithful to the intent of Congress in passing the new statute.

One of the fruits that readers and users of this book will hopefully harvest is a sense of the same intellectual stimulation that I have enjoyed over several years in studying these issues and crafting a new form of industrial property law. The problem that the Congress faced in drafting this legislation was how best to weigh equities between the public interest and proprietary rights. We had to remember that the primary objective of intellectual property law is not to reward the author, but rather to secure the general benefits derived by the public from the labors of authors. We had to devise a way to compromise among competing interests, maximizing them individually and in the aggregate as best we could. Of course, the interests are incommensurate in that we lack any common or quantified standard for their measure. But the art of legislation inevitably involves the balancing of competing concerns, and it requires doing so within a practical time frame. That is part of the challenge of the legislative process.

In studying the problem of how best to protect semiconductor chip products and in devising a legislative scheme that solves the problem and also promotes the public interest, I found the study and the solution to be a paradigm of the industrial property protection problem for all new technology at the end of the 20th Century and at the rise of what may be a new information society. One of the things that I believe we learned was that the era of “shoehorning,” or of pouring new wine into old legislative bottles, should end. We learned, as the author puts it in chapter 11 of this book, that different bodies of intellectual property law strike different respective equilibrium points for balancing the values and interests at stake, that what is an acceptable or desirable balance of interests for authors and artists is not necessarily acceptable in the case of industrial property products, and that “it would be pure serendipity for a law designed to deal with literary and artistic rights to realize the needs of new technology.” We are no longer so blindly self–confident as to expect that such a serendipitous result will automatically occur; we know that we will have to devise the desired result through our own efforts, by having hearings at which knowledgeable persons inform us about technological change, by extensive study of current law, and by painstaking statutory drafting. I believe that our experience under this new law will instruct all of us in learning how better to meet the needs of other new technologies that will require industrial property protection to flourish and thereby improve the quality of our lives.

Should We Consider Another Model for
Industrial Property Rights in Software?

Address By Richard H. Stern Before
The Technology Workshop on Computer Software Protection
The Geo. Wash. Univ. School of Engineering and Applied Science
Washington, D.C. – Oct. 1, 1993

Another version of this paper was published as A Sui Generis Utility Model Law as an Alternative Legal Model for Protecting Software, 1 U. Balt. Intell. Prop. L.J. 108 (1993). The present version of this paper has been abridged and its footnotes have been omitted.

The conventional approach to legal protection of intellectual property in the United States has relied almost entirely on two models — those of patent law and copyright law. These two models of legal protection have severe limitations, however, particularly in regard to late 20th century computer software technology. Systems based on these models are particularly unsuited to protecting non–code aspects of computer software — to protecting against non–verbatim, non–literal copying of computer programs — the subject on which my remarks will focus.

Moreover, the recent trend of appellate decisions in software cases has been in the direction of limiting legal protection to code and literal aspects of computer programs. Accordingly, the time may now be ripe for consideration of other legal models for protecting software, at least abstract, non–literal, non–code aspects of software.

In the remarks that follow, I will summarize the characteristics of the patent and copyright models of legal protection and their difficulties when applied to software. Then I will discuss possible characteristics of an alternative model of legal protection for non–code aspects of software, point out areas of difficulty, and propose further study and discussion of those areas.

The Two Principal Models

The Patent Model

The patent law model for industrial property rights contemplates allowing legal protection only for innovations that have a high level of technical merit or technological advance: Creative work substantially above the routine, advances in technology that persons of merely ordinary skill in the field would not have conceived and reduced to working form.

To determine whether a product that is a candidate for patent protection meets the requirements for protection, a patent system relies on examination by technical experts before any legal protection becomes effective. This is an expensive and time-consuming procedure, both for the applicant and the government. But the filtration procedure is considered worth the high front-end costs, for several reasons. One reason is that it relieves the courts of the burden of making such an assessment in the first instance. That is a task for which they are not well equipped. Another reason is that competitors, potential investors, and the general public have greater assurance that issued patents cover true inventions and thus valid, enforceable pieces of intellectual property.

Patent law does not protect ideas, as such. In principle, the system of protection is limited to machines and other products implementing or carrying out a novel idea. Thus, mathematical principles cannot be monopolized under a patent. By the same token, algorithms and computer programs, as such, cannot be patented. Generally speaking, the less a technological advance is tied to specific machinery the less likely it is that the advance will fit under the patent system. Hence, sweeping advances that cut across the borders of traditional technologies are likely to fall outside the scope of the protection that the patent laws offer.

