Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Posted Jun 29, 2014 20:23 UTC (Sun) by nybble41 (subscriber, #55106)In reply to: Software patents take a beating at the US Supreme Court by kleptog
Parent article: Software patents take a beating at the US Supreme Court
Only because (as your own comment illustrates) people tend to understand neither the nature of software nor mathematics. It is a fact that all software is math; the Church–Turing thesis and the Curry-Howard Isomorphism both prove that beyond doubt, from two different points of view. Unfortunately, you pretty much have to be an expert in both abstract mathematics and computer science to fully understand the implications.
> Consider for example compression algorithms like gzip. ... Some sequences will get shorter, some will get longer. And that it for the math side. ... Now a real compression algorithm is going to try to make strings you see in
real life shorter, and that has nothing to do with mathematics.
Um, no. The math extends to making the strings consistently shorter when there are compressible patterns in the input. (Random input, of course, being uncompressible in accordance with the pigeonhold principle--something else you can learn from the math.)
> And then there's also the requirement that it be efficient to implement, which is also nothing to do with math.
Wrong again. Math has quite a bit to do with the efficiency of the implementation. Evaluation strategies correspond to systems of logic, and evaluation time maps to the number of steps required to complete the program's corresponding proof. Making a program more efficient is equivalent to simplifying the proof. And, of course, the fact that some formulas are more immediately useful than others for solving certain problems doesn't many them any less mathematical.
> And MPEG compression is even more tricky, since it's not even invertible.
It's not invertible because you throw away part of the data (applying a mathematical filter--not all formulas are invertible). If you look closely, MPEG compression is even more obviously mathematical in nature than a universal lossless compression scheme like LWZ. It consists of linear transformation, difference, DCT, filtering, and encoding stages, all of which are well-known mathematical concepts.
Software is math. No other argument is required.
Posted Jun 29, 2014 20:34 UTC (Sun)
by ms-tg (subscriber, #89231)
[Link] (30 responses)
Posted Jun 29, 2014 21:16 UTC (Sun)
by mgb (guest, #3226)
[Link] (28 responses)
Posted Jun 30, 2014 13:00 UTC (Mon)
by ms-tg (subscriber, #89231)
[Link] (27 responses)
Really? That would seem to define most software as hardware.
Posted Jun 30, 2014 17:38 UTC (Mon)
by nybble41 (subscriber, #55106)
[Link] (26 responses)
What do you mean by "effects"? We're taking it to mean physical changes in the real world, which are clearly a function of hardware. If you mean "side effects" like assigning to memory locations, reading input, and displaying output, those are also part of the math, though it may not be obvious if you've never worked in a pure functional programming language like Haskell.
A program is a pure mathematical function from a set of abstract inputs to a set of abstract outputs. This can be modeled as streams or continuations, or using monads as in Haskell. The inputs can include data from external systems (like a keyboard or disk) and the outputs can be routed to external systems as well (like a disk or monitor). Or it could all be a simulation. To the software, external I/O is all abstract; it's the hardware that turns abstract states into concrete, physical I/O, and vice-versa.
Posted Jun 30, 2014 18:18 UTC (Mon)
by ms-tg (subscriber, #89231)
[Link] (23 responses)
The *operation* of anything to change state. So, for example, if one had a piece of hardware that emitted a vibration. And a software program that could caused it to emit certain patterns of vibrations which could, for example, kill cancer cells while leaving healthy cells. I would describe the vibration as an effect, but I would also describe the software as the causative agent of the effect.
> A program is a pure mathematical function from a set of
(Yes -- this is all fine -- I know enough Haskell and work in Scala)
> To the software, external I/O is all abstract
This understanding, I think, would render all complex machines unpatentable .
Let's imagine a discovery that a certain set of state changes in the material world, heretofore unknown, is extremely valuable. Please consider the example above, of certain vibrations in certain patterns, timed to physiological responses observed, being effective in killing cancer cells. But this is just an idea, and not reduced to practice, so not patentable.
Now let's imagine that I conceive of a machine which successfully reduces this idea to practice. It so happens, since the invention of software, that decomposing this problem into basic hardware units and software control on a general purpose computer is the most efficient and effective way to experiment to create the machine, and also a very efficient and effective way to construct and deliver the machine. But for the purposes of the desired observable effect (reducing idea to practice in order to kill cancer cells), the fact that complex behavior is delivered as software is immaterial to patentability of the machine, isn't it?
As I understand it, the goal of the patent system is to encourage the disclosure of *exactly* this type of discovery! By providing incentive to disclose in a protected fashion, multiple actors in industry can advance the state of art, and deliver faster improvement on this promising cancer treatment, ultimately benefiting society.
For that reason, perhaps the coordination of effects within one notional "machine", for the purposes of patenting, by reducing a complex abstract idea to practice, is the reason that "all software" is *not* considered to embody only abstract mathematical functions. Because it would be impossible to patent any complex machine invented in the computer age, if that were the perspective adopted, wouldn't it?
Instead, and I think equally intellectually consistently, we can say that software reduces mathematical functions to practice. In practice, the execution of these effects *cease* to be abstract. And as that is the heart of any complex machine, in the context of such a machine, they remain patentable?
