rms@AI.MIT.EDU (Richard Stallman) (12/11/90)
I expect software patents will put an end to the GNU project within a couple of years if they are not eliminated. It will not be possible to write most of the programs that a user of GNU would expect without infringing patents, and licensing patents is not an option for us. To illustrate the problem, it appears that a patent was recently granted on include files. (Usenix told its members this patent exists, but I don't have any details yet.) It may be invalid, but proving this in court could cost half a million dollars. If we are unlucky and the patent is upheld, there go GCC and G++ down the drain. GNU developers are not the only ones afraid of patents. In a workshop with software developers in September, the Office of Technology Assessment found that most of them were opposed to patents. Many prominent entrepreneurs including jimad's employer seem to think they are a bad idea. For more information about the danger of software patents, contact league@prep.ai.mit.edu.
ark@europa.att.com (Andrew Koenig) (12/11/90)
The basis for patents in the US is Section 8, paragraph 8 of the US constitution: Section 8. The congress shall have power: . . . (8) 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; That is, the legal ability to patent an algorithm stems from the same part of the constitution as the legal ability for the FSF to prevent people from selling proprietary extensions to GNU software. Moreover, there is the same basis for both: to give people an incentive to develop new ideas instead of keeping them secret. When you patent something, you get exclusive rights to it for a while, in exchange for which you are spared the bother of keeping it secret. After the patent expires, everyone gets the benefit of it. A key phrase is `for limited times,' and it is there that I think the entire problem lies with software patents. If I invent a new method of fastening picture frames to wallboard, the construction industry would not be seriously impeded if I have exclusive rights to those fasteners for seventeen years. People have been building things for millenia; seventeen years is negligible by comparison. On the other hand, seventeen years is a substantial fraction of the total time the software industry has been in existence! This argues that software patents should be much shorter-lived than other patents. I submit that if the duration for software patents were changed to seventeen months, most practical objections to them would vanish, especially if that were coupled with the restriction that nothing can be patented if it is already in widespread use.
rms@AI.MIT.EDU (Richard Stallman) (12/12/90)
Yesterday I sent a message to help-g++ stating the FSF position on software patents, including how we expect them to affect the GNU project. Now it seems that people are starting to argue against the FSF position. This list is not the right place for that. If people want to have a discussion of the issue, they should use gnu-misc-discuss@prep.ai.mit.edu. (I should have said this for the discussion of the Apple boycott also, but I didn't remember until too late.) Andrew Koenig's message contains some accurate information and some views that partly agree and partly disagree. Rather than continue an argument here, I urge anyone interested in the issue to ask league@prep.ai.mit.edu for information on the League for Programming Freedom position and the arguments behind it. If and when I get more information about the patent on include files, I will inform people.
dlw@odi.com (Dan Weinreb) (12/13/90)
In article <9012111808.AA20219@life.ai.mit.edu> ark@europa.att.com (Andrew Koenig) writes:
That is, the legal ability to patent an algorithm stems from the
Just for the record: according to a patent lawyer I have spoken with,
by the way, you cannot, technically speaking, patent an algorithm as
such. An algorithm is considered to be like a mathematical law; it is
a fact of the universe, not something that can be patented. He also
said categorically that you cannot, technically speaking, patent
software as such, and what people are calling "software patents" are
not actually patents on the software. They are patents on a method of
achieving some goal, and they would protect equally well a hardware or
a software implementation of that method.
This probably doesn't have any important impact on the substance of
the conversation, but I thought people might be interested.
Also, it is already the case that you cannot patent something that is
in widespread use before the patent exists. The legal jargon phrase
here is "prior art". To defend a patent on, say, include files, you'd
have to show that include files were not part of the state of the art
prior to the patent. (I can't remember whether we're talking about
the date of application or the date the patent issues, but you get the
idea.)
Shortening the lifespan of patents is one of the possible reforms that
has been brought up many times. I think one complexity is that the
appropriate period varies by industry. For example, the time clock
for a new medical drug starts at the time that they begin certain
types of clinical trial; the drug company is unlikely to get FDA
approval until quite some years after that. So if patents were, say,
five years in duration, there would be no effective patent protection
for drug companies. Software, of course, is quite different.johnb@srchtec.UUCP (John Baldwin) (12/13/90)
Honestly, this discussion does not belong on comp.lang.c++. It needs to be moved to a more appropriate place. Having said that, in violation of my own request.... In article <1990Dec12.185220.9629@odi.com> dlw@odi.com writes: > Also, it is already the case that you cannot patent something that is > in widespread use before the patent exists. The legal jargon phrase > here is "prior art". To defend a patent on, say, include files, you'd > have to show that include files were not part of the state of the art > prior to the patent. The problem which endangers the whole software industry is that the U.S. Patent Office specifically requires patent lawyers to be trained at the undergraduate level in the technical discipline that they plan to apply their law training to. For example, lawyers dealing with aviation patents must have a baccalaureate-level degree in aerospace engineering. [Or something like that. I may not have that at 100% accuracy.] Further, the Patent Office SPECIFICALLY PROHIBITS Computer Science degrees from being considered valid for this purpose! The closest allowable area is EE with a specialization in computers. Hopefully, most of these EE/Lawyers are well enough versed in the software side of things to avoid and prevent prior art from being patented. Unfortunately, due to the pressures and complexities of law and business, this seems not to be the case. Prior art IS being patented; it is possible, but VERY expensive, to prove the invalidity of these patents in court. -- John Baldwin | srchtec!johnb@gatech.edu | johnb%srchtec.uucp@mathcs.emory.edu