On Mon, 31 Jan 2011 21:23:52 +0000 Richard Dobson <richarddob...@blueyonder.co.uk> wrote:
> On 31/01/2011 12:53, Andy Farnell wrote: > Er, they aren't, and never have been. Hey Richard, Sure they "aren't allowed". But they de facto _have_ _been_ allowed, and that's why we're having this discussion. If the key to the argument for this stumbling mistake is the ill formed notion of a virtual device then it will fall over easily. Software patents were simply never a considered, rational move. Instead they are a sleepwalk into a dream of market appeasement, based on a foggy understanding of the relationship between a real "device" and a virtual one. As you say they "recognise the notion", but haven't "thought about the definition". Since there is nothing to divide the line between "this" virtual device and a Universal Turing Machine, then there's no partition between the abstract and the concrete. In the former case the machine definition will be written into the claim which is concretized and becomes copyrighted, in the latter case the claim fails the "idea" (abstractness") test. --snip > Hence the classic original FM patent. It uses multiplication (can't be It's cool you picked FM, it helps develop an argument surrounding ambiguity and broadness: In fact there never was a patent on FM. There was a patent on phase modulation based on manipulation of an accumulator in a specific way. "Well they're the same thing", you may say. "Exactly!" I say. So, why do we make this mistake? Because FM and PM are congruent, that is to say there are different mathematical representations with possibly different code flows that amount to the same thing. "Frequency Modulation" became the marketing phrase. Maybe because a bunch of execs thought it sounded cooler. But that's not the end of it because, I forget who, maybe Beauchamp or Arfib, who showed at a similar time, that FM and wave-shaping could be considered congruent. You can see modulation as dynamic wave-shaping in the case that the shaper is another oscillator rather than a table. Therefore, it really became an interpretation (in code) based on whether you were using a stored or generated function whether what you call FM or wave-shaping is your technology. This is why in my book I was quite clear to draw a distinction, as is found throughout design theory, between model, method and implementation. So what significance does this have, given we all agree that so called "ideas", abstract mathematical formulations, even if they are functional, cannot be admitted as patents in the absence of a concrete design and purpose? Since Aristotle, a trisection of realms, often encountered in social and psychological enquiry, distinguishes the "real", the "imaginary" and the "symbolic". Symbols and the rules of their combination are a shared, public realm, though unique ideas may be communicated by combinations of atomic symbols. This is something we grasp easily in computer science and has bearing on much of software engineering. Classically the symbolic mediates the real and the imaginary. Outside computing it's normally seen as a madness or social malady when there is sufficient confusion of any of these realms. The patent system does not properly distinguish these things for software. It was never designed for software. Software (purely symbolic) was shoe-horned into the patent system to meet industrial demands much too fast. It stops at the symbolic and merely implies the real (design). Since both abstract and concrete symbolic forms are possible there is a SIGNIFICANT AMBIGUITY surrounding any attempt at a "software patent". What did this cause? IMHO, a great injustice. One of many mischiefs software patents perpetrate echoing through the last decade. Should Yamaha have been able to obtain the patent they did? Let's say "yes", (keeping aside my other objections to software patents). Should Yamaha have been able to monopolise the use of FM in music synthesis as a result? Categorically no! No! No! No! Notwithstanding that there were other uses of FM in music synthesis prior to the Yamaha patent, the point is that the interpretation of the patent, in reality, by people who were not qualified or diligent enough to understand its narrow symbolic meaning, was too broad. You may conclude development was stifled. Why? Because the patent, while for a very specific implementation, was interpreted and defended as a claim on a broad class of methods. Everything else in that class was effectively prohibited during this time. Were Yamaha rationally justified in filing a patent, which was on a design, but was interpreted as being on a process? I'm sure they didn't intend to muddy this boundary but given that it was actually a VLSI/ASIC job that would have taken at least an electron microscope and months to reverse engineer, probably no. Maybe Yamaha were over-eager and overreached in their claims. The claims were certainly over-interpreted. Ross raised another interesting question on the FM debacle. I paraphrase: "Didn't that actually stimulate Korg, Casio, Roland and others to develop better and equally interesting methods?" I've been thinking hard about this one. Initially I see it as happy hindsight (in psychology: a framing bias based on cognitive dissonance where you only see good outcomes from adverse conditions - e.g. you justify abuse or war as "making you stronger") Although I now think there's something in it, because Korg could have built wave-shaping technology with a similar sound and interface to Yamaha, but probably chose to make it as different as possible to avoid a market conflict. Interestingly, that would still have been a rational choice with or without Yamaha's patents. Problem is, it works in areas of software where there is clearly more than one way to do it (as Perl hackers would say), but there are strategic singular behaviours or methods, choke points, that when seized can only work to the detriment of industry in general. Nobody wants to talk about these because they are both desirable to those invested in the current protectionist system and very hard to know in advance (if there is another accessible solution). People will claim that they are "obvious". But that isn't necessarily so. Therefore, obviousness alone shouldn't be the principle objection. There is a moral question surrounding a mechanism that can block progress for others, whether the intentions are good or not. I realise that in 2011 many are frightened by the word "moral". They dismiss morality as woolly and subjective. I would argue that some moral choices are far clearer, in terms of consensual correctness, than interpretations like what is "obvious". Anyway I rest my point on broadness here: The definition of Patents is not sufficiently mature to encompass software because it cannot do so with unambiguous symbolic boundaries. The necessary symbolic precision required means the Copyright would be a better mechanism. If it's not code you are trying to protect then it's an abstraction, which by definition is prohibited. QED. sincerely Andy -- Andy Farnell <padawa...@obiwannabe.co.uk> -- Andy Farnell <padawa...@obiwannabe.co.uk> -- dupswapdrop -- the music-dsp mailing list and website: subscription info, FAQ, source code archive, list archive, book reviews, dsp links http://music.columbia.edu/cmc/music-dsp http://music.columbia.edu/mailman/listinfo/music-dsp