Posted Jun 30, 2014 20:59 UTC (Mon)
by nybble41 (subscriber, #55106)
[Link] (22 responses)
In that case, the effect is entirely due to the hardware. Software doesn't cause the hardware to emit vibrations; the hardware emits vibrations in accordance with software instructions. The instructions by themselves are passive; they don't *cause* anything. They just sit there in memory until the hardware acts on them. To draw an analogy, sheet music doesn't cause a singer to produce a particular note; the singer produces the note in accordance with the instructions. The cause is the singer, not the sheet music.
> But for the purposes of the desired observable effect (reducing idea to practice in order to kill cancer cells), the fact that complex behavior is delivered as software is immaterial to patentability of the machine, isn't it?
Indeed, the software would be immaterial. The patent would cover the delivery of particular vibrations to the cancer cells in response to specific physiological observations, regardless of how those vibrations were produced. In other words, this example is irrelevant to the question of software patents, since it wouldn't be a patent on the software, but rather a particular set of physical inputs intended to cause a particular physical change in a set of physical cancer cells. Much like Diehr's rubber-curing patent, in fact. If someone produced the same results with hard-wired analog circuits and no software, they would still be infringing on the patent. That same software, running on a general-purpose computer without the vibration-producing output device, would not infringe.
If this is your main concern, rest easy--eliminating software patents would have no effect on the patent-eligibility of physical devices or processes which merely happen to incorporate software in their implementation. The addition of software to something otherwise patent-eligible would not render it ineligible. The patent simply wouldn't cover equivalent software in other contexts. Of course, when the patent is all about the software (like a patent on an encryption or compression scheme) then there's nothing else left which qualifies; these are the kinds of patents which would be eliminated.
If you were hoping for a patent on the device (probably cobbled together from off-the-shelf parts in a fairly obvious way in accordance with their intended uses) you might be out of luck. However, this example is really about the process of killing cancer cells with timed vibrations, not the device which implements it.
> we can say that software reduces mathematical functions to practice
No, we can't, because software isn't a mathematical function reduced to practice. It is, itself, a mathematical function.
Posted Jun 30, 2014 22:32 UTC (Mon)
by dashesy (guest, #74652)
[Link] (19 responses)
Posted Jul 2, 2014 21:28 UTC (Wed)
by nix (subscriber, #2304)
[Link] (18 responses)
I can't figure out what on earth patenting a play could even do, though I'm sure some smart and sufficiently unethical patent lawyer could figure it out: probably it would ban people from using those plots again, which is self-evidently ludicrous given that he pretty much entirely copied those plots from elsewhere himself. But I can't see any way in which it would prevent people from speaking English -- after all, his intended audience was already speaking English, and had been for hundreds of years.
There are lots of legitimate reasons to hate software patents. Surely you can come up with some better analogies.
Posted Jul 2, 2014 21:38 UTC (Wed)
by dashesy (guest, #74652)
[Link] (17 responses)
Posted Jul 3, 2014 7:53 UTC (Thu)
by Wol (subscriber, #4433)
[Link] (16 responses)
By the way, did you know that apparently, if you collect all the known ways Shakespar spelt his name, he rarely used the same spelling twice?
Cheers,
Posted Jul 3, 2014 9:18 UTC (Thu)
by marcH (subscriber, #57642)
[Link] (14 responses)
Mmmm, this is new to me... Are you implying that British English went through the following stages:
1. Original, French spelling
Whatever the (Hi)story was, I like the following irony: British English is using not just more French spellings but even French words and sounds[*] than American English, even though the US won independence from Britain with the backing of France = Britain's arch enemy.
[*]
Posted Jul 4, 2014 0:01 UTC (Fri)
by Wol (subscriber, #4433)
[Link] (13 responses)
What do you mean by "original French spelling"?
Modern English is a fusion of AngloSaxon, and Norman French. Probably pretty much achieved by 14 or 1500. I think by the time of King John, England had gone from being a province of the Normans to being their last remaining stronghold.
Never heard of the Great Vowel Shift.
Don't know anything about the American spelling simplification.
But the major English dictionary writers were heavily influenced by the French affectation current at the time, hence the codification of French spellings into official English. And the reason the Americans didn't follow suit is why should they? I believe it was about the time of the War of Independence or a bit later, and it was very much a London thing. But London being the centre of English Culture, it stuck.
> British English is using not just more French spellings but even French words and sounds[*] than American English
And what is "British English", pray? British (the language) is probably approximately modern Welsh, ie absolutely nothing to do with English. Actually, English isn't even the language of all of England if you want to be pedantic - as I like to put it, "the Saxons speak English, the Angles speak Scots, and the Scots speak Gaelic". Okay, English is now the dominant language in Britain, but it's actually the native language only from about the Tyne south down the East Coast. On the west coast they speak Breton/Cornish/Welsh, and in the north the sassenachs speak Scots.
Cheers,
Posted Jul 4, 2014 7:01 UTC (Fri)
by marcH (subscriber, #57642)
[Link] (9 responses)
It's just the spelling you find in a French dictionary.
> Never heard of the Great Vowel Shift.
Ever heard about Google? :-) Like many other languages and unlike English, French has (usually) just one sound per vowel (and diphtong). Practically this means you can (almost) read French ALOUD correctly without understanding it. Unlike this: http://en.wikipedia.org/wiki/English_orthography#.22Ough....
