10 ms·

That is a (probably the most) difficult question. Intuitively, I would say, "Of course not". But when you learn more about the patent system and how it works, y

That is a (probably the most) difficult question. Intuitively, I would say, "Of course not". But when you learn more about the patent system and how it works, you understand the reasons why something like this may have gotten through. Let me lay it out. 1) Hindsight is a powerful effect. Once you've been shown something new, a person of skill may instantly deduce how it works. This does not necessarily mean it was obvious in retrospect. 2) This presents a significant problem for any system that purports to reward innovation. How do you objectively tell if anything was truly novel and non-obvious at the time it was invented? Any opinion on the quality of an invention is inherently subjective. [1] 3) As such, the way patent offices have historically decided this, is that novelty or non-obviousness is completely based on the delta with the prior art that came before it, where prior art := any published material with an identifiable date prior to your invention. [2] This is because, in true CYA style, when contended, you have an established historical record of documents proving your stance. Novelty is easy: Anything that a single prior art reference doesn't cover is novel. Obviousness is the tricky part: all inventions are combinations of existing elements. But was that combination obvious? How's that for subjective! The USPTO seems to have a rule of thumb that any claim that requires combining more than three+ references of prior art is probably non-obvious. That seems like a pretty low bar to clear (and you'd not be far off) but the catch is, those references could be anything that mentions anything resembling any element of the invention. Under this standard, it is my belief that truly innovative inventions like Diffie-Helman or RSA would not have been patentable [3]. 4) And herein lies the rub: To be fair, you must be objective. But to be objective, you must abandon all measure of quality to the power of raw data. And as data scientists tell me, data often lies. All the data in the world (at that time) may not be enough to prove non-obvious the specific combination of elements that a patent claims. Such as this patent. Really, who's going to document and publish at a venue with a proper date the fact that they developed something to skip audio files? Yes, to be fair, you must judge all patents by the same standards, so the patents in areas that have been most published about (because, you know, they are interesting research problems) are less likely to get through than those that cover some obscure, unimportant aspect hat nobody thought to document. And that, ladies and gentlemen, is why we have this situation 1. Imagine if Einstein was your patent clerk as opposed to the guy you interviewed week before last who failed Fizzbuzz. 2. This, of course, relies on good search, and IMO Google has been the biggest boon (or bane, depending on your perspective) to the patent system than most imagine. 3. Choosing primes and taking mods? Is that all?!

harshreality 13y ago
RSA and DH should not have been patentable. They are mathematical constructions. They're more beneficial when implemented in software, run on a general purpose machine, and applied to data that needs to be encrypted or signed, but they're math nevertheless. Same with this patent. It's a patent on a process, only given form through software (which is itself math). The patent references communications ports, digital memory, output units, media players. (Claim 1). The idea is that by combining physical parts with programming, they've developed a claim on a machine in the classical sense that is patentable. There's simply a divide between people who think this sort of reasoning and patent is perfectly reasonable, and people who don't (most of HN). It doesn't matter how clever any of the other individual claims are. They're specifying an abstract procedure and not a thing. And the software implementation of such procedures is math. That's not supposed to be patentable.
- throwawaykf03 13y ago
  The "software is mathematics" argument has a few flaws: 1) It is reductio ad absurdum, like saying "machines are metals". It is not the naturally occurring metals that are patentable in physical inventions, it is how you configure and use them. Similarly it's not the mathematics that is patentable, it is the application thereof to a practical problem. 2) More importantly, it misunderstands what abstract math is from a legal perspective. People will invoke the Church-Turing thesis and several related theorems to prove that software being executed is math, but they miss the point. An abstraction, by definition, cannot affect or effect anything in the real world. An idea in your head is abstract. The moment you act on it in the physical world, it is real. Similarly, if you can run some piece of software and get useful, practical, real world results, it is not abstract.
  - harshreality 13y ago
    You're splitting math into separate categories that are not adequately defined by court precedent. The legal profession has at best an inaccurate and non-uniform concept of what math is. Relying on what the legal profession views as math therefore is not a valid argument and depends on which part of the legal profession you survey. If software by itself can be a patentable device/machine for the purposes of patent law, that needs to be made explicit, and I happen to think it will cause serious problems; it already is causing major problems, even without explicit supporting precedent. Almost every software patent you can find mentions network communications, memory storage, disk storage, display, or other artifacts of general purpose computing. If software by itself is, or should be, patentable, why is everyone trying so hard to patent complete machines rather than only software algorithms? Maybe it's that they don't agree with you that pure software implementations of solutions to real world problems are patentable? According to your paradigm of allowing patents for application of math to a practical problem, you're fine with someone patenting a moon slingshot maneuver to get back to Earth? That's an application of pure math. We agree you can't patent using equations to predict motion, right? Can you patent using equations to predict motion and reversing that to generate a solution to a specific orbital problem? Can you patent running that algorithm on a general purpose computing machine? Can you patent an embedded platform that is built out of a general purpose computing machine, but is sold as a black box running only that algorithm with suitable inputs and outputs?

harshreality 13y ago

RSA and DH should not have been patentable. They are mathematical constructions. They're more beneficial when implemented in software, run on a general purpose machine, and applied to data that needs to be encrypted or signed, but they're math nevertheless. Same with this patent. It's a patent on a process, only given form through software (which is itself math). The patent references communications ports, digital memory, output units, media players. (Claim 1). The idea is that by combining physical parts with programming, they've developed a claim on a machine in the classical sense that is patentable. There's simply a divide between people who think this sort of reasoning and patent is perfectly reasonable, and people who don't (most of HN). It doesn't matter how clever any of the other individual claims are. They're specifying an abstract procedure and not a thing. And the software implementation of such procedures is math. That's not supposed to be patentable.

throwawaykf03 13y ago
The "software is mathematics" argument has a few flaws: 1) It is reductio ad absurdum, like saying "machines are metals". It is not the naturally occurring metals that are patentable in physical inventions, it is how you configure and use them. Similarly it's not the mathematics that is patentable, it is the application thereof to a practical problem. 2) More importantly, it misunderstands what abstract math is from a legal perspective. People will invoke the Church-Turing thesis and several related theorems to prove that software being executed is math, but they miss the point. An abstraction, by definition, cannot affect or effect anything in the real world. An idea in your head is abstract. The moment you act on it in the physical world, it is real. Similarly, if you can run some piece of software and get useful, practical, real world results, it is not abstract.
- harshreality 13y ago
  You're splitting math into separate categories that are not adequately defined by court precedent. The legal profession has at best an inaccurate and non-uniform concept of what math is. Relying on what the legal profession views as math therefore is not a valid argument and depends on which part of the legal profession you survey. If software by itself can be a patentable device/machine for the purposes of patent law, that needs to be made explicit, and I happen to think it will cause serious problems; it already is causing major problems, even without explicit supporting precedent. Almost every software patent you can find mentions network communications, memory storage, disk storage, display, or other artifacts of general purpose computing. If software by itself is, or should be, patentable, why is everyone trying so hard to patent complete machines rather than only software algorithms? Maybe it's that they don't agree with you that pure software implementations of solutions to real world problems are patentable? According to your paradigm of allowing patents for application of math to a practical problem, you're fine with someone patenting a moon slingshot maneuver to get back to Earth? That's an application of pure math. We agree you can't patent using equations to predict motion, right? Can you patent using equations to predict motion and reversing that to generate a solution to a specific orbital problem? Can you patent running that algorithm on a general purpose computing machine? Can you patent an embedded platform that is built out of a general purpose computing machine, but is sold as a black box running only that algorithm with suitable inputs and outputs?