From: Paul Robinson <PAUL[_at_]TDR.COM>
Organization: Tansin A. Darcos & Company, Silver Spring, MD USA
The following represents the text I will be reading at the Patent Office hearing on the relationship between computer software and patent issues, Crystal City, Virginia, 9:15am February 10. This text has been edited to allow me to fit it, and possible questions, into an 11 minute space.
This is a short portion of my remarks on the matter. This will be part of a complete comment on the Federal Register text. My comment will be posted to the Internet once completed.
Good Morning Commissioner Lehman, Mr. Kushan, the staff here, members of the audience, people reading this report in the future and anyone else I've forgotten.
My Name is Paul Robinson. I am Chief Programmer for Tansin A. Darcos & Company, a software development firm specializing in text processing applications; I also do work on Commercial Philosophy and metaphysics of computer systems. My special interest and my personal hobby is collecting compiler and other program sources. My reasons for this are that these all solve problems. By reading the manner and method other people have solved other problems, it gives me insight into how to solve mine.
This is a common practice in the computer world in order to, as the expression goes, "Not reinvent the wheel." I assume this is common in other industries. In fact, this is most likely the reason that we have a patent system; someone is granted the exclusive right over commercial use of an invention for a limited term in exchange for telling the world about it.
For most computers, every application such as word processing or spreadsheets has at least two and possibly three or more different applications fighting for market share. The fights in this industry are usually referred to by the expression "Dinosaur mating dances" as huge companies fight for market share by releasing new programs to introduce new features that the companies believe the customers want. Version 3 of Turbo Pascal was an excellent language compiler and less than 40K. Version 4 would fit on one 360K diskette. Today, Turbo Pascal for Windows version 1.5 takes 14,000K of disk space. The program that is probably the premiere application for graphics design is Corel Draw!, which has so much material it is now being released on not one, but two 500 megabyte CD-Rom disks.
But there are probably still niches for smaller companies to move into.
With the rapid changes in the marketplace, it is necesary to be ready to have new programs and new releases of old programs out to encourage people to move to the next release. In some cases, companies make more money from upgrades, and need to do so to stay alive. These kind of cycles mean new releases have to be out very quickly; in a matter of weeks to months.
With this kind of rapid development cycle, delays in the release of a program could be fatal and the time available to create the work is sometimes barely enough. Until recently, the only legal issue that anyone had to worry about was copyright infringement. That could be avoided by creating new work from scratch.
Now we have another issue altogether. A programmer can independently create something without ever knowing about any other developments, and yet be sabotaged by the discovery that the method that they used is patented. This is a standard problem that all industries have had to face, and it is part and parcel of living in an industrial society. But there is another problem. A computer program is the written instructions by a human being to tell a computer how to perform a particular task. As such, there are only two parameters: the input supplied to the program and the expected output. Everything else is literally a figment of someone's imagination.
This bears clarification. A computer program is the means of manipulating the internal data paths of a computer system. There is no requirement that the manipulations have any correspondence to the real world. In this, the real world, doing anything requires the expensive movement of people and goods from one point to another, the possible refinement of materials into other materials, and the expenditure of energy and resources. Doing anything in a computer is merely the essentially cost-free movement of electron paths from one direction to another; it brings forth the apportation of the concepts of the madman Imanuel Kant into reality: a world in which anything is possible:
There are things that can be done within a computer program that cannot be done in the real world, or would have undesirable consequences. As such, we should ask whether the patent rules, which are designed to apply to real-world conditions where doing something requires the expenditure of energy and resources, should apply in a world where the known rules of the universe do not apply. Because the entire design starts from scratch, and the designer doesn't just get to play God, he <italic>is<\italic> God.
Despite the ease under which someone can do something, we still live under real-world constraints. Once a design choice is made, it is very expensive in time and effort to change it. Worse, because most programs have interactions that cover every part, a change to one part can cause unexpected and even undesirable side effects in unknown and unexpected places. Computer programs may be "the stuff that dreams are made of" but once placed into concrete form as written in software instructions, it's just as expensive to repair or change as if it was carved out of real materials.
It may be necessary to change the rules on patents to comply with the conditions that exist for computer programs.
There has been talk of instituting "first to file" in order to "harmonize" with the systems in other countries; I think that is not a good choice; most countries have fewer patents, and provide protection which is much narrower than our system does. This would also mean that someone who does invent a new and useful technique for use in a computer application would be unable to collect any royalties from someone else who is using the same invention, who thought of it after they did, but started using it before they filed.
The two really large problems that exist in our system are probably two part: the secrecy under which patent applications are filed, and the problems if a program uses parts of several patents, which might not be discovered until later.
As I mentioned earlier, computer programs are created out of the figment of someone's imagination, then mass copied, the way an original painting can be reproduced by lithograph. A single large application might have a dozen people working on it, and upwards of 50 different features, and might have upwards of 200 or more different parts, any one of those might be infringing on zero, one or more patents depending on what the claims are. I doubt seriously that all but the largest corporations have the resources to do 200 patent searches on a single software application, which would be prohibitive for a small company, because it is likely that a large program could infringe dozens of patents, due to the continued development of ever larger applications that do multiple simultaneous functions.
But more than that, you can't do patent searches on works which are under application form, until after the patent has been issued. And more importantly, with more than 1,200 patents issued every week, checking them all for possible interconnection would make it impossible to do any serious work.
Seventy years ago, fears that the major piano player manufacturer would tie up the entire song market and prevent other companies from creating player piano rolls caused Congress to institute compulsory licensing. This may be an idea whose time has come again.
Therefore it might be considered to make two changes in the patent law with respect to computer programs: to implement a standard compulsory license, perhaps 10 percent of the manufacturer's suggested list price, and to eliminate secrecy provisions in the filing of patent applications.
Either of these could certainly help the situation. Eliminating secrecy and publishing applications once filed would let people know about pending inventions: they could endeavor to avoid infringements in advance; it might also allow them to file interferences early, if it turns out that they invented the concept earlier, while it is cheap to do so; and would allow people to be aware of what is being developed which would comply with Article 1, Section 8 of the Constitution, where patent protection was designed "to encourage the improvement of the useful arts".
The other option of setting a standard royalty via compulsory license would eliminate the worries of someone infringing upon an existing patent or one that is filed after their work is created. It would also grant to inventors an income stream from those who use their inventions, which started before they filed their application but after they reduced the invention to practice. It would also limit liability and exposure to sustainable limits. As it stands, if someone develops a program that infringes upon 40 patents, and they each want a 3% royalty, it isn't hard to see that 120% of the program's income is not going to be possible.
--- Paul Robinson - Paul[_at_]TDR.COM Voted "Largest Polluter of the (IETF) list" by Randy Bush <randy[_at_]psg.com> ----- The following Automatic Fortune Cookie was selected only for this message: If the odds are a million to one against something occurring, chances are 50-50 it will.Received on Thu Feb 10 1994 - 06:11:00 GMT
This archive was generated by hypermail 2.2.0 : Mon Mar 26 2007 - 00:35:11 GMT