Conversely, this explains why a significant number of people have such a hard time spelling English.
As non native speakers we frequently correct the spelling of native speakers. This is because we learned the words in books whereas native speakers learnt it while speaking. Conversely, native speaker don't understand us the first time we try to read a "book" word aloud :-)
Reading French alound is not as simple and straight forward as German but much easier than English
> Don't know anything about the American spelling simplification.
It's "simplified" compared to British English because "center" for instance is written like it sounds in English and not like it sounds in French ("centre")
> And what is "British English", pray?
In the context of this discussion it's merely the variant of English vocabulary in use on the British Isles, including Ireland (where Irish is struggling), and more specifically the words and spelling which are different from American English. No need to over-interpret me.
Posted Jul 4, 2014 8:58 UTC (Fri)
by Cyberax (✭ supporter ✭, #52523)
[Link] (2 responses)
From my experience - reading Italian or German aloud is easy, Russian is a little bit more complicated (you might sound over-correct but still perfectly understandable), Ukrainian is the simplest (you can actually transcribe unknown words correctly - try that in English). English is impossible, a couple more centuries and it risks to become as hieroglyphic as Chinese...
Posted Jul 4, 2014 15:37 UTC (Fri)
by giraffedata (guest, #1954)
[Link] (1 responses)
Posted Jul 4, 2014 20:40 UTC (Fri)
by Cyberax (✭ supporter ✭, #52523)
[Link]
Posted Jul 6, 2014 0:44 UTC (Sun)
by Wol (subscriber, #4433)
[Link] (5 responses)
>It's just the spelling you find in a French dictionary.
So you're telling me that you'll find English words like "cow" and "pig" in a French dictionary from about 1200AD?
Sorry, I know I'm being facetious, but the great majority of English words came from German roots, not French, so "original French spelling" is a weird one to me.
> > Never heard of the Great Vowel Shift.
> Ever heard about Google? :-) Like many other languages and unlike English, French has (usually) just one sound per vowel (and diphtong). Practically this means you can (almost) read French ALOUD correctly without understanding it.
Maybe I should have Googled it. But this *could* well be the era in which Norman French and Anglo-Saxon really gelled into what we now think of as English. I notice the wikipedia article says that Vowel Shifts are peculiar to *Germanic* languages ...
Oh - and while I'm unable to cite my sources, unfortunately, I have heard of a text-to-speech program that reads English text. When given TRUE ENGLISH words, it only needs about 30 rules to handle a vocabulary of 50- to 70-thousand words. (In contrast, the average well-spoken person uses a typical vocabulary of about 25,000 words.) The problems that most foreigners have with English words is that those words are NOT ENGLISH! Would you like a cup o' char? That last word is Asian, and there are an awful lot of words in regular English that are poorly assimilated foreign immigrants.
> As non native speakers we frequently correct the spelling of native speakers. This is because we learned the words in books whereas native speakers learnt it while speaking.
What a marvellous comment on the excellent status of our education system. NOT! I have to admit that I learnt pretty much ALL my English grammar and spelling from studying foreign languages, in particular Latin. So I'm forced to agree with you. Most English people can't spell ...
> > Don't know anything about the American spelling simplification.
> It's "simplified" compared to British English because "center" for instance is written like it sounds in English and not like it sounds in French ("centre")
Except that "simplification" implies that something was done to make American spelling simpler. It wasn't. In the 1700s (when America and Britain parted company) we spelt it "center" in Britain too. It was only in the 1800s that Britain changed to French spellings. The Americans (sensibly) never followed our lead. It was the London-based dictionary writers who pushed English to use French spellings.
> it's merely the variant of English vocabulary in use on the British Isles ... No need to over-interpret me.
Except that "the variant" implies that "British English" is a homogenous whole. Try telling that to a Sassenach, Scouser or Geordie! Even going to visit my daughter in Yorkshire we sometimes have trouble understanding the people there because they have a *different* *vocabulary* to us southerners.
And while I'm almost certainly more easily offended than most, as someone who speaks English I object strongly to being labelled "British", just like you get a fair few Canadians and Mexicans who object to being lumped in with USAians just because they are "Americans".
Cheers,
Posted Jul 6, 2014 7:57 UTC (Sun)
by marcH (subscriber, #57642)
[Link] (1 responses)
Glad you realize that different English words may have different origins.
> When given TRUE ENGLISH words,
... or maybe not?
> it only needs about 30 rules to handle a vocabulary of 50- to 70-thousand words.
What's even more impressive, it has 10 rules for just this one: http://en.wikipedia.org/wiki/English_orthography#.22Ough....
... which leaves only 20 rules for all the rest of the language! Sorry I'm being facetious :-)
> and while I'm unable to cite my sources, unfortunately,
It is very unfortunate because these 30 rules would literally save the lives of hundreds of millions of non-native English speakers. Search harder please?
> What a marvellous comment on the excellent status of our education system. NOT!
I'm sure you can fix any problem with more education, including world hunger, but my comment was only about the above average complexity of spelling English. Again you are over-interpreting me.
> Except that "simplification" implies that something was done to make American spelling simpler. It wasn't. In the 1700s (when America and Britain parted company) we spelt it "center" in Britain too. It was only in the 1800s that Britain changed to French spellings. The Americans (sensibly) never followed our lead. It was the London-based dictionary writers who pushed English to use French spellings.
