This article was published in the Communications of the ACM, June, 1992
Jealousy and Envy deny the merit or the novelty of your invention; but Vanity, when the novelty and merit are established, claims it for its own... One would not therefore, of all faculties, or qualities of the mind, wish for a friend, or a child, that he should have that of invention. For his attempts to benefit mankind in that way, however well imagined, if they do not succeed, expose him, though very unjustly, to general ridicule and contempt; and if they do succeed, to envy, robbery, and abuse.
-Ben Franklin, 1755
All Nature is but Art, unknown to thee;
All Chance, Direction, which thou canst not see;
All Discord, Harmony not understood;
All partial Evil, universal Good;
And, spite of Pride, in erring Reason's spite,
One truth is clear, Whatever is, is Right.
-Alexander Pope, Essay on Man
The issue of software patentability is an important topic because it affects the environment in which programmers and designers work, software innovation, the health of the software industry, and U.S. competitiveness. This paper while motivated by "Against Software Patents," by the League for Programming Freedom in Communications (Jan 1992), is an overall defense of software patents.
U.S. Patent 4,736,308, the first patent under the heading "Absurd Patents" in "Against Software Patents" is described:"For example, Apple was sued because the HyperCard program allegedly violates patent number 4,736,308 a patent that covers displaying portions of two or more strings together on the screen, effectively scrolling with multiple subwindows. Scrolling and subwindows are well-known techniques, but combining them is apparently illegal." The League calls this an "outrageous result." Based on this description alone, any reasonable person would have to agree.
But I am that inventor and Apple was actually sued on a prior related patent, 4,486,857. Because my patents were misrepresented, I researched the other patents described in the League's article and am reporting my results.
There is much the League did not say about my patent and the circumstances surrounding it. First, it did not describe my background. In 1963 I worked on the software for the first computer designed to be a timesharing computer. I was at Xerox PARC in its early days, wrote two articles for the CACM[17,20] and a book on user interface design.[16] My patent covers a commercial product called Zoomracks [19] which introduced a new computer metaphor called the card and rack metaphor. Although there is a MS/DOS version, Zoomracks was marketed primarily on the Atari ST. Zoomracks developed a strong base of users who used it for a very broad range of applications, but it was a financial struggle largely because Atari did poorly. In August 1987, Apple Computer introduced HyperCard which is based on a similar, but more limited card and stack version of the metaphor.
I was then faced with having invested six years of raising money, developing a product, marketing it, and proving its value in the market, only to find I was in debt, my customer base was on a dying computer and Apple was giving away free a more polished and featured, although less elegant, version of the metaphor. While Apple may not have set out to rip Zoomracks off, it was aware of it (having seen it under non-disclosure), of HyperCard's similarity to Zoomracks, and Zoomracks' patents.
HyperCard created expectations that Zoomracks could not meet, and other companies began to develop HyperCard clones. Meanwhile, I asserted my rights, sued and settled with Apple licensing the patents. Apple is to be applauded for respecting my patents.
IBM was less respectful: We had twice brought our patent to its attention with respect to products like HyperCard and we had visibly asserted our patents and sued and settled with Apple by the time IBM decided to bundle what many consider to be a HyperCard clone. I spent six months patiently trying to deal with IBM. Finally IBM representatives flew to San Francisco to show us prior art-earlier technology-invalidating our patents that they claimed to have. When they arrived, they refused to show us the prior art, "for fear the patent office would recertify our patents in error." Even if IBM had been straightforward with me during the six months, to accept such an assertion without evidence would have been naive.
Faced with a choice of accepting IBM's offer of 0.2% of the $5 million IBM is said to have paid to license the token ring patent, or to accept its challenge to "sue us" if we wanted to see the prior art, IBM left me no choice but to fight. But I have chosen to fight in the court of public opinion where possible, rather than the civil courts where, because of its financial strength, IBM has the advantage. I added a detailed description of my dealings with IBM to my book[16] and sent copies of my book to the members of the Commission on Patent Reform when they asked for comments.
Based on my experience I formulated Heckel's Principle of Dealing with Big Companies: There is no such thing as a free lunch; unless you're the lunch.
With Apple and IBM, I did battle against large companies who were sophisticated about intellectual property, rather than small ones that were not. I felt it was in everyone's interest to force companies and the courts to make decisions about software patents so the rules and the marketplace realities can be clear to all. not just the sophisticated few. This article is written in that same spirit. While it is a personal issue, I write to clarify the software patent issues in general, to raise he level of discussion and because like most good inventors, I am curious about what the truth is.
One can only understand the need for patents in the light of a competitive marketplace. We need a heavy to show what the innovator faces as Humphrey Bogart needed Sidney Greenstreet in The Maltese Falcon. IBM has already presented itself in that role; it will reappear as did Sidney Greenstreet.
We will visit the other eight patents mentioned by the League in its article and show that the patents it selected, on examination, disprove its case.But first, we take the broader view.
Should software be patentable like other technologies? The primary issue is a policy one and so we have been influenced by Neustadt and May and their book on governmental decision-making[33]. In the background of our analysis are these questions: What analogies (to software) exist? What are the similarities and differences? What are the assumptions, explicit and hidden? "What is known?" "What is the history of the issues?" What are the interests of the various players?" We will follow the Goldberg rule and ask, not "What is the problem?" but "What is the story?" Most important, we ask, "How did things turn out in the past?"
History and innovation economics, more than law and computer science, must be the foundation on which to make policy. We have framed 10 points which are, we believe, the consensus of informed opinion on software patents. We hope they help you crystallize your thoughts on patents and enable you to better articulate your differences, if any, with the informed opinion.
Most of the arguments against software patents turn out to be arguments against patents per se. These arguments are advanced most credibly on the basis of established technologies where data and research already exist.