Finally, and particularly important for present purposes, a patentee is entitled to withhold the use of the patented invention from others, entirely or selectively, and to set whatever price for permission to use the invention that the patentee decides to charge. Moreover, a patentee is ordinarily entitled not only to damages for patent infringement, but also an injunction to prevent any future use of the invention without consent. Subject to very narrow qualification, the United States model of patent protection (unlike that of other countries) does not permit compulsory licensing or anything like it.

The Copyright Model

The copyright law model of legal protection requires only a de minimis level of merit or creativity — little more than failure wholly to plagiarize the work from another author. Accordingly, there is no need for examination of novelty and technical merit. Thus no filtering by technical experts occurs before copyright protection attaches to a work. Ordinarily, the first time that the creative level of a work is examined is in the course of a copyright infringement suit before a federal court. The result is lower front-end cost for copyright protection to attach, but possibly higher administrative costs in litigation, if that ever occurs.

A copyright owner has the right, typically for a 75-year term, to exclude others from reproducing, importing, and distributing the copyrighted work. Copyright law provides no general right to prevent use of the subject matter. Thus use of infringing software (for example, execution of an illegally copied program) is not copyright infringement unless the use involves the making of a copy.

Like a patent, a copyright confers an absolute property right, in the sense that a copyright owner may at will withhold from others, entirely or selectively, the right to reproduce the work. A copyright owner may secure both damages from, and injunctions against, those who infringe the copyright.

United States law does not ordinarily permit, and the Berne Convention prohibits, compulsory licensing of copyrighted works. The Berne Convention also prohibits discrimination against particular kinds of literary work (the category in which computer programs are classified). Thus a national copyright law must not give computer programs less favored treatment than other kinds of literary work, for example, by not allowing injunctions in cases of infringement of computer program copyrights while allowing injunctions in book copyright cases.

A copyright, unlike a patent, has no claims defining its scope. Accordingly, the scope of a copyright is whatever a court ultimately holds that it is. That scope is largely unpredictable in advance.

Ordinarily, a copyright will protect against verbatim copying, such as bit–for–bit or instruction–for–instruction copying of code, but that is often not the major issue in software cases. In recent years, ingenious counsel have frequently persuaded courts to protect against non–literal, non–verbatim imitation of copyrighted works—for example, imitation of user interfaces and command languages for application programs. This has been done on the imaginative theory that computer programs and other works of new technology are equivalent to poems and novels, for which copyright law may protect plot and other non–literal aspects. Therefore, according to this line of reasoning, non–literal aspects of computer programs, such as their command languages, deserve similarly broad legal protection against imitation.

These developments have created great tension with the basic legal principle that copyright protects only expressions of ideas, and does not protect any idea, procedure, process, system, method of operation, concept, principle, or discovery. This list is what we may appropriately term the index prohibitorum of the Copyright Act, since one section of that Act states that the scope of copyright protection does not extend to the foregoing list of things.

In attempting to extend copyright to such “non–literal” protection of software, some courts have confused copyrights with patents. (The Whelan and Lotus–Paperback decisions are examples of this.) Very recent decisions have tended in the opposite direction, however, and have tended to restore the previously recognized differences between the copyright and patent models of legal protection. (Such recent decisions include those in the Galoob, Sega, and Altai cases.)

These recent decisions have also reemphasized the principle of copyright law that predominantly functional and utilitarian aspects of works — in particular, computer programs — are not protected by copyright. While the issue still remains unsettled, and courts in some parts of the country have not yet issued decisions on these issues, it seems fair to conclude that the recent trend of copyright law decisions is against protection of relatively abstract aspects of computer software and of predominantly utilitarian and functional aspects of computer software. Nevertheless, some of these aspects of software — for example, instruction sets, programming languages, and algorithms — may be the most creative, innovative, and valuable aspects of software technology. Their creation and marketing may contribute much more to technological progress in software than does creation of code itself.

