My name is Jim Warren. I have been working in and reporting on the computer industry for almost a quarter-century.
First, I am a member of the Board of Directors of Autodesk, Inc. -- a multi-national software company specializing in computer-aided design. As a $400-million company, we have been recently identified as the 6th- largest PC software publisher in the world. I am presenting its views and recommendations.
Second, I have been a computer professional since 1968, have founded multi-million dollar companies in Silicon Valley, and have held numerous leadership roles in personal computing, essentially since its inception in the 1970s. I am speaking from that perspective -- from which, I have also been the "futures" columnist for more than half a decade, for MicroTimes which has a 220,000 circulation. I also write monthly columns about government and computing for Government Technology magazine and for BoardWatch magazine. I was founding President of the Microcomputer Industry Trade Association, founding Chair of the First Conference on Computers, Freedom & Privacy, received the Electronic Frontier Foundation's Pioneer Award the first year it was presented and hold graduate degrees in Computer Engineering from Stanford, in Medical Information Science from the University of California Medical Center in San Francisco, and in Mathematics and Statistics from the University of Texas at Austin. I was founding Editor of microcomputing's first software periodical (Dr. Dobb's Journal), founded microcomputing's first free newspaper and first subscription newspaper -- InfoWorld, was founding host 15 years ago of television's oldest weekly computer show -- the "Computer Chronicles," and founded and chaired the world's largest public microcomputer conventions in the first decade of the industry. Although all that I "invented" were innovative, all utilized complex procedures and all were valued by those who paid millions to use what my innovative entrepreneurial risk created, it never occurred to me to patent them, and I could not have patented those "useful arts" if I had wanted to. Fortunately for us all, patent protection is not available for innovative services, as opposed to gadgets. We could all go home if you would simply declare that software is a service -- which, in many ways, it is.
But we are here, and I am speaking for Autodesk, and also as a long-time industry observer and leader. I speak as a successful innovator; but not as an intellectual-property lawyer.
My remarks are in three parts: principles, pragmatics and specific recommendations.
We all know that software is somehow different from all traditional inventions. But how does it differ from the devices that are surely what the framers of the Constitution envisioned when they mandated patent protection? The difference is that all traditional inventions enhance our physical capabilities, whereas software mimics the mind and enhances our intellectual capabilities. That is what makes software different from all patentable devices, and that is what justifies sui generis.
Software mimics the stimulus functions of the mind. But the patentable input devices that permit software to mimic vision, hearing, touch, taste and smell are as different from the software, as human senses are different from the mind. Software mimics the memory of the mind, but the patentable storage devices are as different from software, as the physical brain is different from the intellectual processes it performs. And software mimics the response functions of the mind, but the electronic and electro-mechanical output devices are as different from the software as thought is from action. Software is not a gadget.
So, let me define what software is, for the purpose of our discussion, based on its functionality -- its utility; the useful character of its art: Software is what occurs between stimulus and response, with no physical incarnation other than as representations of binary logic.
Prior to the 17th Century, analytical intellectual processes and algorithms -- sometimes called mathematics, sometimes called logic, often called philosophy -- were conducted in natural language, written and verbal. It is amazing to realize what intellectual advances were made using no more than those primitive "inventions" -- language and writing. Then Rene' Descartes "invented" symbolic mathematics. And in the next four centuries, we came to depend on manipulating the hieroglyphics of symbolic mathematics -- and to some extent of symbolic logic -- to aid in performing our analytical intellectual processes to solve problems of the mind, primarily but not exclusively mathematical problems and processes. That "invention" massively enhanced our ability to perform analytical intellectual processes. Could Descartes have patented symbolic mathematics? I think not.
In the 19th and 20th Centuries, the performance of analytical intellectual processes -- "algorithms," if you will -- was further greatly enhanced by the use of mechanical and electro-mechanical calculators, where the intellectual processes were "inventions" of sequences of button-pushes and crank-turns. And they permitted problem- solving sophistication that was simply not possible without the machines. The machines that aided the mind were patentable. The intellectual processes for logically solving problems -- the computational sequences or algorithms -- were often novel, often wildly non-obvious, and certainly useful. But did those arithmetic and logical intellectual processes become patentable once they were assisted by a machine? I hope not.
As a telling aside, the name, "computer," derived from the first so- named machines replacing World War II women who were calculating ballistic trajectories by executing algorithms using row upon row of Marchant calculators. The name of the government job-classification for those women was, "computer." Furthermore, Eckert and Mauchly invented the first all- electronic computer, the Electrical Numerical Integrator and Computer -- the ENIAC. And John Mauchly's wife was a computer -- and a delightful human being, at that.