Patents have been accepted around the world as promoting innovation. Many giants of U.S. industry such as G.E., AT&T, Polaroid, Xerox and Hewlett-Packard, started as small companies that used patent protection to protect their inventions.
Yet, most of the articles on patents in the trade, business and even academic press read by the computer community [5, 13, 26,27,28,41,42,48] have a an antisoftware patent bias. The reason is that for every patent there is one patentholder who is reluctant to speak because the issue is complex and what someone says could be used against them in litigation. And there are a dozen who might like to use the patented technology without paying for it and so are willing to malign the patent and patent system and pass on unsubstantiated rumors and misinformation
Economists have researched innovation in other technologies [24, 31,43] and found the following: patents encourage innovation; and small entities-individual inventors and small companies-are a very important source of innovation. According to Jewkes et al. [24],
It is almost impossible to conceive of any existing social institution so faulty in so many ways. It survives only because there seems to be nothing better. And yet for the individual inventor or the small producer struggling to market a new idea, the patent right is crucially important. It is the only resource he possesses and, fragile and precarious as his rights may be, without them he would have nothing by which to establish a claim to a reward for his work. The sale of his ideas directly or the raising of capital for exploiting the ideas would be hopeless without the patent.
While several articles discuss software patent and copyrights [8,46,47], few have been written for the software, as opposed to the legal, community [11,16,37]. Such studies have only rarely been linked to software[7], and, we are unaware of any empirical studies of the effect of software patents on innovation other than this one.
If we are to reject patents in principle, we should argue that case. If we accept patents as promoting innovation elsewhere but not in software then we should differentiate software from other technologies.
The courts have developed a patent jurisprudence as a unifying mechanism to support many technologies and foster evolutionary improvement while balancing the rights of patentholders and potential infringers.
Patents have a long history (see sidebar) , Most of the concerns about patents raised by the League have been raised long ago in the context of other technologies and addressed in case law and legislation and have stood the test of time. The patent system, like MS/DOS, is not perfect. MS/DOS has a long history of evolutionary improvement: it is a derivative of CP/M, which is a derivative of TOPS-20, which is a derivative of the SDS-940 timesharing system which evolved from the first timesharing system developed at BBN about 1960. Patent jurisprudence has a similar history of evolution.
Part of the value of patents is they are a proven, public domain standard of intellectual property protection that have a history of evolutionary improvement over 500 years, compared to the 30 or 40 years of experience developing operating systems. TopView and OS/2, demonstrate how developing a new operating system and crystallizing a new infrastructure around it are fraught with dangers known and unknown. An infrastructure has crystallized around MS/DOS . It includes developers and consultants who know it, books explaining its use, and commercial products based on it. Similarly, an infrastructure has crystallized around that patent system. It includes patent lawyers, case law examples of valid and invalid, infringed and not infringed patents, and books and articles explaining patents to both lawyers and non-lawyers.
There are problems with the patent system. Only that which is not real is perfect. The patent community and the Patent and Trademark Office (PTO) are aware of the problems and have been working to address them. A Commission on Patent Reform is considering improvements. The problems that the patent community sees are these: (a) the difficulty and expense of prior art searches; (b) patents being issued with inadequate examination; (c) imprecise claims making it difficult to know whether infringement has occurred; and (d) the length of time the examination process takes. (This hurts both the patentholder who can't enforce unissued patents and potential infringers who won't know about them. Patents are published automatically 18 months after filing in all other countries.) None of these problems are unique to software.
If lack of perfection was a reason to get rid of something, no teenager would survive to adulthood. Other industries find patents useful in spite of these problems; software will too.
Patents, it is said, inhibit standards. They do not; they inhibit the expropriate of intellectual property without just compensation in violation of the Fifth Amendment. Where patents exist standards are created two ways:
Arguments that software is different should be treated critically; you can be sure those same arguments will be used by those who do not believe that the protections of the Bill of Rights extend to areas where computers and software are used.
Fred Brooks, following Aristotle, suggested the distinction between essence and accident[3], and that distinction has guided our analysis in the distinctions raised by the League and the academics (see sidebar). The question is whether the differences between software are essential or accidental in their encouraging innovation. The League says software is different and should be protected differently. They present two arguments.
Why, so they are; but so are airplanes, silicon chips, silicon chip fabrication plants, potato chip plants, oil refineries and many things. But people find the patent system beneficial in these other technologies.
Other industries have cottage manufacturers and they deal with patents. Outside of software much invention is a cottage industry; about 5000 independent inventors belong to the 37 organizations that are members of the National Congress of Independent Inventors. And most cottage industries don't rely on invention.
We should no more optimize an intellectual property system for cottage developers then we should for Fortune 500 companies.
When one talks about marketing and maintaining commercial software products, the costs are much greater than the estimates made by the League. At the other extreme, IBM is reported to have spent 2.5 billion dollars to develop OS/2, including applications.
It is expensive to develop software if the task is to design it from scratch and make it a success in the market; it is not, if the task is to clone something that already exists or is precisely specified.
Indeed, that clone software is so much cheaper to develop argues for the necessity of patent protection if one wants to stimulate the development of products worth cloning.
Making software non-patentable or subjecting it to a different form of protection creates practical difficulties, rather like a state seceding from the Union and setting up checkpoints on its border. And if one state secedes, they all can. If each technology has a sui generis (unique to itself ) form of protection, we would have to set up boundaries between the different technologies and would need rules for what happens at the boundaries.
This situation occurs in software development. Should programmers be able to define their own conventions or should they conform to the system conventions even where they are not optimal? Do new programmers get to define their own conventions just because they were not involved in the original decision? Aren't programmers expected to abide by the conventions so the code will integrate better and others can maintain it later. Of course, as problems surface, it is foolish to resist all change in conventions just because changes have repercussions. Changes are made, but as part of a deliberative process in which the burden of proof is on those who advocate the changes.