Thanks: this was my main question from the start a couple of posts ago, i.e., "back to original French spellings". Even when the "we" carries an "homogeneous whole" implication which risks to offend some English speakers, it's interesting. Any references on the timeline?
> Except that "the variant" implies that "British English" is a homogenous whole.
Over-interpretation again.
> Try telling that to a Sassenach, Scouser or Geordie!
Since this is (or... "was", *sigh*) a discussion about spellings, how many of these people use different spellings for the same word?
Posted Jul 6, 2014 14:56 UTC (Sun)
by Wol (subscriber, #4433)
[Link]
Is it a Loch, a Lake, or a Lough?
(Scots, English and Irish)
:-)
Cheers,
Posted Jul 6, 2014 16:55 UTC (Sun)
by mathstuf (subscriber, #69389)
[Link] (2 responses)
It's not as if "American English" is all that homogeneous either (though it might be *more*, it certainly isn't perfectly uniform). I definitely have "isms" I use that catch people up every now and then. For example, I've met people who didn't know what "quarter of 3" meant in response to "what time is it?" and I thought that one was pretty universal…
Posted Jul 6, 2014 22:46 UTC (Sun)
by ncm (guest, #165)
[Link] (1 responses)
Prepositions in all languages are delightfully fluid.
I challenge Wol's assertion that most words in English come from German. The Germanic words just get repeated a lot more than the others. Really, though, German and French both have tenuous claims. Practically all trace to Proto-Indo-european, with French, German, Latin, and Greek only squatting for a short time.
Posted Jul 6, 2014 22:56 UTC (Sun)
by mathstuf (subscriber, #69389)
[Link]
[1]Basically act as if everyone is punning all the time and react accordingly.
Posted Jul 4, 2014 7:36 UTC (Fri)
by marcH (subscriber, #57642)
[Link]
Sorry can't resist mentioning this too... British English is one variant of English where the XX/YY/ZZZZ date format makes some kind of sense :-)
Posted Jul 4, 2014 11:21 UTC (Fri)
by jezuch (subscriber, #52988)
[Link] (1 responses)
Well, that tells how much of an authority on language matters you are :)
Posted Jul 10, 2014 17:13 UTC (Thu)
by nix (subscriber, #2304)
[Link]
(As for the nonsense about 'TRUE ENGLISH' words having simple pronunciation, that's only true if you define 'TRUE ENGLISH' as 'those words that can be pronounced according to arbitrary rules I devise'.)
Posted Jul 3, 2014 9:32 UTC (Thu)
by mpr22 (subscriber, #60784)
[Link]
Posted Jul 15, 2014 11:14 UTC (Tue)
by ms-tg (subscriber, #89231)
[Link] (1 responses)
So what about patents on the following types of effects:
I've seen software patents that relate to each of these types of effects. In each case it's the fact that a concept is reduced to practice (ie "the computer is really doing this new effect now, and here's how") that is the core of the patent.
How are you squaring that circle?
Posted Jul 15, 2014 13:26 UTC (Tue)
by dgm (subscriber, #49227)
[Link]
> 1. Moving packets of data among heterogeneous and flaky hardware (networking)
The computer here is not making the effect, the network adapter is (in fact, the transceiver). What the computer does is orchestrate and control.
> 2. Simulating hardware (virtualization)
Virtualization is a pure software thing. You cannot generate sound, for instance, without a sound card, even if you virtualize it. In fact, the very fact that you can run a sound program without a sound device (and have it produce data instead of sound) should be a hint that the program is in fact much more abstract than the machine it runs on (and that it needs to produce the effect). But you will not get any sound without *real* hardware, that is clear.
> 3. Accepting a new form of human input (user interface)
Input (human or else) is just using sensors to read data from the real world. This is done by hardware. The case is the reverse of 2. In this case, there's no effect unless you add some output device.
> 4. Providing a new form of human output (user interface)
See point 2. You need real hardware for that. Say you want to print a sheet of paper. Who will do that, the computer or the printer? The software running on the computer, or the printer? The printer, of course. What the software can do is orchestrate. It provides the organization, the instructions, the "music".
> In each case it's the fact that a concept is reduced to practice
Not the concept. Except for 2 (which is a pure software thing) there's no concept, but a real hardware device. And for 2 there's no new effect.
Posted Jun 30, 2014 22:05 UTC (Mon)
by kleptog (subscriber, #1183)
[Link] (1 responses)
While that definition is fine as far as it goes, it basically suggests that a block of text on a computer is somehow abstract, whereas if it were on paper it would not be. It also suggests that if we were inside the Matrix nothing would be patentable, which seems an absurd conclusion.
But to go back to the point about abstracts ideas not being patentable, it's worthwhile to note that (AFAICT) the US Patent Law doesn't say that anywhere. Instead it's a definition the Supreme Court invented that rule themselves so they can change it. There is a trend in society to to treating digital objects more like physical ones, for example the theft of objects in games.
I think this trend will only continue, so it seems only a matter of time before someone would consider a process manipulating such objects to be patentable. After all, one of the things defining an abstract idea is the lack of utility, but I don't think you can claim compression algorithms don't have utility. I can easily see a court making the decision that patenting the idea of a compression algorithm is not allowed by being too abstract, whereas a specific algorithm might actually pass.