Descartes' symbolic mathematics and symbol-manipulation provided an alternative for representing and performing algorithmic processes of the mind, and greatly expanded the intellectual limits of our mind -- but its symbols and processes were not patentable.
Calculators offered another alternative for representing and performing algorithmic processes of the mind, but the processes were not patentable.
Now, software offers another alternative for representing and performing the logical and computational algorithmic processes of the mind. So instead of assisting our intellectual processes by unpatentable manipulations of symbols on paper, or unpatentable sequences of button- pushes, we now represent the content and performance of our intellectual tasks in binary form and let the computer mimic our mind. And it, again, greatly expands the intellectual limits of our mind. Small wonder that there are those who want to monopolize anything that works like it.
The fundamental question is: Do we want to permit the monopoly possession of everything that works like logical intellectual processes. I hope not.
The mind has always been sacrosanct. The claim that intellectual processes and logical procedures (that do not primarily manipulate devices) can be possessed and monopolized extends greed and avarice much too far. Algorithmic intellectual processes must remain unpatentable -- even when represented by binary coding in a computer; even when executed by the successor to the calculator.
We know that computers pose a threat to our well-known Constitutional rights. They threaten our privacy, our freedom from unreasonable search and seizure, our ability to speak without monitoring and censorship, and assemble without the chilling effect of electronic surveillance. The threat is so serious that, in 1991, Harvard Constitutional scholar Laurence Tribe, for the first time in his life, proposed an amendment to the Constitution to assure that, "This Constitution's protections for [the Bill of Rights' freedoms], shall be construed as fully applicable without regard to the technological method or medium through which information content is generated, stored, altered, transmitted or controlled." It is becoming clear that he should have also included the protection of the freedom to continue to utilize the exclusively- intellectual algorithmic functions of the mind.
What frightens and infuriates so many of us about software patents is that they seek to monopolize our intellectual processes when their representation and performance is aided by a machine. What happens from the time a stimulus is input to the mind or a machine, through its arithmetic and logical processing by the mind or a machine, until a response is output from the mind or a machine -- must be free from monopoly. The government and the money-counters must not lock away from us, our intellectual processes -- not even when we execute them with the aid of symbols on paper, or performances on calculators, or programs in computers.
What occurs between stimulus and response, with no incarnation other than as electronic representation of binary logic and all of its algorithmic interpretations, must remain free.
The representation of an intellectual process -- in a book, or source code or binary, electronic form -- should be protected against unauthorized copying, but everything that performs similar to a thus- described and implemented process that is exclusively-intellectual, simply must not be monopolized. And that includes the representations or output created by computer programs when the primary purpose of such representations is the utilitarian function of communicating results for our further intellectual use -- especially if they are anything less than amazingly novel and non-obvious.
What programmers do is figure out how to solve intellectual problems, rather than physical problems. And we represent our thought processes and their performance in software, rather than symbols on paper or key- strokes on calculators. This nation would do well to think long and hard before it decides to grant monopolies over intellectual problem-solving.
One other precedent worthy of mention is the Constitution. It declares that the intent of intellectual-property protection is, "To promote the progress of science and useful arts, by securing for limited times to authors and inventors the exclusive right to their respective writings and discoveries." [Sec. 8, Par. 8]
The function is to promote progress by protecting the artistic and non- utilitarian works of authors through copyrights, and protecting the useful devices of inventors through patents. It says nothing about monopolizing intellectual processes -- not even when they are mimicked by a machine designed explicitly to aid the intellect.
Everything that is represented or performed by software is first a completely-detailed algorithmic intellectual process. There are no exceptions, other than by error.
Thus, I respectfully object to the title for these hearings -- "Software-Related Inventions" -- since you are not primarily concerned with gadgets that are controlled by software. The title illustrates an inappropriate and seriously-misleading bias. In fact, in more than a quarter-century as a computer professional and observer and writer in this industry, I don't recall ever hearing or reading such a phrase -- except in the context of legalistic claims for monopoly, where the claimants were trying to twist the tradition of patenting devices in order to monopolize the execution of intellectual processes.