The evolution of the law works the same way; computer law is just another subsystem to be integrated into the fabric of jurisprudence.
The problem of having different conventions in different areas is demonstrated by the Cadtrak patent which is discussed later. It is a hardware, rather than a software, patent. It requires a display device but no software to infringe it. A computer can be designed so it does not infringe, although a simple program running on it can. This demonstrates that simple software programs can infringe almost pure hardware patents and suggests the difficulty of drawing a legal distinction between hardware and software.
In the legal debate (see sidebar) women are playing important roles. Susan Nycum, the founding chair of the ACM Section for Legal Issues advocated in 1974 that software be patentable.[34]She recommended I get patents in 1981 and suggested Hugh Finley, another long-time ACM member, as a patent lawyer. She now heads the computer law group at, Baker and MacKenzie. This view is now the mainstream view taken now by the practicing and academic patent bar and most computer lawyers. The contrarians who argue against software having the same breadth of protection as other technologies are led by Pamela Samuelson. Unless overruled by the Supreme Court., the issues will be decided by the Court of Appeals for the Federal Circuit whose chief judge is the Honorable Helen W. Nies
The League for Programming Freedom has launched an offense in the debate. Like the Battle of the Bulge, up close it looks formidable. But, while it must be treated seriously, it is the last gasp of a dying cause.
The PTO and those patent lawyers who prosecute software patents have much more experience in the nitty gritty of protecting software than the contrarian academics. And the PTO and the courts have more experience weaving new technologies into the fabric of the patent system than the software community has in creating forms of intellectual property protection.
The pioneers in each new technology see that technology as new, different, and central, and expect the world to accommodate it. To some extent the world does. But slowly each new technology becomes woven into the tapestry of knowledge encompassing other technologies-each distinctive in its picture-but using the same threads and the same weave.
This is the paradox that one must confront if one argues software patents decrease innovation. The essential difference between Marxism and Capitalism is property rights. Patents and copyrights create intellectual property rights which can be brought, sold, rented and licensed like other property rights. Marxism may be better than Capitalism in some areasÐcertainly not in Russia, but Capitalism, with all its flaws, has outperformed Marxism.
The paradox of Marxism is not just a theoretical issue. Stallman, the founder of the League for Programming Freedom, heads the Free Software Foundation which is developing and planning to distribute a clone of the Unix¨ operating system. AT&T has invested in Unix based on its ownership as manifest in patents and copyrights. AT&T can't be pleased when Mr. Stallman comes along and gives away free copies of a clone of a product it invested millions in developing and marketing.
If AT&T had not used patents and user interface copyrights to protect its intellectual property rights, Mr. Stallman would have no trouble making and distributing a Unix clone. But AT&T must pay its bills with money it receives from customers and has asserted its rights. If it is acceptable to clone Unix or any program, will anyone invest in new ideas? Should we optimize an intellectual property jurisprudence for, not large entities, not small entities, but companies that distribute free clones of other people's software?
For all his talk about wanting to promote innovation, Mr. Stallman seems to get his ideas for technology from AT&T, 1969 and his ideas for intellectual property protection from IBM, 1965.
Many software developers do it for the fun of it. But the distinction is based, not on the technology, but on amateurism: amateurs flourish in the early stages of a new technology. From the time the world's oldest profession began, professionals have accepted that others give it away, but they bristle if they are expected to work at the same rates.
Software patents have unique problems: prior art libraries are limited, the search classification system was designed for hardware patents, few computer scientists are examiners. Still, when it gets to specific cases computer scientists and the PTO see invention similarly. (See Document Comparison)
For the last two years the PTO has been improving the situation. It is improving its prior art search facilities in software, has published a new software classification system, and is actively recruiting computer scientists.
The PTO has still not been able to rid itself of the prejudice against software patents as patent practitioners in the software area will tell you. It still is conservative in its interpretation of what constitutes patentable subject matter and has rejected several applications that that are being appealed.
Prior to 1982, about 30 different software related patent cases went through the Appellate Courts. The range of technologies-seismic, medical, petrochemical, telecommunications, firmware, and software-demonstrate that software is both well grounded in patent law, and basic to the advancement of American Industry. Software has become pervasive in industry that it has been basing business decisions on software's being patentable for 10 to 20 years. This has created a sophisticated broad-based constituency for keeping software patentable. Congress has not given in to demands to make less pervasive technologies, such as biotechnology, unpatentable; it is less likely to do so with software. Software has been clearly patentable longer than it has been copyrightable (See Patents and Copyrights).
Chisum, the leading authority on patents, wrote an article on the patentability of software and concluded:
The continuing confusion over the patentability of computer programming ideas can be laid on the doorsteps of a single Supreme Court decision, Gottschalk v. Benson, which held that mathematical algorithms cannot be patented, no matter how new and useful. A careful analysis of that decision shows the holding is not supported by any of the authorities on which it relied, that the Court misunderstood the nature of the subject matter before it, and that the Court failed to offer any viable policy justification for excluding by judicial fiat mathematical algorithms from the patent system. The Benson decision is inconsistent with the later Supreme Court ruling in Diamond v. Chakabarty that the patent system applies impartially to new technologies and that any policy issues for excluding new technologies should be addressed to Congress. Policy considerations indicate that patent protection is appropriate for mathematical algorithms that are useful for computer programming as for other technological innovations. [8]
Chisum is in the mainstream in saying the courts made a mistake by making software unpatentable. But courts are reluctant to overturn previous decisions directly, and then only after their scope has been eroded. A similar situation where prejudice had become part of the jurisprudence occurred earlier: Plessy v. Furgesson(1896) held that "separate but equal" facilities for whites and blacks were lawful. The courts did not directly overturn it, but eroded its vitality on a case- by-case basis over a period of years in a number of decisions starting with Murrey v. Maryland while appearing to show respect for Plessy. Finally, when faced with Brown v. Board of Education (1954), ample precedent had been created for the Supreme Court to overrule Plessy directly.