Difficulties With the Two Principal Models

The difficulties with the two principal models, patents and copyrights, center on the fact that they do too little — and do too much — for software and its non–code aspects. For example, in principle neither copyrights nor patents protect ideas. Yet, some of the most creative and valuable aspects of software may be classified as ideas, or at least they operate at so high a level of abstraction that they are, for purposes of patent law, unprotected ideas or mathematical principles. For purposes of copyright law, they are unprotected ideas, procedures, processes, systems, methods of operation, concepts, principles, or discoveries. Algorithms and user interfaces are salient examples of this. So, too, are languages and instruction sets. The patent and copyright models, unless they are greatly altered, do not allow for protection of highly abstract subject matter, regardless of what may be its great economic value.

The fairly rigid requirement of the two traditional intellectual property legal models that injunctions must be issued against infringement creates problems, too, for “idea” aspects of software. Thus, a given case may present a court with a difficult either–or (or XOR) choice: Choice 1 is to order a total prohibition against any infringing use by the defendant of the plaintiff's allegedly protected software idea. Alternative choice 2 is to determine that the allegedly infringing material is legally unprotected — that it is outside the scope of the plaintiff's statutory intellectual property rights.

Faced with those polar choices and no available legal middle ground between them, courts and Congress have often preferred the legal underprotection of the second alternative to the perceived legal overprotection of the first. Presumably, they conclude that overprotection of rights would stifle the innovative efforts of other workers in the field.

Here we see the unfortunate impact on software rights of the “XOR principle” of our traditional patent and copyright legal models. The owner of intellectual property rights either gets it all or else gets nothing. There are no halfway measures, no compromises. The copyright and patent models insist on doing too much for copyright and patent owners, in some ways, if they do anything at all. The end result, however, may be that our laws give too little to encourage the creators of new software or to encourage those who invest in bringing new software to the market. This is a major defect of the traditional models.

Indeed, the principal models of intellectual property protection not only at times do too little or do too much for “idea” aspects of software, but they are often completely askew to software. Software and the intellectual property laws are often like ships passing in the night. Or perhaps you will prefer the metaphor of the Second and Ninth Circuits in the Altai and Sega cases: using one of them for software is like “forcing a square peg into a round hole.”

There are other models for industrial property protection, however. Copyright and patent law do not exhaust the known legal schemes for protection of intangible rights in intellectual creations. Other legal models may help us overcome the difficulties and problems inherent in the traditional copyright and patent models.

Another Industrial Property Model

Other legal models for protection of industrial property rights are known in this and other countries. One alternative model to consider for industrial property protection of software is that of the utility model. This model is known in many European countries and Japan. A utility model is a kind of petty patent. It typically has narrower scope than a regular patent. It may be allowable upon mere registration with a government agency. Its validity is sustained on the basis of a lesser showing of technical merit than a regular patent. And it may have a shorter life span. The U.S. Semiconductor Chip Protection Act of 1984 — which protects integrated circuit chip layouts — is basically a utility model law. The chip law represents the first major departure in the United States from the patent and copyright models of intellectual property law. As many others have suggested, some of this law's provisions may suggest a more appropriate legal scheme for protection of computer software than the patent and copyright models do.

Using utility-model laws as an alternative model for industrial property protection suggests that some of the following features may be appropriate for a software protection law directed to non–code aspects of computer software and non–verbatim copying of computer programs. The following sketch of characteristics of an industrial property law for protecting rights in future software technology is tentative. I will often use the word “probably.” Where I do not, you should understand it as implicit. This is a matter calling for study, discussion, identification of the interests at stake, balancing them and making appropriate trade–offs, and a considerable number of other compromises.

At this stage, we need to raise questions, identify issues, and point to possible models. That is more appropriate now than purporting to prescribe definitive solutions.

Protected subject matter

First, something like utility–model protection could be afforded to anything in the index prohibitorum of section 102(b) of the Copyright Act. (That is the section that excludes any idea, procedure, process, system, method of operation, concept, principle, or discovery from copyright protection.) By the same token, patent law's exclusion of laws of nature and mathematical principles need not be a limitation on property rights in a utility model. Congress is free to protect any of these things under a different law, if Congress considers that wise. Of course, common sense and considerations of policy may call for some of the same limits, but that would be a matter addressed to the wisdom of the legislature, not its power.

Utility–model–like protection could be considered, for example, for algorithms, programming languages, instruction sets, icons, and various aspects of user interfaces for computer programs. More generally, utility–model–like protection could be provided for any economically valuable aspects of future software technology for which Congress made a judgment that industrial property protection was needed to encourage their creation, disclosure, and commercialization.