For the first 40 years of the computer industry, and most of the 20 years of the microcomputer industry, there were almost no claims of monopoly over intellectual processes. Nonetheless, investors invested and profited; companies grew wealthy; employment expanded; salaries were and are high; un-patented innovation was amply rewarded; and, explosive progress was promoted -- as the Constitution intended. Can you imagine how computing would have developed if everything from b-trees to Shell-sort to n-way tape-merge to block- structured programming to object-oriented design were each monopolized for 17 years, like xerography? Where would Microsoft be if Gary Kildall had patented CP/M, of which MS DOS was a look-alike? Or Digital Equipment Corporation had patented the TOPS-10 CCL interface, from which Kildall derived the CP/M command interface? Where would Windows be if graphical user interfaces had been patented by Apple? For that matter, where would Apple be if Xerox Palo Alto Research Center had locked up GUI techniques for 17 years? Neither the innovators, nor the investors, nor the consumers would have benefited, and national progress would have been greatly suppressed.
Of the thousands of programmers I have known in the last quarter- century, I have never heard a single one say they didn't develop a program because they couldn't monopolize.
Of the thousands of programs I have known about as a multi-decade industry observer, I don't know of a single one that was innovative enough to promote progress -- much less perhaps-qualify for a patent as a "useful art" -- that couldn't find funding. This is at least partially due to the fact that -- unlike patentable devices -- software can be created with relatively little investment, and quickly duplicated and distributed at even less cost. And almost every major software publisher, today -- including Autodesk -- is proof of it.
The system was not broken when there were no software patents.
Now however, there is growing evidence that software patents have begun to harm and deter progress in what your hearing notice stated was, "the fastest growing industry in the United States ... [that] employs about 4 percent of the American work force ... [with employment growing] at an annual rate of 6.6 percent." Wealthy corporations are beginning to use their software patents to attack weaker opponents. But small, under-funded innovators have little more than the false delusion of protection when they receive a patent. (Software patents do, however, provide endless employment for intellectual-property litigators.) Here are examples illustrating that the waste of resources and danger to innovation are increasing:
Microsoft reportedly paid $25-million to IBM in 1992 to license IBM's portfolio of about 1,000 software patents -- not for use, but simply for self-defense. That's $25-million that Microsoft did not spent to reward innovation and promote progress.
Lotus inappropriately used copyright protection -- intended for artistic and non-utilitarian works -- to monopolize displays and key-strokes that were useful and should never have been protected under copyright in the first place. But they were also obvious and were hardly novel by any stretch of the imagination -- thus, should not have been protected at all, certainly not by patent. It appears to most of us that Lotus won only because they could outspend their opponents -- under-funded Paperback Software. Now, we have dangerous case-law precedent and have destroyed what was otherwise a viable, cost-competitive, job-producing, tax-paying, innovative software publisher. Innovation was not rewarded, and progress was not promoted. Fortunately for Lotus, VisiCorp had not copyrighted, much less patented, the displays or intellectual processes that they chose to mimic by machine, a decade earlier.
Paul Heckel of HyperRacks, who spoke yesterday, was once an innovative programmer --Jwho foolishly thought a software patent would protect his innovation. But he is neither an attorney nor wealthy. So he has wasted most of his last decade in bitter, consumptive attempts to protect his patent, but has reaped little reward. His patent left him no time to use his technical talents for personal benefit nor to promote progress.
The company for which I am speaking, Autodesk, holds some number of software patents and has applied for others -- which, of course, remain secret under current U.S. law. However, all are defensive -- an infuriating waste of our technical talent and financial resources, made necessary only by the lawyer's invention of software patents. Autodesk has faced at least 17 baseless patent claims made against it and has spent over a million dollars defending itself, with millions more certain to pour down the bottomless patent pit unless we halt this debacle. Fortunately -- unlike smaller software producers -- we have the financial and technical resources to rebuff such claims. We rebutted all but one of the claims, even before the patent-holders could file frivolous law-suits, and will litigate the remaining claim to conclusion. Note that your Office has issued at least 16 patents that we have successfully rebutted, and we never paid a penny in these attempted extortions that your Office assisted. But it was an enormous waste of resources that could have better been invested in useful innovation. These unending baseless claims benefit patent lawyers, but they certainly do not promote progress.
Ask each major software publisher in your hearings to disclose the money and talent they are wasting on software patents and defenses, instead of investing it in progress and innovation. I would guess that, across the software industry, such claims may consume as much as a billion dollars. Much worse, they are consuming more and more time of the programming talent in each company, for programmers are the only ones technically competent to assist attorneys in preparing one defense after another after another. Still other programmers remain unproductive in one deposition after another after another, and one court hearing after another after another. This does not reward innovation nor "promote progress."
