Software Patents
The transcript below is my testimony at a hearing held by the US Patent and Trademark Office (USPTO) in 1994, on the question of whether to allow patenting of software. At that time, a few software patents had been issued, and many, many more had been filed but not yet acted upon. I felt strongly, along with many others in the software industry, that this was a lousy idea, and the testimony I gave expressed that.
Unfortunately, the PTO decided otherwise, and software patents today are widely issued. In fact, many of the predictions of those of us who argued against such patents have come true. We regularly see large patent fights and awards, which are generally detrimental both to consumers and to innovation. Remember, the core idea of patents is to foster, protect, and reward innovation. It does so by, in many cases, protecting profits as well, but that is not their reason for being, merely the tool. All too often, patents are used by their holders solely for profit, and to the detriment of innovation. This is perhaps nowhere more true than in the case of software.
All of this represents my personal opinion, not any official position of Wind River, even though I was CEO / Chairman of Wind River at the time of this testimony. I am no longer part of Wind River management, though I am on the board. I am not an official Wind River spokesman on this issue, nor do I want to be. What I would say is that, today, software patents are the rules of the game, whether I like it or not. I think that every company has a responsibility to its shareholders to create the most value that it can for them. It does so by playing by the existing rules as effectively as it can, whether it likes those rules or not. For many software companies (though definitely not for all of them), that means aggressively pursuing patent protection, at least for defensive purposes, and sometimes for offensive purposes as well.
Software Patent Policy
Public Hearing on Use of the Patent System to Protect Software Related Inventions
Transcript of Proceedings Wednesday, January 26, 1994 San Jose Convention Center
Jerry Fiddler, CEO/Charman, Wind River Systems
Mr. Fiddler: Thank you.
I just dashed in the door. Traffic was terrible.
I stand before you not as an expert in intellectual property law. I’m not a lawyer. Most people in this room know far more about intellectual property law than I ever will. Rather, I wish to speak to you as an expert software engineer and the founder and CEO of a successful software company, Wind River Systems.
Wind River Systems is a 3O million dollar public company, with 4O percent of our revenue from overseas. We create software for embedded systems, the microprocessors found inside our cars, fax machines, telephones, robots, factories, consumer electronics. According to Software Magazine, last year we were the 92nd largest software company in the US.
My perspective on software patents is simple: stop issuing software patents. Software patents should not exist. I say this for a number of reasons.
First and foremost, I look at the reasons patents exist, which is for the benefit of society. Certainly there are fields where patents are essential because of the large investment involved for creation of technology and the ease of copying that technology. In such situations, patents incent the major investments necessary for those inventions which benefit society.
This doesn’t apply to software. Availability of patent protection is not necessary to incent creation of software. Copyright and trade secret protection are entirely adequate and more appropriate. Yes, major investments are necessary to create software, but that investment is primarily involved in quality implementation and support of the software, not development of the algorithms and ideas that might be patentable. Therefore, unlike a drug, for instance, it’s not substantially cheaper or quicker to copy a program’s functionality than it is to develop the original.
The deal society makes with the inventor, “Tell us about your invention and you can have a monopoly for 17 years,” is not a fair deal today when it comes to software. In a field changing as fast as software is today, 17 years might as well be a millennium. The deal might as well be phrased, “Tell us about your invention and you can monopolize it forever,” so the fact that we, society, know about it is meaningless.
In fact, patenting of software is actively harmful to society. People don’t need software monopolies. They need software that’s open, compatible, and that adheres to their expectations and standards. They need the software equivalent of expectations like “accelerator on the right, brake on the left”. Patenting of software could only impede these goals.
Furthermore, patenting of software will not accelerate its creation or advancement. Rather, it will impede that advancement, which is far better driven by the free market than by monopoly.
Imagine where we would be today if patents had been granted on technology or concepts critical to word processors or spreadsheets. Rather than the sophisticated and elegant tools we now have available thanks to competition, we would still be using something very much like the primitive first versions of those tools. Worse still, we must remember that word processors and spreadsheets have been largely responsible for spawning an industry and making the personal computer a part of most of our lives. The quality and advancement in those tools have created opportunities for computer manufacturers and for other software vendors who can sell to users who have computers primarily to run those primary tools.
It’s not too strong to say that if there had been strong patent protection for the first word processors and spreadsheets, the personal computer industry today might be five to ten years behind where it is. As another example, if aspects of TCP/IP, the network protocol, had received patent protection, today the Internet might very well not exist.
Creation of software will also be impeded by the difficulty of writing software that doesn’t inadvertently trip across a patent somewhere. This is true in other fields where patenting is less controversial, but it’s far worse in software. It’s not unusual for a program to be a million lines long and consist of many thousands of subroutines and functions. Algorithms and ideas are embodied in each of those components and in combinations of them. Some of these algorithms may be studied in school or found in books, but many are developed “on the fly” as the program is created. Many of these subroutines and functions might be far afield from the purpose of the program as a whole.
An operating system, for instance, might contain routines for sorting and searching, handling queues, parsing text, controlling hardware, testing memory, et cetera. It will be impossible to know which of these routines, algorithms and ideas violate a patent, because every programmer would need to understand every software patent — every software patent that’s active. Software is simply too complex, composed of too many pieces which are too easy to create, to lend itself to being broken down into patent-sized chunks.