Nature/scope of protection

Protecting such aspects of computer software would extend legal protection to economically valuable things now left largely without any effective legal protection. That extension is probably inconsistent with adopting for software the full scope of patent and copyright relief and the whole arsenal of remedies now in use for the traditional subject matter of patent and copyright law. Indeed, insisting on protection that extensive would probably torpedo the whole project.

Many would consider that granting injunctions against use of such features of software technology was inconsistent with promotion of rapid software progress. Locking up new algorithms or instruction sets tightly for a period of years is probably a mistake. It is one thing to make users pay reasonable compensation for enjoying the use of such new ideas. But it is quite another thing to permit their creators to withhold their use at will or levy any charge that they please, with the aid of state compulsion.

The scope of software rights should probably be more like those of patent law than copyright law. The software creator's right to control use is a major legal right omitted by copyright law. Doubtless that is for good reasons in the case of copyright's traditional subject matter. But the right is economically important for software. Hence, software rights, like patent rights, should cover use of the protected subject matter, at least to the extent that use of unauthorized copies of software should be regarded as infringement. In the case of things capable of repeated and multiple use in different programs, such as algorithms and user interfaces, their use in preparing a computer program should probably be considered an act of infringement, giving rise to a right for compensation.

Conditions for awarding protection

What should the conditions or prerequisites be for entitlement to legal property rights in software features? Probably originality, and novelty relative to the prior work of others in the field, should be minimum requirements. In addition, some level of technical advance over what has gone before should be required. As in the case of the U.S. semiconductor chip law and many utility model laws, the legal model for protection of software features should probably call for a level of required technical merit somewhere below patent law's very high, inventive standard and above copyright law's very low standard.

Any in-depth evaluation of the technical merit of particular software, however, should be saved for litigation, if it ever occurs. Again, that is as the U.S. semiconductor chip law provides. A utility–model–like law for software would probably call for registration of software rights, and immediate attachment of industrial property rights, while avoiding the high front-end costs of a patent-type examination procedure. The expense of a thorough analysis of the prior work of others in the same field, and a determination whether the registrant's advance over such work was sufficient to justify legal protection, should probably be deferred until a plaintiff and a defendant are concerned enough about the matter to be willing to spend the kind of effort and money involved in a lawsuit. However, there are respectable arguments on the other side, and this is another aspect of the model that requires more discussion and study.

Modification of Existing Intellectual Property Laws, Instead

In any discussion of adopting other legal models for software, one may ask: Why depart from the copyright and/or patent law models at all? Would it not be preferable just to modify one or both of these existing schemes to put some corners in the round holes so that they will accommodate square pegs? Are they not familiar schemes with established, known bodies of case law helping us to fill in necessary omissions or ambiguities in any statute applied to software?

Congress answered many of these questions in the House Report on the Semiconductor Chip Protection Act of 1984, when the House refused to follow the Senate's copyright approach to chip protection and insisted on enactment of a sui generis chip law instead. It recognized that, to a significant extent, the Berne Convention and the existing body of United States copyright and patent law do not permit us to square off the legal hole to make the software peg fit it. Indeed, to modify copyright law and patent law sufficiently so that they accommodated software properly would mean changing them so much that they would no longer be copyright or patent law. Modifications that extensive would make us reinvent the patent and copyright laws. Not only would that be infeasible because of these laws' traditional constraints, but it would be unacceptable to the users and beneficiaries of traditional copyright and patent law. They would not tolerate the amendments or reinvention.

Finally, the concept of saving effort by relying on the existing body of copyright law is an illusion. Earlier, in sketching the outline of a software law based on the utility model, I gave a few examples of difficult choices. There are many more of these difficult choices. They should be addressed explicitly, not by pretending that some existing body of law that has never addressed such issues, or had any reason to address them, offers a ready-made solution to be incorporated by reference. The likelihood of coming to rational solutions for these problems is enhanced more by addressing them explicitly, than by hoping that buying a lottery ticket — meaning incorporating a preexisting law by reference — will lead to a serendipitous outcome. (Murphy's Law is inconsistent with free lunches.) The right way to craft an appropriate scheme of industrial property protection for software is to do it purposefully.

What Next?

Clearly, a sui generis software utility model type of law will not draft itself. Crafting it is a substantial undertaking. But the effort of undertaking study and drafting is a better choice than trying to sweep the difficult issues under the rug by purporting to incorporate some other existing body of law in its entirety.