Samuelson, having asked Professor Chisum to write his article, now attempted to refute him, but after arguing for over 100 pages that the basis for software patentability is weak, was forced to conclude that:
...the only principle which seems to have guided the court's decisions is one of upholding the patentability of as many program-related inventions as possible while appearing to show respect for the Supreme Court's decisions. [39]
Samuelson's observation seems to be compelling evidence that while she has not been persuaded that Benson, like Plessy, is a fundamental error in giving prejudice the force of law, the Court has and will, in due course, reverse it. The original intent of Congress will again become law, and statutory subject matter will "include anything under the sun that is made by man."
The software community will be best served by articles about how to avoid infringement, how to deal with infringement notices, how to find prior art, how to use patents to protect new ideas, how to differentiate products, and how to make the patent system work better for software (based on experience rather than speculation). In brief, we should direct our energies towards making the system work so as to increase innovation and U. S. competitiveness, rather than fighting patents.
Patents, like a cat's claws. are weapons to punish when necessary. A declawed cat will not survive long in the wild; nor can a defenseless startup once it succeeds and attracts substantial competitors. Patents are not the only defense, but they are a vital one to innovative startups that must survive in a real world. In business, as in the jungle, respect is given only to those who can protect themselves.
Microsoft, IBM and others are applying for patents in quantity. Those who don't understand the situation, are not. Many are happy to have software patents attacked. Why let your competitor in on a good thing? Why pioneer new product ideas when it is less risky to copy competing products and incorporate useful features once market success is proven.
From the perspective of large companies, a loud voice, like the League, yelling against software patents can be useful, as a meat packer finds a Judas goat useful in leading sheep to slaughter.
The Japanese are aggressively filing for U.S. patents on software. While our strength is innovation, Japan's is in adapting innovations and steady improvement. If they have the improvement patents and we did not file for the basic patents, we lose. If we arrogantly dismiss the Japanese as incapable of creating good software or cavalierly dismiss patents as undesirable, then 20 years from now we will be trying to get back the software market from Japan just as today we are trying to get back the automobile and semiconductor markets. We aren't even trying to get back the consumer electronics market.
Who is responsible for the misperception about the desirability and legality of software patents? In a certain sense, it is the League for Programming Freedom. But it knows not what it does. And its arguments are the ghosts of arguments for IBM's corporate self-interest of a bygone era. Is not the origin of the problem IBM's attempt in the 1960's to declaw a competing technology by depriving its practitioners of their constitutional rights as inventors? (See Software Patents: IBM's Role in History.)
If it was just a question of IBM outfoxing its competitors, we might learn our lesson and let it pass. But we think it useful to ask some questions: Is it in the interest of the United States to have strong, competitive, innovative software industry? Is it in IBM's interest? Did IBM use its position on the 1966 Patent Commission to put its corporate self interest ahead of that of the United States? Should IBM be held responsible for its role in creating the current software patent mess?
Some have proposed making software patents unenforceable. Might a law making IBM patents unenforceable make more sense? Or a law that would prevent IBM from obtaining patents for a period of time, say 5 or 10 years? At a time when when competitiveness with Japan is a major concern, what kind of a message should we send about what happens to those who use their positions on government commissions to sacrifice their country's interest to their corporate self interest?
Similarly, if we should eliminate patents to avoid patent litigation as the League suggests, should we not eliminate all laws so as to avoid all litigation?
If we are to have a meaningful debate on whether software should be patentable, I suggest we take our standards, both of debate and of where the burden of proof lies, from Abraham Lincoln:
I do not mean to say we are bound to follow implicitly in whatever our fathers did. To do so would be to discard all the lights of current experience-to reject all progress-all improvement. ... if we would supplant the opinions and policy of our fathers in any case, we should do so upon evidence so conclusive, and argument so clear, that even their great authority, fairly considered and weighed, cannot stand...
If any man [believes something], he is right to say so, and to enforce his position by all truthful evidence and fair argument which he can. But he has no right to mislead others, who have less access to history, and less leisure to study it, into [a] false belief...thus substituting falsehood and deception for truthful evidence and fair argument.
What I find most frustrating in this debate is that the mode of argument used against software patents by so many [15,26,27,28] is to throw as much mud against the wall as possible and hope some of it will stick. I have expended some effort here removing some of the mud. I don't claim to have removed it all, but I hope that I have wiped away enough to show you that the rest will wash off too.
In its article, the League lists nine patents, mine and eight others to make its case. It is unlikely that the members of the League considered the positive side of any of the patents they cited. It is as if they went searching for quarters with heads showing, and finding several, reported their findings without turning any of them over. Here we turn over the other eight quarters in an attempt to produce some empirical results.
U.S. Patent 4,197,590: The inventor founded a company to develop and market what appears to be the first personal computer to write directly from memory to the display. This invention has been widely licensed to the personal computer industry by Cadtrak. The "XOR" is only part of the invention. Cadtrak filed and has won at least one lawsuit against a larger company. The idea behind the "XOR" claims of this patent is simple. A program XORs a cursor icon onto a display device; later a second XOR to the same place erases the cursor restoring the original display. To move the cursor one XORs the cursor to its old location, then XORs it onto the new location. There are many ways to get around this patent. One can use an underline as a cursor or "logically or" the cursor onto the display, erasing later by rewriting the display with its original information. This approach is fast, lets you change cursor icons easily, accesses the minimum possible data, and requires no space be reserved on the screen for the cursor.
