Path: utzoo!attcan!uunet!mcsun!ukc!dcl-cs!aber-cs!odin!pcg From: pcg@cs.aber.ac.uk (Piercarlo Grandi) Newsgroups: comp.arch Subject: Re: Patents and Architecture Message-ID: Date: 4 Jul 90 14:50:48 GMT References: <62864@sgi.sgi.com> <=Y943A7@xds13.ferranti.com> <37297@ucbvax.BERKELEY.EDU> <63007@sgi.sgi.com> <21F4V63@xds13.ferranti.com> Sender: pcg@aber-cs.UUCP Organization: Coleg Prifysgol Cymru Lines: 75 In-reply-to: peter@ficc.ferranti.com's message of 3 Jul 90 22:29:45 GMT In article <21F4V63@xds13.ferranti.com> peter@ficc.ferranti.com (Peter da Silva) writes: > They had > to try to fool the patent office with a patent claim describing an > encryption device based on RSA (and the claim has been thrown out in > other countries as ridiculous) because mathematics cannot be patented. This just illustrates another problem with the patent laws, and how they're not up to dealing with software. So? But RSA is an algorithm. Algorithms are mathematics. They had to patent a device because algorithms cannot be patented, hoping that no device could be built for RSA which was different in design from their own. I do not see the problem with patent laws; RSA were trying to really patent an algorithm, not the use of the algorithm as a program. Just a moment here: what you mean by "software" when you say it has been patented: 1) a specific implementation, e.g. a program 2) the design of an implementation 3) the algorithms used in that implementation My understanding is that 3) cannot be patented, because it is maths. 1) can be copyrighted, not patented. This leaves 2), but then either it essentially coincides with 1) or 3); maybe it makes sense to grant a patent on the design of a specific program, but what on the design of a class of programs? Or do you want to make program writing techniques to be patentable? It is very easy to build unbelievable examples for software patents: * For example: Hoare invents quicksort, so AT&T have to get a license to implement it as qsort(3)? * Or even more fuzzy: Church invents lambda calculus, so McCarthy has to pay royalties every time lisp 1.5 was run? * Maybe more reasonably: Boyer and Morris of Manchester invent compiler compilers, so every time you want to write a parser generator you speak to their lawyers? * Or again: Baker of MIT invents incremental copying collectors, so when Hewitt derives from it generational collection he is infringing on Baker's patent? * Stretching it a bit to methodologies: Dijkstra invents the use of weakest preconditions to write programs, and every time you use one to design your program, you owe him royalties? The problem is that patents are there to protect how a specific gadget is built, not mathematics, algorithms or concepts; you can patent transistors, and even then the more specific your transistor type the stronger your patent, but not the use of semiconductors in electronics. I know that there are patents that claim ownership of concepts, but they usually get thrown out by the courts. There are good pragmatical reasons to avoid patenting mathematics and anything that is akin to mathematics (ideas in general), and they apply to many so called "software" patent matters. Hey, there has been a lot of progress in software without patents; we are now confronted with the idea that somebody want to patent, under the guise of software, *architectures*. Is this reasonable? Is it necessary given currently available protection, as copyright for programs, patent for industrial *applications*, and trademark for look and feel? -- Piercarlo "Peter" Grandi | ARPA: pcg%cs.aber.ac.uk@nsfnet-relay.ac.uk Dept of CS, UCW Aberystwyth | UUCP: ...!mcsun!ukc!aber-cs!pcg Penglais, Aberystwyth SY23 3BZ, UK | INET: pcg@cs.aber.ac.uk