With some googling I found [1] which describes my point better than I could, showing the Supreme Court flip-flopping all over the place. Like I said earlier, I agree that allowing patents on e.g. compression algorithms would be bad, I'm just not convinced the Supreme Court is ever likely to some to that conclusion. We need better arguments.
[1] The Paradox of Abstract Ideas by Alan L. Durham, http://epubs.utah.edu/index.php/ulr/article/viewFile/641/462
Posted Jul 1, 2014 14:42 UTC (Tue)
by nybble41 (subscriber, #55106)
[Link]
I don't think it suggests anything of the sort. A block of text would be just as abstract on paper as on a computer. It doesn't matter whether the data is stored as electrons or magnetic domains in a computer memory or as graphite marks on paper.
> It also suggests that if we were inside the Matrix nothing would be patentable, which seems an absurd conclusion.
When you start with an absurd premise, expect an absurd conclusion.
> There is a trend in society to to treating digital objects more like physical ones, for example the theft of objects in games.
Yes, and this trend is very disturbing. Real property rights are being eroded in favor of fake ones. Patents are part of this trend. While I do think software patents can be distinguished from other patents, I wouldn't mind seeing the entire system go away. It's not like some patents are good but software patents are bad; it's a matter of bad and worse. We'd be better off without either.
Posted Jul 3, 2014 7:47 UTC (Thu)
by Wol (subscriber, #4433)
[Link]
Software is anything (abstract) that goes through the ALU. As should be obvious from what the TLA stands for - Arithmetic Logic Unit.
As for the effects, well the ALU is hardware, therefore stuff going through it will make it do things, but what it does is not software.
If you understand what is abstract, then the difference between software and hardware is easy to grasp. Think of a clock LED. It gets fed the abstract byte code 0100 (I don't know, and more importantly don't NEED to know) the physical representation thereof. That's software. The relevant LED strips light up and I can SEE (and need to know what they represent) the physical representation of a four. That's hardware.
Basically, if you can describe what it is without knowing (or caring) about the physical representation, it's software. If the physical embodiment is important, it's hardware.
Cheers,
Posted Jun 29, 2014 21:56 UTC (Sun)
by dvdeug (guest, #10998)
[Link] (28 responses)
MPEG compression uses mathematical tools, in the exact same way that every patent uses physics-based tools. But the goal of MPEG compression (to look good to humans) is not mathematically definable, and there's a lot of choices in MPEG that aren't dictated by the math. For example, what colorspace compresses the best (on real life data) without losing what humans consider important is not mathematically definable.
Posted Jun 29, 2014 23:59 UTC (Sun)
by nybble41 (subscriber, #55106)
[Link] (27 responses)
While true in one (not very useful) sense, the standard way to judge the effectiveness of a compression algorithm is to consider the size of the decompression program along with the compressed encoding. It's obviously trivial to specify an encoding that turns any one particular finite byte string into an empty string, while making any other input longer. However, you can't recreate the information from nothing; it's either in the encoding or in the decoder.
> The game with a compression algorithm is to handle real-life byte strings well, in a way that can be run quickly on the hardware. Calling that math is like saying that any patent just uses physics.
A device or process may *use* the laws of physics, but it isn't itself a law of physics. Software doesn't just use math, it *is* math.
A patent on rubber-curing, to pick one example, makes use of the laws of physics but is actually a patent on a particular way of physically transforming raw materials into cured rubber. Creating cured rubber in that way would infringe on the patent; a simulation would not, even though it references the same physical laws. A patent on a compression algorithm would be a patent on the actual math. It would be like trying to patent the laws of physics related to curing rubber, rather than the act itself.
Running quickly is largely independent of the actual hardware. Those sort of platform-specific micro-optimizations tend toward a constant factor speedup at best. Efficiency mainly comes down to picking the right algorithms for the evaluation strategy, which, as I've already pointed out, is equivalent to simplifying a mathematical proof in the corresponding logic system. In any case, the result is still an algorithm, and thus math.
> But the goal of MPEG compression (to look good to humans) is not mathematically definable, and there's a lot of choices in MPEG that aren't dictated by the math.
But those choices *are* dictated by the math. You start with a mathematical model of human perception, determined by scientific observation and statistics, and use that model to decide which information you can throw away and which information you need to keep. For each model there will be a particular set of optimum choices.
In any case, whether you painstakingly calculated the optimal choices or arrived at them by trial and error, the final result is still a mathematical algorithm.
Posted Jun 30, 2014 2:01 UTC (Mon)
by neilbrown (subscriber, #359)
[Link] (22 responses)
Interesting claim. Could you provide (a link to?) a precise definition of "math" which would allow me to assess this claim. My own imprecise definition doesn't seem to lead to this identity.
> For each model there will be a particular set of optimum choices.
Maybe it is the model that is the innovation then. I do not think that a set of scientific observations and statistics will leading inexorably to a single model. Extracting a model from observations could easily qualify as an inventive step for me.
Nor am I convinced that a given model leads deterministically to a particular algorithm, though maybe the difference there would be more in the realm of copyright than patent.