The League says this patent can be infringed in "a few lines of a program." It can be, but not on a computer that was commercially available at the time the invention was made. The invention is largely the invention of the frame buffer. As such, it requires hardware which has since become common, making it possible to infringe the XOR claims with a few lines of code. Many, if not most, computer manufacturers including Apple and IBM have taken out licenses which covers programs running on their computers.
This patent illustrates that it is usually easy to design around a patent one accidentally infringes. If this patent, a hardware patent, is a "bad" patent as some claim, it only demonstrates that the electronics industry tolerates "bad" patents because it finds patents beneficial on balance. Software should be able to tolerate "bad" patents similarly. To discard the patent system because some bad patents exist would be the same as suppressing free speech to stamp out lies.
U.S. Patent 4,398,249: This is what the League mischaracterized as the "Refac recalculation patent." In 1970 Rene Pardo and Remy Landau invented the concept of an array of formulas which would enable businessmen to write their own programs to create business applications. Although the word was not used, their invention is in essence the modern computer spreadsheet. That the claims cover recalculation is an artifact of how the patent claims were written. Pardo and Landau marketed a commercial spreadsheet-like product based on this technology. This invention has been widely adapted in the personal computer industry-over 250 spreadsheets have been marketed.
This patent was originally rejected by the Patent and Trademark Office (PTO) as a mathematical algorithm and thus unpatentable subject matter. Pardo and Landau felt so strongly about their inventive contribution that they appealed their case pro se, which means they, not a lawyer, wrote the brief and argued it before the appeals court. The decision, In re Pardo[23] is a major legal precedent which establishes that an invention is patentable whether or not the novelty involves software. If their experience was typical, they were stonewalled when they tried to enforce their patents. This would explain why they approached Refac a white knight in the fight against the patent pirates. If Pardo and Landau have the same deal Refac offered others, then they can expect to collect royalties only as Refac does.
U.S. Patent 4,633,416: This patent is held by Quantel, a company that developed a line of commercial video editing products protected by its patents. Quantel filed for patents when it was small. and it has grown from being small to large because it has used its patents to prevent competitors from using its technology. The first company Quantel sued was much larger than it was.
U.S. Patent 4,777,596: The League tells us XyQuest was notified that their product XyWrite infringed Productivity Software patent protecting the ability to accept an abbreviation or correct a spelling error by hitting a space bar. When licensing negotiations failed, XyQuest removed the feature from future releases.
Productivity Software was founded in 1984 to develop data input systems where minimal keystroke data input is important. Based on its patented technology, Productivity Software has grown to seven employees and markets 31 specialty products. It has found niches in the medical and legal transcription and the handicapped marketplaces.
Patent problems are generally minor compared to the other problems pioneering companies face. At about the same time XyQuest had postponed introduction of its latest version for about a year so it could upgrade to IBM standards as part of an agreement where IBM would market XyWrite exclusively. At the last minute IBM reneged on the deal[25].
While the first five patents are held by small entities, the last four patents are held by large entities and they also protected commercial products.
U.S. Patent 4,558,302: Unisys licenses this, the LZW compression patent, for 1 percent of sales. It has threatened a large entity with a lawsuit, but no small ones.
U.S. Patent 4,555,775: The League describes AT&T's backup store patent as "Too Obvious to Publish." Yet, in a letter in this issue of CACM, Dennis Richie points out that this technology was published in the ACM[35] and was recently called "a seminal paper" whose ideas are seen in X Windows, Macintosh and many other windows systems [14]. While AT&T has sent notification letters on this patent, it has put the patent into reexamination and has not threatened suit or sued anyone on this patent.
U.S. Patent 4,656,583: This is an IBM patent on compiler speedup.
U.S. Patent 4,742,450: This is an IBM-shared copy on write patent. These two patents are what IBM calls Group 1 patents whose royalty is 1% of sales. They have been licensed by IBM as part of general licensing agreements but have not been licensed individually. (About 50 of IBM patents are Group 2 patents. Group 2 patents can be licensed for 2% each; the entire Group 1 portfolio, for 2%; and the entire IBM patent portfolio for 5%.)
These two patents have not been litigated and I do not believe that IBM has aggressively asserted these patents against anyone. IBM, like most companies, normally files for patents only to protect what they expect to become commercial products. We treat these patents as protecting commercial products.
Most patents are never asserted. Much of the value of patents, like that of the Swiss Army, is that they act as a deterrent. The patents described here are typical of the small number where the patentholder forces a resolution: the infringer may take a license, designs around the patent, or produce prior art showing there was no infringement.
Many letters asserting patents are no trespassing signs, putting potential infringers on notice should they infringe or telling them not to. They require no action. The notified companies might send prior art back to the patentholder, who might send it to the PTO for reexamination. The "infringer" may ignore the notice, waiting to see the reexamined patent or for the patentholder to become more assertive. The resolution may be hidden, in that an infringer may design around the patent. Rarely, a product is withdrawn from the market. The statistics on the cited patents are summarized in Table 1.
Whether any patent including those described here, is valid and infringed is a complex legal and technical question. An advantage of the patent system is that the question is an objective one based on the patent, prior art, and the "infringing" device. Such a dispute is less acrimonious than one where the task is to evaluate testimony where one person yells "thief," and the other "liar." Whether infringement actually occurred in any of the cases is irrelevant. The relevant question is did the original patentholders bring commercial products to market based on the patented technology and motivated by the rights a patent bestows?
The 9 patents cited by the League summarized in Table 1 lead us to these conclusions:
All nine patents protected commercial products.
Four of the nine patents were from startups funded to exploit the patented technology. A fifth filed for its patent in its seventh year. All five companies struggled for years.
The technology pioneered by at least three of the small patentholders was significant in that they started new product categories or were widely adopted in the industry.
Only Quantel seems to be unwilling to license its patent.
The only patent dispute between similar sized companies (XyQuest) was settled readily. No small entities were faced with a lawsuit brought by a large entity without the advantage of the patentholder having settled earlier with a large infringer.