Some personal experience may be informative here. In the early 1980s, I spent a great deal of time on behalf of the U.S. semiconductor industry attempting to persuade Congress to pass a copyright amendment to protect semiconductor manufacturers against the copying of integrated circuit chip layouts. We were successful in persuading the Senate to endorse the approach of blanket incorporation of copyright law. But Chairman Kastenmeier wisely blocked that move when the bill came to the House. His opposition to incorporating copyright law by reference for chip layouts forced a choice of drafting, instead, a so–called sui generis chip law — that is, a new, self–contained law drafted to address the specific perceived problems of the semiconductor industry. That eventually led, I believe, to a much more rational and effective chip protection law.

The chip law's lesson translates perfectly to software. Unfortunately, however, crafting a sound software protection law is probably one or two orders of magnitude more difficult than drafting a chip protection law was. That is a challenge, however, not an insurmountable obstacle.

I said earlier that crafting a sound law in this field is a substantial undertaking. A first step in this direction is elaboration of the appropriate model. That would call for much more study. Perhaps more conferences such as this one. Perhaps a series of seminars or workshops on individual issues such as the unanswered questions raised earlier.

Any proposals for legislation should follow — not precede — careful, detailed study of the issues that model–making raises. Any proposals should follow an attempt to ascertain the views of, and to build an informed consensus among, software professionals, industry, and users regarding an appropriate legal model for industrial property protection of the valuable aspects of future software technology.


The most recent trend of decision in software copyright cases has been away from the Whelan decision, away from protection of non–literal aspects of computer programs, and away from treating copyrights as if they were patents. That does not mean that industrial property protection of non–literal, relatively abstract aspects of software is a bad idea. Nor does it mean that courts should or do consider these aspects of software to be without economic value and undeserving of any kind of industrial property protection. It means only that the traditional intellectual property law models, particularly that of copyright law, do not accommodate such protection. If we believe that such protection is needed, we need to consider another legal model. That in turn would mean that legislation of some kind is needed to accord the appropriate protection.

A possible new legal model for protecting the abstract, but nonetheless valuable, aspects of late 20th century software technology is provided by utility model and petty patent laws. Such a model could select from copyright, patent, and other industrial property laws those features most suited to protection of advanced software technology. At the same time, this model of legal protection can avoid many of the significant pitfalls of U.S. patent and copyright law when they are used to regulate software proprietary rights. Given the most recent trend of court decisions in the field of software protection, the time may well be ripe now to consider this alternative legal model if and when proposing software protection legislation.

Before legislation is attempted, however, we need much more study. Perhaps more conferences such as this one, perhaps a series of seminars or workshops on the many different issues that model–making raises. I hope that this conference will be the initial step in an attempt to ascertain the views of, and to build an informed consensus among, software professionals, industry, and users regarding an appropriate legal model for industrial property protection of the valuable aspects of future software technology.

A petty patent system for algorithms and other software abstractions is proposed as a compromise to restore predictability to computer–related patent law and as a way to permit software innovators to obtain rapid, inexpensive intellectual property protection that would not preempt the innovative activity of others in the field. The subject matter of the petty patent system is algorithms and other computer–related advances that are rejected under section 101 of the patent statute as directed to nonstatutory subject matter. The same system may, however, appropriately include software abstractions or “nonliteral aspects” of computer programs that courts similarly deem unprotected under the copyright statute.

An important feature of the system is a predictable, relatively bright–line means of distinguishing claims to algorithms, as such, from claims to algorithm–using machines and algorithm–using industrial processes. The distinction must be made under any system regulating this field, for clearly it has never been a bar to patentability of a technological advance that it operates in accordance with a rational, definable scientific principle. The problem with the present system, particularly after recent patentability decisions, is lack of predictability and clarity, with consequent impairment of security of expectation. It is therefore important to provide a clear standard in the statute or equip it with a mechanism for accomplishing clarity in discrimination between algorithms, as such, and algorithm–using machines or processes. The proposal does this by first setting a standard believed to provide predictability and clarity,* and then authorizing the PTO Commissioner to issue regulations to fill definitively any interstices that may exist or develop.