Posted Jun 30, 2014 17:24 UTC (Mon)
by nybble41 (subscriber, #55106)
[Link] (21 responses)
It's a complicated topic, and you're welcome to do your own research, but Wikipedia has a fairly good overview:
http://en.wikipedia.org/wiki/Mathematics#Definitions_of_m...
There are a number of different competing definitions, but for the most part it doesn't matter which one you choose; all the major variations (logicist, intuitionist, and formalist) are in agreement on this point.
> Maybe it is the model that is the innovation then.
Your problem is that you haven't really been arguing that software isn't math; you're arguing that (some kinds of) math should be patent-eligible. And now you're dropping the software part entirely and suggesting patents on pure mathematical models.
Posted Jul 1, 2014 2:18 UTC (Tue)
by neilbrown (subscriber, #359)
[Link] (20 responses)
As there appears (from that link) that there is no agreement about what math is, I think it would be very hard to argue that software *is* it.
But you are correct that I am really arguing that (some kinds of) math should be patent-eligible - at least in part.
The purpose of a parent is to motivate inventors to make the substantial investment required to invent something of value. I think that "mathematical" inventions (E.g. compression algorithms) can be just as valuable and require just as much investment as any other inventions.
I would generally disagree with software patents, not because the subject matter has any character which makes it ineligible, but because the economic case for patents doesn't appear to apply to the current market for software.
i.e. the extra innovation that patent rights might generate is less than the cost to innovation that enforcing those rights imposes.
Posted Jul 1, 2014 7:54 UTC (Tue)
by marcH (subscriber, #57642)
[Link]
Sounds very sensible, but even this would be still far from the most urgent problem.
It's entertaining to see engineers discussing how to fix laws, but I think it is much more productive when engineers' energy is focused on engineering: i.e., fixing the USPTO rather than the laws. The main patent problem right now is not to decide whether software is patentable or whether software is pure maths - quite entertaining but not terribly useful. The main patent problem is: how come the USPTO issues patents on "1-click buy" or "slide to unlock". Clearly, the USPTO is missing quite a few PHOSITAs, i.e., engineers. Or maybe just an massive injection of common sense. How about IT patents requiring approval from the IEEE or some other similar body? This site has a number of other, more practical, more agreeable and immediately productive fixes:
While it's always more pleasant to have a pub discussion about American constitution and/or science, chances are common sense and good old engineering could be enough to get rid of all patent trolls.
Posted Jul 1, 2014 8:23 UTC (Tue)
by dgm (subscriber, #49227)
[Link]
Why? They do quite a lot of software. I swear, I'm a former math student. In fact I learned C++ just for that.
Posted Jul 1, 2014 15:15 UTC (Tue)
by nybble41 (subscriber, #55106)
[Link] (17 responses)
Then the problem is with you, and what you find convincing. You said you wanted something specific, and yet picked the vaguest and least-serious definition on the page. If you're going to go with that definition, however, then it implies that whatever a mathematician does is mathematics. Some mathematicians do software, so software is math. (And some mathematicians walk their dogs, so walking a dog is math... I did say it wasn't a very good definition.)
> As there appears (from that link) that there is no agreement about what math is, I think it would be very hard to argue that software *is* it.
There is no agreement on a single, concise, formal definition. That doesn't mean there isn't consensus on whether particular subject areas fall under the heading of "mathematics". Nearly all mathematicians can be classified as logicist, intuitionist, or formalist, and all three definitions treat software as branch of mathematics. Theories of computation, Turing-completeness, the typed and untyped lambda calculus, proofs and their equivalence to type systems--these are all well-established as being within the mathematical domain.
> But you are correct that I am really arguing that (some kinds of) math should be patent-eligible - at least in part.
With this admission, I believe my work here is done.
Posted Jul 1, 2014 15:58 UTC (Tue)
by marcH (subscriber, #57642)
[Link] (16 responses)
It's fortunate non-mathematicians consider software as their branch too, so we can get something practical and useful and not just theories.
> With this admission, I believe my work here is done.
Bye!
Posted Jul 1, 2014 17:39 UTC (Tue)
by nybble41 (subscriber, #55106)
[Link] (15 responses)
There are, of course, both theoretical and applied branches of mathematics, just as there are theoretical and applied branches of physics, chemistry, etc. An accountant adding up column of numbers in a spreadsheet is doing math just as much as some tenured university professor working toward a proof of the Cherlin–Zilber conjecture. The programmer who writes the accounting package is also doing math, whether he/she realizes it or not. It takes all kinds. Without the theories there would be no practice; without the practice the theories would serve no purpose.
A programmer can be thought of as a mathematician specializing in applied computer science. The fact that most programmers don't concern themselves with the theoretical underpinnings of their field doesn't make the field any less a branch of mathematics.
Posted Jul 1, 2014 21:48 UTC (Tue)
by marcH (subscriber, #57642)
[Link] (14 responses)
Finally a clear definition of mathematics! :-)
> A programmer can be thought of as a mathematician specializing in applied computer science. The fact that most programmers don't concern themselves with the theoretical underpinnings of their field doesn't make the field any less a branch of mathematics.