The only disputes where a small entity paid patent royalties or was sued was where the patentholder had previously settled disputes with larger companies. No case was cited where a big company aggressively went after a small one over patents unless a large company had respected the patents first. The only big company this author knows about that has gone after small ones is IBM (see sidebar).
Four of the five small entities had large entities use their technology without first licensing it. All four were forced to sue. This makes it unlikely that all the patent disputes were an honest difference of opinion although some likely were. It is for this reason "piracy" seems a fair characterization. These same small entities have had their patents mischaracterized and their motives impugned in the academic, trade and business press read by the software community.
It can cost over a million dollars to litigate a patent through to trial. The data shows that large entities are quick to use their power to try to intimidate small ones into abandoning their rights or accepting nuisance settlements rather than address infringement issues on their merits. It appears that this high rate of patent piracy is caused in part by the Federal Rules of Civil Procedure which tilt the scales of justice against the weak.
Our results confirm the League's suggestion that big companies will readily bully small ones but refutes its suggestion that a patentholder who asserts a patent will get showered with gold. The yellow matter is not gold.
The Japanese adapt innovations from sources outside the company twice as fast as United States companies[31]. The technology protected by at least two of the patents (CadTrak and HyperRacks) was exposed to companies that later became the first infringers.
Japan's ability to accommodate outside [the firm] innovation may be one of the reasons it has been so successful in dominating markets. If the U.S. is to exploit its strengths in innovation, it must learn to adapt outside innovations without the inefficiency of legal confrontation. Fast and efficient patent enforcement should encourage U.S. companies to license outside technology early on rather than waiting until they have an infringing product in the market and face substantial legal exposure.
If large companies are forced to deal with a infringement issues early, they might see it is in their advantage to work with the inventors to use their knowledge. Now, the legal system keeps the patentholder and infringer at war until such time as the patentholders knowledge is of little value to the "infringer," thus wasting one of our most valuable resources: the creativity and experience of innovators.
The only example that was cited where someone faced infringement issues from more than one patentholder seems to be XWindows facing the CadTrak and AT&T patents, but this has not been resolved and no lawsuits seem to have been filed or threatened.
We think the League is right in alleging that bad patents have been issued. The League, however, fails to identify a patent that was rejected by the courts. We think this is because issues of prior art and patent invalidity are considered early on in the patent assertion process. Patentholders rarely continue to assert patents in the face of solid evidence of invalidity or non-infringement.
Product development effort seems to have focused on creating many versions of an invention once its value was proven. Over 250 different spreadsheets and at least four products generally considered to be HyperCard clones were marketed.
The innovations protected by small entity patents listed here seem to have been more widely adopted than those of big companies in their industries. Big companies are better at commercializing and protecting their minor innovations, than their major ones.
That small entities seem to introduce the more fundamental innovations to the market is telling. Big companies are often unsuccessful in transforming innovations into commercial success: Xerox PARC pioneered much of modern day personal computer and its software. Although IBM invented a predecessor to the spreadsheet (expired U. S. Patent 3,610,902), it did not market a commercial product based on it, nor did it assert the patent even though its claims seem to read on (i.e. be infringed by) modern spreadsheets. These technologies became major product categories primarily through the efforts of small entities.
Table 2 shows a rough estimate of the efficacy of three major sources of innovation: federally funded, large entity, and small entity. Our results show that small entities are 7.5 times as cost-effective at stimulating innovation as large ones, and 200,000 times as cost-effective as federal funding. The U.S. grants patent rights to universities as part of its research contracts thus patents are issued in all these areas and patents asserted is a reasonable measure of innovation. We believe that a more scientific study would refine these results, but doubt it would change the basic conclusion.
As a software developer, you might review the patents discussed and put yourself in the place of each of the parties involved. If your product finds satisfied users, do you think better financed companies with stronger marketing organizations will market competitive products using your innovations? If so, will patents be useful to you? If a patent is enforced against you, do you think you will be able to design around it? If you have to license it, do you think your competitors will also have to license it, thus passing the cost on to the end customers? Which problem would you rather have: a big company entering a market you developed, or finding out you were accidentally infringing a patent? Do you think the effect of software patents might be more innovation, higher software prices and an industry with more long term profitability?
If you are protected by patents, your success depends in part on your patented inventions as others must deal with them. If you accidentally infringe a patent, designing around it is within your expertise. If you do not have patents, success depends much more on the ability to finance and market products-capabilities outside of your expertise and control. If you are a software developer, don't patents benefit you by manifesting your contributions in rights you can bring to the bargaining table, while confining the problems largely to your area of expertise and control?
Issues in Science and Technology (Winter 1991/2) contains a letter from Commissioner of Patents Harry F. Manbeck who said of another article by the same authors that they[15]:
... demonstrate they do not understand the current law...Most of their statements ... do not appear to be the result of a balanced and reasoned inquiry and do not appear to be supported by the facts. ... they cavalierly dismiss the view of those who appear to have used the patent system successfully and impugn their motives...
The PTO issued about 89,000 patents in 1990 from which the League, with the advantage of hindsight, can pick and choose the ones to attack. Consider the information presented here on patents the League selected to demonstrate the PTO's mistakes. Whose standards are higher, the League's or the PTO's?
After reviewing our results we can make some general recommendations.
This is consistent with the results described here. The operational implications are to continue to let the system operate as it is accepting evolutionary changes based on experience rather than speculation.
PTO fees should be reduced, especially for smaller entities, and the PTO should receive a higher level of funding to improve its ability to examine patents so it can issue better quality, more timely, patents in software and other technologies. European patent offices are much better equipped and much better funded. It seems that the PTO should compete with the NSF and other organizations for federal funding on the basis of their cost effectiveness in encouraging innovation.