I can easily envision a world in which progress in software is totally blocked by a web of patents owned by a very few very large companies; not the best or the most creative companies but rather those with the most lawyers. In a world like that it would be completely impossible to start and build a company like mine — and this nightmare could come to pass very quickly.
To date there has been little litigation regarding software-related patents. God help us all when that litigation does begin. Judges and juries will be asked to rule on whether a large complicated program, potentially millions of lines long, written in an obscure computer language, violates an arcane patent. The claimed violation will be built into the very fiber of the program, hidden within the program’s structure and data in complex and subtle ways. One expert will say one thing, another expert will say the opposite, neither judge nor jury will be competent to understand the nature or veracity of the patent, much less which expert is closer to the truth. The patent will have been issued by an examiner who is not expert in the specific software field and might not understand the concepts essential to operating systems, fuzzy logic, or whatever the specific field is, much less the prior art. The chances of a fair and informed decision will be vanishingly small.
Software is, perhaps, more analogous to literature and music than it is to mechanical invention. It would be silly to think about patenting the first-person novel or the sonata form, yet there are software patents already that to a software engineer are just as absurd.
As a software company CEO, I am perfectly content to compete based on the quality of the software we create and the support we provide for it. I am fully satisfied with copyright, contract and trade-secret protection for the software we write. We have begun to work on some patent applications because I think we may need them for defensive purposes, but I would far rather we didn’t need to do so.
If software patents become prevalent, it will seriously interfere with our ability to continue improving our products and our ability to continue developing new ones. It will also interfere with our ability to provide openness and compatibility to our customers — a key part of the value we provide to them.
The best possible result of these hearings for us, for our customers and for society would be for software patents to simply go away.
Commissioner Lehman: Thank you very much, Mr. Fiddler. I would just make an observation and ask a question.
You indicate where would we be if we had had patents on the spreadsheets, and so on and so forth, and I think that suggests that I think it’s one of the reasons why we may not have patents on spreadsheets and the idea of a word processing program, and so on, is because those particular items were not patentable, they didn’t meet the test of patentability.
And I think herein lies a lot of the problem when you say software shouldn’t be patentable. Well, it well may be that there’s a lot of confusion as to where that threshold is drawn, and that indeed some software-related inventions could and, you know, are very appropriately patentable, but there seems to be a lot of confusion about where the test, where the threshold, what kind of innovation meets the test of novelty and unobviousness, where that’s drawn.
And how would you feel about a more vigorous examination of where that line of nonobviousness is drawn?
Mr. Fiddler: You know, obviously, to the extent patents exist, I’d like them to be as narrow and as well-defined as possible. Clearly, that’s in everybody’s benefit.
But I think that, yes, it’s true that probably the concept of a word processor is not a patentable concept, but there certainly are key components of those that very well may have passed patent law, particularly as patents seem to be being issued, you know, very recently.
There are, I think, that even if the Patent Office is perfect, even if it issues only patents that are entirely appropriate, are entirely correct, are novel and nonobviousness, and so forth, which is I think a very unlikely place to get to, but even if we can assume that the PTO is perfect in those respects, I still think that it will have — it makes it far more difficult to create software.
If I sit here, I mean you can set for me a problem and say please write a program that does something, and, depending on the problem, in somewhere between five minutes and a couple hours I may be able to do that. Is what I have done patentable? Maybe. Maybe there’s an idea in there that is, maybe there isn’t.
To find out if there is, it will take me far longer to find that out, and there’s no way in the world I can be familiar with it and it will be very difficult for me to find it. It may be in a field far away from the one in which I’m working. It will multiply my work not by 1O or 2O or 5O percent, but potentially by thousands of percent.
Commissioner Lehman: How is that any different, really, from an engineer that’s working in electronic components of aircraft in —
Mr. Fiddler: I think it’s different —
Commissioner Lehman: — in Seattle where there’s obviously a lot of innovation and they’re constantly asked to design all kinds of gizmos and do things and yet that’s an area clearly where there’s been patentability for a long period of time and they aren’t, you know, suggesting that somehow or other engineers can’t make a move and put pen to paper or turn on their workstation without consulting the legal department?
Mr. Fiddler: I think it’s different in a couple ways. For one thing, copyright doesn’t work for them and it works fine for us. For another thing, it’s far easier to create software ideas and to make them work.
When I start and write a program, I may write however many lines of code it is, I may start with a design and do that, and I may actually have it debugged within a very few minutes. I can make changes to it by saying, “Change this line of code.” I can make it work in a very few minutes. That’s very different than a hardware concept or building something in hardware, where the turnaround time is much longer, the number of concepts probably embodied — certainly the number of novel concepts embodied in any specific project are probably much smaller.
As I said, a very small number of programmers, two or three or five programmers, can certainly write a million-line program with many thousands of ideas that may potentially be patentable. Have they been patented? Have they not been? Is there prior art? Isn’t there? It’s almost a question of luck and almost impossible to find out and it will make it extremely difficult to work to create these kinds of programs.