I don't in which ivory tower you live but sorry; you clearly have no idea about what is the day job of the average software engineer. Just like for any other engineering field, (computer) science, theory and maths tend to be a extremely small part of it. We could talk about design patterns, continuous integration; bug tracking, workflows, version control, project management, teamwork, release processes, validation, etc. but I'll keep it short and say only this: probably the largest effort in coding is actually communicating with your colleagues and peers - not with the machines.
Posted Jul 1, 2014 23:45 UTC (Tue)
by zlynx (guest, #2285)
[Link] (4 responses)
But without communicating with the machines you wouldn't be a programmer. You'd be a project manager.
A large part of any job is in the communication. Only very simple jobs don't need it. But each job does have that one important part that separates it from other jobs. The plumbing, programming or scalpel cutting may be a small part of each day but without it you wouldn't be a plumber, programmer or surgeon.
Posted Jul 2, 2014 6:26 UTC (Wed)
by marcH (subscriber, #57642)
[Link] (3 responses)
There is a massive difference. When a plumber is actually "plumbing" or a surgeon operating their main task and focus and energy and time are not spent communicating with other human beings; as opposed to when a programmer is writing code. By nature software (or hardware) development is an activity orders of magnitude more "social" than plumbing or surgery (Surgery is actually one of the very least; a significant number of surgeons think they sit somewhere between God and the rest of us - but I digress)
The need to highlight this difference on the *Linux* Weekly News community web site is a bit surprising considering Linux is probably one of the largest collaborative community project ever but there you go.
I am aware that many junior programmers don't realize this until they have gathered some experience; I've witnessed this realization process many times. Sometimes it may come as a disappointment, as in: "what do you mean teamwork and code reviews, I entered this job so I could avoid human interaction?"
Posted Jul 2, 2014 19:15 UTC (Wed)
by zlynx (guest, #2285)
[Link] (1 responses)
No when a programmer is writing code the primary purpose is "does it work?" If it does not properly tell the computer how to do the job, the code is a FAILURE no matter how pretty or well documented it is.
Posted Jul 2, 2014 23:58 UTC (Wed)
by marcH (subscriber, #57642)
[Link]
Yes, very "primary".
Even if we limit writing code to "make it work" for one second: how much of a mathematical/theoretical effort does "make it work" typically involve? Unless you are 1) successfully fooling your manager and having fun re-inventing some wheels, or 2) working on some exceptional R&D project, the answer is: practically none.
Posted Jul 4, 2014 22:18 UTC (Fri)
by jubal (subscriber, #67202)
[Link]
Posted Jul 2, 2014 9:21 UTC (Wed)
by dgm (subscriber, #49227)
[Link] (8 responses)
Posted Jul 2, 2014 9:41 UTC (Wed)
by neilbrown (subscriber, #359)
[Link]
I have it on good authority that they stare at dots and dashes all day, and think.
http://devlinsangle.blogspot.com/2014/07/the-power-of-dot...
Posted Jul 2, 2014 15:03 UTC (Wed)
by marcH (subscriber, #57642)
[Link] (6 responses)
If you like you can call software engineering: "Very applied branch of mathematics where you can find plenty jobs even if you barely understood anything in maths class at school", and RTL design: "Very applied branch of physics and chemistry for people who don't need to know them - rather software instead". However I'm afraid this won't ever make much difference to patent law.
Posted Jul 2, 2014 21:53 UTC (Wed)
by nix (subscriber, #2304)
[Link] (5 responses)
I did really, really badly at school mathematics. So did an awful lot of actual working mathematicians (among whom I am not numbered, I hasten to add). I edged into maths via the lambda calculus and proofs, and was astonished to find that I found it wonderfully easy compared to procedural computation: it felt like programming.
This is, I think, not specific to me being weird. What you need for working mathematics is very much not what you need for school maths classes, and is very much more like what you need for programming: a skill at spotting patterns and rigour in reasoning. It's just that maths has several hundred years' more knowledge behind it, so there's a larger knowledge base than in programming, and it's more interconnected than random programming facts are. But to my eyes that makes maths *more* like the ideal of programming than programming is: both are all about the connections and the patterns and what you can do with them, and maths has more of those than programming does.
Posted Jul 2, 2014 22:26 UTC (Wed)
by marcH (subscriber, #57642)
[Link] (3 responses)
Sorry for the confusion.
Posted Jul 10, 2014 17:14 UTC (Thu)
by nix (subscriber, #2304)
[Link] (2 responses)
Posted Jul 10, 2014 20:27 UTC (Thu)
by marcH (subscriber, #57642)
[Link] (1 responses)
US English? Mmmm... http://www.lse.ac.uk/
Looks like the word "school" can be confusing in many places! For sure the direct translation in French can be very vague too.
Posted Jul 16, 2014 18:46 UTC (Wed)
by nix (subscriber, #2304)
[Link]
Posted Jul 2, 2014 22:33 UTC (Wed)
by Cyberax (✭ supporter ✭, #52523)
[Link]
Posted Jun 30, 2014 2:31 UTC (Mon)
by dvdeug (guest, #10998)
[Link] (3 responses)
Note also, if the MPEG algorithm is just math, then so is any song. An MP3 player is a function of, say, input MP3 and output format, and can be thus partially specialized to be a Tom's Diner player, which is no more or less a mathematical function then the original. You can summarize that way, but society has chosen not to, and hence it's not a productive argument. Society will not end all copyrights because they are in some sense math, and thus will so no reason to end all software patents because they are in some sense math.