In 1990 only $2 million of the PTO budget of $419 million came from federal funding; the rest came from user fees. Superficially, it might seem that investing in the patent system will have a multiplier effect of 80,000 in creating innovation as compared to federally funded science. We suggest no such thing. We do however ask the question: If taxpayers were to spend on additional $160 million per year to support innovation, we could either increase the $64 billion federal funding on science by one-fourth of one percent (0.25%) or increase PTO funding by 40% enabling the PTO to issue better patents and restore reduced user fees for small entities. Which will likely produce a more innovation? Which will achieve a greater multiplier effect by encouraging additional private investment?
An example of federally funded science is fusion power research which has been going on for at least 25 years, has cost hundreds of millions of dollars and has produced little practical result. Pons and Fleishman developed (and filed for patents on) cold fusion without government funding yet they, having invested their own money and not being in the mainline of governmental funding, are heavily criticized. While it is not clear that Pons and Fleishman have produced cold fusion, respected people in the field believe that they have, even if no one yet understands what is happening. This is just an example of how the system is biased in favor of governmental funding of expensive conventional solutions, and against individuals and small companies who risk their own time and money to innovate.
Individuals taking a contrarian view have been the major source of new ideas in both science[38] and engineering[24]. This is why small entities and the patent system is so important. Most will fail, but the successes more than make up for the failures
It would be interesting to evaluate the results of federally funded science to determine which projects are worth the cost. Some projects seem to have become like those welfare mothers, who generation after generation, are entrapped in a governmental support system.
The data shows that it is commonplace for large companies to pirate the technology of small entities. No case was cited where a large company licensed a small entity's technology without first being sued suggesting that the existing laws do not motivate large companies to resolve patent disputes with small companies quickly. The issue here is not just fairness to inventors and improved efficiency in settling disputes, it has to do with avoiding the waste that occurs because U.S. companies are so much slower at adopting new innovations than Japanese companies.
Congress responded with anti-piracy legislation where software copyrights are concerned; we would hope it would similarly pass legislation to prevent patent piracy. Remedies similar to the criminal penalties for copyright infringement and Rule 11 sanctions for attorneys who file frivolous suits are worth considering. We also suggest the following as possible remedies for patents disputes to stimulate discussion:
These suggestions which should induce speedier resolution of patent disputes, like all our recommendations, are suggested for all patents disputes, not just software.
The patent system, an enormously productive system for inducing innovation, is being stymied by a cumbersome dispute resolution process. Is it in the public good to have a system of conflict resolution which discourages conflict resolution? Should innovators spend their time innovating or litigating?
If the courts could resolve software patent and copyrights issues more quickly, it would clarify the law so everyone can make decisions with some predictability. The problem are not unique to patents but occurs in all litigation. That the judicial and even the legal, community are beginning to address the inefficiency of dispute resolution and litigation is grounds for cautious optimism.
We are keenly aware that the sample is small and unscientific and so our results should be considered suggestive rather than definitive. A more definitive study should be useful in bringing out facts that would be useful in evaluating future changes to the patent law.
The recommendations can be summarized as: Redress the balance of incentives so innovators will prefer to develop their ideas commercially, using patent protection, rather than search for federal funding.
The software industry is getting more competitive. Almost every company that has hit products uses its cash flow to develop entries in other product categories. As a result, product categories are getting very competitive. Since most software companies have confined their intellectual property to source code copyrights, trademarks and, sometimes, user interface copyrights, whenever they come up with a successful innovation, their competitors will often quickly replicate it. As a result, the impetus is toward similarly featured products competing on price, differing only in the mistakes which the originators must maintain to support their existing customers.
Now companies are recognizing that by using patents, they can compete on features and functionÑnot just tactically, but strategically. Even if competitors do replicate the features they will likely make them different enough to avoid infringement. Companies following this approach will support standards, but their products will have a substantial proprietary component engendering products with more diverse feature sets. This will enable the industry to compete more on the profitable playing field of unique capabilities and market position and less on price. This is consistent with standard business school product marketing, where product differentiation and market segmentation are basic.
Intellectual property has already driven the market for those who got in early and established standards. Lotus owns the 1-2-3 standard. Novell owns a major network standard and Wordperfect a major word processor standard. Apple owns the Macintosh User interface standard and Intel and Microsoft own the IBM compatibility standard. Patents give companies the opportunity to establish and own something of value in the market based on their innovativeness, rather than their marketing and financial capabilities.
While the problem of people accidentally infringing software patents has been greatly exaggerated, several patents will be successfully asserted against existing products. This will be primarily between those companies that focused on innovation and have patents, and those that focused on exploiting recognized business opportunities. This kind of confrontation happened earlier on in the aircraft and other industries[21].
During these confrontations, the businesses with a large volume of infringing product will, if they are not patent aware, understandably feel "extorted". Such businesspeople will take support for their position from those who argue against software patents and advocate or suggest invalidating existing software patents[15,39].
The software innovators who advanced the technology and made business decisions based on their patent rights will similarly feel cheated especially where they pioneered commercial products based on their inventions. When depositors made decisions based on government guarantees of S&L deposits, no one suggested that the government default on their obligations to insured depositors as people suggest the government invalidate existing software patents. No one vilifies the S & L depositors because the government has to pay them money, yet software innovators find themselves vilified with lies and half truths.
In this confrontation, both sides start out feeling cheated.
Many "infringers" will react emotionally and view it as a problem to be gotten rid of and many will fight to the bitter end. This raises the stakes, as a company having been put on notice may be liable for treble damages and attorney's fees. Patentholders will not likely pursue these cases for four or five years while the infringer's liability builds up. After a suit is filed these companies will be getting much of their advice form those who have most to profit from the litigation: their litigators. This seems to be what is happening with Lotus.