Furthermore, software patents can provably fail to reward innovation and at the same time deter progress. As an example: There is an arithmetic coding compression algorithm that is an option to the JPEG standard. But such arithmetic coding algorithms are patented by IBM -- and possibly subject to patents held by Mitsubishi and others. As a direct consequence, almost no JPEG implementations support this option -- they only implement Huffman coding, that is less efficient but not subject to monopoly. Thus IBM receives no revenues, users must settle for a less powerful tool, and IBM cannot even use the technology to their own advantage because their products would then be incompatible with those of other vendors.
Additionally, patents impede progress and harm the market by forcing incompatibility -- designed exclusively to avoid infringing a patent. Innovation is diverted from genuine progress into finding a path to circumvent the visible alligators in the patent swamp, with other alligators invisible, hidden beneath this nation's disclosure delay. Products are not interoperable, and standards are thwarted because components in them are patented.
It's bad enough to inflict this harm and impediment to progress on physical devices. To extend it to machine-mimicked intellectual processes is outrageous.
Finally, there is the intense danger that software patents pose to our industry's global competitiveness. We all know that Wall Street's near-term bottom-line myopia strangles U.S. corporate long-term research-and-development investment -- unlike corporations in some other nations, notably on the Pacific Rim. Thus, we are at a disadvantage in producing basic innovation that is patentable. But foreign companies can do it, and are doing it. As an example, several Japanese companies are pursuing extensive research in fuzzy logic, and are reportedly filing for U.S. patents on every comma and semicolon. Are they likely to license such patents to their U.S. competitors?
We have seen what foreign investments in long-term R&D have done to our competitiveness in traditionally-patentable areas such as consumer electronics, automobiles and optics. Nonetheless, few publicly-traded U.S. corporations have dared to endanger next quarter's bottom line by investing great amounts in R&D that can pay off only in the long term. Shall we inflict this danger on our software industry?
And the debacles deriving from differing national patent policies are intolerable. For example, the patent on public-key cryptographic algorithms only restricts their use by North Americans -- due to different disclosure requirements. (Strictly incidental to the patent issue, U.S. tax-payers are the ones who paid for public-key research and development at two publish-or-perish academic universities, but are among the few people in the world who cannot freely use what they paid for.)
Let us also agree that the Constitutional intent is to "promote progress."
So let us disregard who wants what for self-benefit, and act on principle. We propose as a principle, that those processes that are exclusively-intellectual and exclusively-algorithmic -- even when mimicked by machine -- must not be monopolized.
These are Autodesk's implementing recommendations, and they are mine, stated as a deeply concerned computer professional and a multi-decade industry observer and entrepreneur. And, they are supported by the vast majority of talented programmers I have known. (They are in less favor, however, with many of the intellectual-property attorneys I know -- who do not create technical innovations, but are well-paid to patent them, threaten them and defend them.)
We offer two recommendations, the second having twelve parts -- so to speak, the 12 Apostles of Redress:
FIRST: Issue a finding that software, as I have defined it, implements intellectual processes that have no physical incarnation; processes that are exclusively analytical, intellectual, logical and algorithmic in nature. Use this finding plus the clearly-stated Constitutional intent, to declare that the Patent Office acted in error when it granted software patents. Declare that software patents monopolize intellectual and algorithmic processes, and also fail to fulfill the Constitutional mandate to promote progress -- that in fact, they clearly threaten it. Declare that only the representation of such algorithmic intellectual processes -- in human-readable or machine-readable form -- is protectable, and then only by controlling the right to copy it. Declare that software, as a mimic of the mind, cannot be patented.
Like Prohibition's speak-easys and the nation's unaware nuclear-test subjects, this will create the problem of how to make whole, those who hold erroneously-issued patents. But that cannot justify continuing to patent intellectual processes. Repaying patent-holders can and should be resolved through the political process -- although patent-holders cannot expect to retain their improperly-granted monopolies of logic. Like the rest of us, the government must accept the consequences of its errors and move to correct them.
SECOND: Until -- and only until -- software patents are definitively prohibited, reject or at least freeze all such applications that have not yet been granted, pending conclusive action on all of the following twelve recommendations:
But, to best promote the nation's progress, reward innovative algorithms, benefit investors, enhance the economy, create and protect high-wage jobs, and permit continuing global leadership -- refute software patents.
Let us stand on each others' shoulders, rather than on each others' toes.
I appreciate your attention to these comments and recommendations and look forward to your timely response and action.