Knuth: I am against software patents on any idea that is so obvious that you would expect a
typical student to invent it. I can maybe be persuaded that it is okay to have a patent on
something really deep, like some interior point algorithm or something. But, certainly if I had to
write the TEX system in today’s environment I would never start, because I would have to worry
too much about getting permission to use hundreds of ideas. Of course, fortunately, we got most
algorithms into software before people started taking this dumb view of patenting. But I think,
anyway, I think software writers should be paid for services, for customization, for adapting
programs, for the knowledge they have on how to maintain things. But not for the algorithms;
not for the methods that are used. The method, that is something that is usually – consider the
analogy I made to the patent office. I wrote an open letter to the patent commissioner. I said that
algorithms are like words to a writer, or like lawyers have precedents. What would the law
become if every time lawyers had a precedent they would charge a fee to other lawyers for citing
their precedent. That would be, I think, very much analogous to what people want for software.
This is something I am writing a paper on. Very interesting I think, how countries determine what is something to own and be paid for if others use it and what is common and for everyone?