Posted Jun 30, 2014 3:55 UTC (Mon)
by mgb (guest, #3226)
[Link] (2 responses)
Posted Jun 30, 2014 4:04 UTC (Mon)
by dvdeug (guest, #10998)
[Link] (1 responses)
Posted Jun 30, 2014 4:32 UTC (Mon)
by dlang (guest, #313)
[Link]
algorithms are one of the abstract idea categories that are not supposed to be patentable, this is why so many software patents recite the characteristics of a general purpose computer, it's to try and make things seem as if it's a real machine being created
copyrighting software/music/books is seen as pretty reasonable because it protects that way of doing things.
getting the rights to all possible ways of doing something may be valid for vary narrowly defined cases (the RSA algorithm for example, which didn't block DSA, ECC or other approaches to Public Key Encryption)
The problem is that 99.99+% of software patents are like RSA getting exclusive rights to anything that looks like Public Key Encryption, not just their one way of implementing it. (or worse, they are on ideas that are trivial)
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
"Software is also effects."
That's hardware.
Software patents take a beating at the US Supreme Court
>
> That's hardware.
Software patents take a beating at the US Supreme Court
>> That's hardware.
> Really? That would seem to define most software as hardware.
Software patents take a beating at the US Supreme Court
> abstract inputs to a set of abstract outputs
Software patents take a beating at the US Supreme Court
> The *operation* of anything to change state. So, for example, if one had a piece of hardware that emitted a vibration. And a software program that could caused it to emit certain patterns of vibrations...
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software is a language, had Shakespeare patented his works no one could speak English.
Um, no. Software is written in languages. Languages themselves are abstract concepts. You cannot point to a piece of software and say 'that is C': you can merely point to programs that accept C, or produce it, or analyze it.
What I meant is that software is the Language to talk to computers, not any particular language (more the reason software is abstract). If computers get smart we might be able to program them using the Natural Language or the language of humans (NLP research already is getting very successful).
Software patents take a beating at the US Supreme Court
Also I chose Shakespear because according to the Wikipedia:
In Shakespeare's day, English grammar, spelling and pronunciation were less standardised than they are now,[166] and his use of language helped shape modern English
Software patents take a beating at the US Supreme Court
Wol
Great Vowel Shift
2. Great Vowel Shift
3. American-like spelling simplification
4. *Back* to French spellings because of some "further influence"?
Words: autumn, courgette, aubergine, toilet, caravan, aluminium, reservation, ...
Sounds: zebra, garage,...
http://www.bg-map.com/us-uk.html
http://www.iloveenglandstudyabroad.com/british-vs-america...
Great Vowel Shift
> 2. Great Vowel Shift
> 3. American-like spelling simplification
> 4. *Back* to French spellings because of some "further influence"?
Wol
Great Vowel Shift
Great Vowel Shift
Hear, hear! I still have this problem after living in the US for more than 4 years.
Are there spelling contests (as in the US) in Ukraine? Russia? Italy?
Great Vowel Shift
Great Vowel Shift
Great Vowel Shift
Wol
Great Vowel Shift
Great Vowel Shift
Wol
Great Vowel Shift
Great Vowel Shift
Great Vowel Shift
Great Vowel Shift
Great Vowel Shift
Great Vowel Shift
I can assure you that it's not the Victorians that decided to write "colour" and "centre"; those spellings were well-established at the time Noah Webster decided they were silly, and his American Dictionary of the English Language (which used "color" and "center") went to press in 1828, when HRH Princess Alexandrina Victoria of Kent was nine years old.
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
>> The *operation* of anything to change state. So, for example, if one had a piece
>> of hardware that emitted a vibration. And a software program that could caused
>> it to emit certain patterns of vibrations...
> In that case, the effect is entirely due to the hardware. Software doesn't cause the
> hardware to emit vibrations; the hardware emits vibrations in accordance with
> software instructions. The instructions by themselves are passive; they don't
> *cause* anything. They just sit there in memory until the hardware acts on them.
> To draw an analogy, sheet music doesn't cause a singer to produce a particular
> note; the singer produces the note in accordance with the instructions. The cause
> is the singer, not the sheet music.
1. Moving packets of data among heterogeneous and flaky hardware (networking)
2. Simulating hardware (virtualization)
3. Accepting a new form of human input (user interface)
4. Providing a new form of human output (user interface)
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Wol
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Why should we devalue our work by saying that it isn't good enough to be patented?
And I think this is because the computing landscape moves so fast. I would have no objection at all to a 2 year patent grant of any "non-obvious" software idea. Maybe even 4 years. More than that and the parent will slow things down more than it speeds things up.
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
By nature software (or hardware) development is an activity orders of magnitude more "social" than plumbing or surgery (Surgery is actually one of the very least; a significant number of surgeons think they sit somewhere between God and the rest of us - but I digress)
A little nitpick: surgeons might be not exactly social and may think they're gods or godesses incarnated, but all of them have to be very efficient communicators. Surgery is a teamwork in the strictest sense of the word and requires constant and effective communication – there are no lone stars in the operating theatre, even if they enjoy behaving in a sligthly operatic style.
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court
Software patents take a beating at the US Supreme Court