  *     The central portion of the proposed statutory standard, see Algorithm Conundrum, for making this distinction is:

§ ___. Relation of software protection to patent laws

    (a) ALGORITHMS UNPATENTABLE.—When the main point or points in which an alleged invention departs from the prior art is an algorithm,...a method of doing business, or other subject matter protectable under this chapter, the sole available form of intellectual property protection, if any, for the alleged invention shall be that which this part of this title provides.


      (1) Notwithstanding subsection (a), even though an invention utilizes or carries out an algorithm, ... method of doing business, or other subject matter protectable under this chapter:

        (A) if the invention is implemented in novel and unobvious apparatus, the apparatus may be patented as provided under part II of this title;

        (B) if the invention is implemented by a novel and unobvious use of new or old apparatus, the use may be patented as provided under Part II of this title.

      (2) Novelty for purposes of Part II of this title or paragraph (1) of this subsection shall not be predicated on newly causing an otherwise conventional apparatus to operate in accordance with an algorithm, in combination with a category of printed matter, or to carry out a method of doing business....

The PTO would issue these petty patents with only negligible prior examination, which would be limited to facial compliance with statutory requirements and implementing regulations, as in the case of copyright and chip layout registrations. Therefore, petty patents would have only a slight presumption of validity. Issues of novelty and technical advance over the prior work of others in the field** would be addressed in detail only when an infringement dispute or similar controversy arose.

  **     The proposed system would make entitlement to a petty patent dependent on novelty and technical advance. The required standard of technical advance would not be as high as the standard unobviousness standard of regular patent law. 35 U.S.C. § 103. See generally the discussion of this issue in Algorithm Conundrum at 219–21.

The preferred method of resolving a controversy over novelty and technical advance of a software abstraction would be an adversary, post–issuance opposition or revocation proceeding in the PTO, rather than a determination in a district court as in patent, copyright, and chip layout infringement litigation. In this administrative proceeding, interested parties in the software industry could tell the PTO about prior knowledge and use by persons other than the petty patentee (registrant). Since the PTO is more suited to making technical determinations than courts are, one might anticipate a higher quality and less expensive way of resolving such intellectual property validity controversies under the proposed regime.

Remedies for infringement would be less preemptive, except in exceptional cases, and they would appropriately be gauged to the relative contributions of the parties to commercialization (e.g., relative enablement) Remedies must not be too preemptive, but they must be sufficient to make the election of a petty patent attractive. The ordinary remedy for infringement would be reasonable and entire compensation for the use made,*** and injunctions would be available only in exceptional circumstances.

  ***     See 28 U.S.C. § 1498, which provides for an award of reasonable and entire compensation, and permits no other remedy, when the United States infringes intellectual property rights.

The remedy mechanism should consider factors that are material in the software abstraction context, even though regular patent law does not address them. For example, compensation for competitive use of a software abstraction might take into account such possible facts as that the defendant's infringing computer program is a substantial enhancement over the plaintiff's contribution to the art; that the defendant engaged in substantial independent work to implement or commercialize the protected subject matter; that the defendant independently created the infringing subject matter and did not copy from the registrant; that the defendant in good faith, prior to the plaintiff's effective date of registration, and without derivation from the plaintiff, exploited the protected subject matter; whether the amount of technical advance in the art or technical merit of the protected subject matter was modest or substantial; whether industry compatibility requirements made it commercially impracticable not to use the protected subject matter; and that the plaintiff made excessive demands for compensation or excessive claims as to the scope of its rights. A court or other tribunal awarding compensation might be permitted to decrease a plaintiff's compensation award because of its unreasonable behavior and the defendant's reasonable behavior, or increase a plaintiff's compensation award because of the defendant's unreasonable behavior and the plaintiff's reasonable behavior.

Portions of Draft Sui Generis Petty Patent Software Statute
Excerpted from Richard H. Stern, On Defining the Concept of
Infringement of Intellectual Property Rights in Algorithms and
Other Abstract Computer-Related Ideas
, 23 AIPLA Q.J. 400 (1995)

§ 1001. Direct infringement

Subject to [later sections describing exemptions and limitations], a person infringes rights in a software abstraction protected under this Act if the person uses, or traffics in an embodiment of, the protected software abstraction within the United States, and the person knows or should know that the software abstraction is protected under this Act.

