Path: utzoo!utgpu!watmath!uunet!tut.cis.ohio-state.edu!snorkelwacker!think!kulla!barmar From: barmar@kulla (Barry Margolin) Newsgroups: gnu.misc.discuss Subject: Patenting software (was Re: Why I do not support GNU) Message-ID: <31007@news.Think.COM> Date: 19 Oct 89 23:57:13 GMT References: <8910160520.AA01740@sugar-bombs.ai.mit.edu> <131@euteal.ele.tue.nl> <17359@watdragon.waterloo.edu> Sender: news@Think.COM Distribution: gnu Organization: Thinking Machines Corporation, Cambridge MA, USA Lines: 68 In article <17359@watdragon.waterloo.edu> afscian@violet.waterloo.edu (Anthony Scian) writes: >There are two distinct issues being addressed by patent (protectionist) >debates: Patentable software (algorithms) and "look and feel". >Imagine if the early computer pioneers placed the almighty dollar >over advancing knowledge. LR(1) parsing or for that matter anything >in the "dragon" book would be royalty costs for any language >translator company. Let's leave patents to physical processes and >techniques. What makes physical processes so special? Why is someone who figures out that putting two gears together in a particular way does something useful entitled to more protection than someone who discovers that putting two gotos together in a particular way is useful? They both have invented new mechanisms, so they should be entitled to similar protection. At the time LR(1) parsing was invented, software patentability was not very well known. Since then, many software patents have been granted: Ritchie patented the Unix setuid mechanism (he has explicitly decided not to enforce the patent), RSA patented their public-key encryption algorithm, I think Apple has a patent on pull-down menus, and someone has a patent on using XOR to implement a raster graphics cursor. Another reason LR(1) parsing might not have been patentable is that it might not be considered a novel and unobvious invention, or there might not be any individual inventor. I don't know the research processes that led to it, but if it were the result of research throughout the industry, and it was the natural result of that research, it might not be patentable -- in effect, the whole computer parsing research community invented it together. There are some people who don't believe that patent protection is reasonable. The notion that an independent inventor can be prevented from marketing his idea simply because someone else who thought of the same thing patented it seems unfair; patent law seems to assume that the independent inventor stole the idea. I can sympathize with this view. However, I have little sympathy for those who distinguish hardware and software engineers' patent rights. >Suppose you prove a theorem. Would you like to see mathematics >come to a stand-still while everybody protects their ideas? Patents might allow you to protect the *mechanism* you used to prove the theorem (if it were a novel technique), but not the theorem itself, which is simply a universal truth. The theorem could them be used by other mathematicians to prove corollaries, but if there were other theorems amenable to proof by your technique you could license the technique. Mathematics would only come to a stand-still if everyone refused to pay the license fees (in which case, the patent holder would probably not bother charging it, since he's not making any money from it anyway -- there's not much money in knowing how to prove a particular theorem all by itself). >Besides, using your analogy you wouldn't be able to package the >CD player in a similar manner to any of your competitors. >(No ON-OFF switch, no LED display, no sliding platform, etc.) Patent protection is not permitted for "obvious" inventions, and I expect it is disallowed for designs that are mandated by industry standards. I'm sure most simple switches fall into these categories (even if not, most switch designs are old enough that the patents have expired). For example, labeling an on-off switch with "1" and "0" is an ISO standard, so it could hardly be patentable. Given the existence of 7-segment digital LEDs, I don't think anyone could get a patent on putting a bunch of them in a row to represent a multi-digit number, but I imagine the original 7-segment LED was patentable. Barry Margolin, Thinking Machines Corp. barmar@think.com {uunet,harvard}!think!barmar