Some software developers on finding out that the rules were not what they thought, face the problems of infringing others' patents while not having patented their own successful innovations. Some will chalk it up as one of many risks and uncertainties of business. Those who react emotionally might find it useful to first ask: Which of the players have acting in good faith? Which have not? Which have been responsible for the patent mess? Which have been innocent victims? Having answered these questions, such developers can more effectively target their wrath.
Companies that act rationally will analyze the patents to ascertain their scope and validity, whether infringement is occurring, and how easy it is to remove the "infringing" capability. The will check with other licensees. They will consider the obvious options, such as taking out a license, removing the infringement, finding prior art and showing it to the patentholder, or fighting in court if that is the only possibility. They will likely try to address the problem early on before the liability builds and consider designing around the patent, negotiating a license or using some form of alternative dispute resolution to resolve infringement and validity issues.
If the problem is associated with purchased product, most companies will stand in back of their products and provide a license, warrant it against infringement, or provide guidelines on how to avoid infringement. It will put its "no problem" in writing.
Astute companies will view infringement as an opportunity in disguise. If the patent is good and competitors are, or soon will be, infringing it, the first licensee can generally get an inexpensive license forcing competitors to pay more if they want the technology. It may be possible to get an exclusive license on some feature which differentiates a product from the competitor's . It can be worthwhile to see if the patent covers useful capability which could be added to the product. The best time to negotiate for the patented technology is when you are not incurring liability of infringement but rather can offer to create a demand for that technology by incorporating it into a product. Invention being the mother of necessity, competitors will be faced with the choice of paying a higher price to license patented technology or leaving it out thus differentiating your product from theirs.
In brief, what superficially looks like another problem to be dealt with in the increasingly competitive, commodities oriented software business, will prove to be what makes products less price competitive. Many industries have worked on this basis all along: patents make industries more diverse in their offerings, more profitable, more innovative, and ultimately will make the U.S. more competitive.
The essence of this article is simple: Software intellectual property issues are not inherently different in substance from other technologies; what motivates people is not inherently different; industry life cycle is not inherently different; marketing and business strategies and tactics are not inherently different; the law and policy issues are not inherently different. The technology is not even new. Software has been around for 40 years. The issues may be new to those who had no experience of them, but the only thing that is different is that software is a mass market industry for the first time and real money is at stake.
I would like to thank Steve Lundberg, John P. Sumner, Susan Nycum, Lewis Gable, George Gates Dave Pressman and Tom Hassing for their many useful comments.
2. Axelrod, R. The Evolution of Cooperation, Basic Books. 1985.
3. Brooks, F. No silver bullet: essence and accidents of software engineering, Computer (Apr. 1987).
4. Bruce, R. Lincoln and the Tools of War. University of Illinois Press, 1989.
5. Bulkeley, W. Will software patents cramp creativity? Wall Street J. (March 14, 1989).
7. CBEMA, Comments on computer-related invention patents, Computer Lawyer (Oct 1991).
8. Chisum, D. The patentability of algorithms, 47 U Pitt. L Review, 959,971, (1986).IMPORTANT
10. Choate, P. Agents of Influence, Touchstone, 1990.
11. Clapes, A. Software Copyright, & Competition, Quorom, 1989.
12. Cringely, R. X, Accidental Empires., Addison Wesley, 1991.
14. Foley, et al. Computer Graphics: Principles and Practice,, Second ed. Addison Wesley, 1990.
17. Heckel, P. Isolating differences between files, Commun. ACM (Apr. 1978).
19. Heckel, P. Zoomracks, designing a new software metaphor, Dr. Dobbs J., (Nov 1985).
20, Heckel, P., and Lampson, B. A Terminal oriented communications system Commun. ACM , (July 1977).
22. Howard, F., WIlbur and Orville, Knopf, 1988
23. In re Pardo, 684 F.2d 912 (C.C.P. A. 1982)
24. Jewkes, J., Sawers, D., and Stillerman, R. The Sources of Invention, Second ed., Norton, 1969. IMPORTANT
25. Judas, J. B., Innovation, a Casualty at IBM, Wall Street J., (Oct.17, 1991), A-23.
26. Kahin, B, The software patent crisis, Tech Rev. (Apr. 1990) 543-58
27. League for Programming Freedom, Against software patents, Commun. ACM (Jan. 1992)
28. League for Programming Freedom, Software patents, Dr. Dobbs Jo., (Nov. 1990)
29. Lewis, T. Empire of the Air. HarperCollins, 1991.
30. Lincoln A. Selected Speeches and Writings, Vintage, 1992.
31. Mansfield, E. Industrial innovation in Japan and the United States, Science, (Sept. 30, 1988).
32. McCloskey, M. Intuitive Physics, Sc. Am, (Apr. 1983), 123.
34. Nycum, S.. Legal protection for computer programs, Comput Law J. 1, (1978).
35. Pike, R, Graphics in overlapped bitmap layers, ACM Trans. Graph. 17.3 (July 1983), 331
36. Pressman, David, Patent It Yourself, Nolo Press, Third Ed., 1992.
37. Ritter, T. The politics of software patents, Midnight Eng. (May-June 1991).
38. Root-Bernstein, R. Discovering, Harvard University Press, 1989. IMPORTANT
41. Samuelson, P. Should program algorithms be patented? Commun. ACM, (Aug. 1990).
42. Samuelson, P., and Glushko R., Survey on the look and feel lawsuit,' Commun. ACM, (May 1990).
43. Schon, D., Technology and Change, Delacorte, 1967.
45. Schwartz, E. The coming Showdown over software patents, Business Week, (May 13, 1991).
47. Sumner, J and Lundberg, S. Software patents: Are they here to stay? Comput. Lawyer, (Oct. 1991).
48. Slutsker, G. Whose invention is it anyway? Forbes Magazine. (Aug. 19, 1991).