§ 1002. Indirect infringement

(a) Subject to [later sections describing exemptions and limitations], a person infringes rights in a software abstraction protected under this Act if the person, in this or a foreign country, while having the state of mind described in subsection (b) — (i) conspires with, (ii) actively induces, or (iii) intentionally commits acts, or engages in conduct, that causes — another person to use, or traffic in an embodiment of, the protected software abstraction within the United States.

(b) A person is liable for infringement under subsection (a) only if, when the person acted, the person knew or should have known that:

(c) A person trafficking in goods, services, or other things that have no substantial noninfringing use is rebuttably presumed to be committing conduct that the person knows or should know will cause a violation of section 1001. A person trafficking in goods, services, or other things that have a substantial noninfringing use is rebuttably presumed not to be committing conduct, by such trafficking, that the person knows or should know will cause a violation of section 1001.

§ 1003. Use of protected subject matter

A person uses a protected software abstraction within the meaning of section 1001 if the person:

§ 1004. Trafficking in protected subject matter

A person traffics in an embodiment of a protected software abstraction within the meaning of section 1001 if, in respect of any computer program, machine, or device that implements or embodies the protected software abstraction, or with respect to any machine or device configured in accordance with a protected data structure, the person:

§ 1005. Additional uses and traffic

A person uses, or traffics in an embodiment of, a protected software abstraction, within the meaning of section 1001, if the person commits acts or conduct that a court, in an infringement action under this Act to which the person and the owner of rights in the abstraction are each party, determines, as a matter of law:

§ 1011. Authorization and exemption, in general

Notwithstanding section 1001 through 1005, a person does not infringe rights by engaging in acts or conduct that the owner of the rights has authorized expressly or by implication.

§ 1012. Particular authorized and exempted acts

Notwithstanding sections 1001 through 1004 [the previous provisions describing infringing use and trafficking]:

§ 1013. Additional limitations on rights

A person does not commit infringement by engaging in acts or conduct that a court, in an infringement action under this Act to which the person and the owner of rights are each party, determines:

§ 1014. Ineffective notices

Notices and other writings are ineffective, under any law, that purport to declare that distribution to the public of copies of a computer program, machines, or devices embodying a protected software abstraction was not a distribution, that transactions otherwise having the characteristics of sales of such copies, machines, or devices were licenses rather than sales, or that title to such copies, machines, or devices was retained or withheld despite occurrence of transactions otherwise having the characteristics of sales of the copies, machines, or devices.

§ 1015. Effective notices of restrictions

(a) Notwithstanding section 1012(a), an owner of rights in a software abstraction may by use of adequate notice impose or cause to be imposed on a direct or indirect customer, other than a consumer end user, a reasonable limitation or restriction on the customer's trafficking in a product embodying the protected software abstraction.

(b) A person who intentionally disobeys a limitation or restriction described in subsection (a) is not authorized to traffic in the product to which the limitation or restriction applies, insofar as such traffic goes beyond such limitation or restriction. Any such unauthorized trafficking is an act of infringement in violation of section 1001.

(c) A limitation or restriction is reasonable, for purposes of this section, if and only if each of the following conditions is satisfied:

(d) A limitation or restriction imposed solely to prevent infringement of intellectual property rights is rebuttably presumed reasonable. A limitation or restriction against conduct described in subsections (b) to (f) of section 1012 [debugging, fair use, etc.] is rebuttably presumed unreasonable. A price-fixing limitation, condition, or restriction is conclusively presumed unreasonable. There is no automatic presumption of market power by reason of ownership of rights in a software abstraction.

(e) Notice is adequate for purposes of this section if the patentee gives the customer full, unambiguous, written notice of the restriction prior to the distribution to the customer of the product embodying the protected software abstraction, and gives such notice to the customer's predecessors, if any, in title to the product prior to the distribution of the product to them.

(f) This section is intended fully to occupy the field of legal regulation of the subject matter, and it shall not be contravened, modified, or supplemented by any other provision of any law.


1. Explanatory material on these sections may be found in the AIPLA Q.J. article from which this material is taken.

An earlier version of the statutory proposal, with explanatory material, was published as a four-part article in IEEE Micro (June, Aug., Oct., and Dec. 1983 issues).

2. § 1015(f) could be strengthened against potential supersession implied by subsequently enacted legislation by adding: “No subsequently enacted provision of law shall be deemed to contravene, modify, or supplement this section unless the provision clearly and expressly states that it is intended to do so.” What would be wrong with adding the quoted language without the part beginning with “unless”?

Return to Table of Contents