| |||||||||||
Courgette meets a dangerous (Red) BendCourgette meets a dangerous (Red) BendPosted Nov 2, 2009 21:12 UTC (Mon) by AlexHudson (subscriber, #41828)Parent article: Courgette meets a dangerous (Red) Bend
Really don't see Google licensing this in the way Red Hat have managed in the past. Would be a nice surprise.
Personally, I think this is a bad patent - there's no amazing technical solution here; if you ask the right question this solution kind of pops out automatically. Patents are supposed to protect ideas for ways of solving problems people recognise as being very tough / impossible.
(Log in to post comments)
Courgette meets a dangerous (Red) Bend Posted Nov 2, 2009 22:41 UTC (Mon) by socket (subscriber, #43) [Link] I think an argument can be made that any invention becomes obvious if you can ask the right question to prompt it and understand the basic prerequisite knowledge.There are likely a large number of patents in the world that aren't obvious to me, but I attribute that to the fact that (for example) no material-science related patent is obvious to me, because I'm not a chemist. But show me a software patent, and it will seem obvious to me. Thus, the "rule" that patents shouldn't be obvious to the typical practitioner with the relevant experience. Sometimes the interesting part of an invention isn't the process used (the content of the patent) but simply the problem it's trying to solve. The solution may be obvious, but realizing the fact that the problem existed in the first place is an insight in and of itself. This particular patent seems to me pretty obvious, once you start thinking about how to do compression: Compression relies on finding repeated patterns in data, and abstracting away that information - that's obvious. The result of compiling code after making changes is an object file with systematic changes. Those changes can involve pointer address changes - and this is obvious to anyone who programs much in C. If the typical C programmer were asked to make a compression program for executable patches, this seems to me that looking into systematic changes like pointer addresses is not just obvious, it's the low-hanging fruit. Nearly every software patent I've heard of is either an analogue of an obvious, everyday process in the physical world, or the description of a mathematical algorithm in terms of software. Simply adding "...with a computer" to an obvious, ordinary process doesn't make it any less obvious. The patent system would be made consistent with itself either by removing the exclusion of mathematical algorithms from patentability (a bad idea, in my opinion) or by adding an exclusion for software-related patents.
Courgette meets a dangerous (Red) Bend Posted Nov 2, 2009 22:58 UTC (Mon) by proski (subscriber, #104) [Link] It's actually a good question for a job interview. How would you implement a binary patch algorithm? What would you do to reduce the patch size? I imagine a good applicant would come with a similar idea, perhaps without implementation details.Of course, it will take some effort to implement and validate the idea, but it should be much easier than validating e.g. a drug or a rocket engine.
Courgette meets a dangerous (Red) Bend Posted Nov 3, 2009 8:58 UTC (Tue) by michaeljt (subscriber, #39183) [Link]
> It's actually a good question for a job interview. How would you implement a binary patch algorithm? What would you do to reduce the patch size? I imagine a good applicant would come with a similar idea, perhaps without implementation details.
Or for the patent office? Like finding twenty people "skilled in the field" with no knowledge of the patent, and ask them to sketch a solution to the problem with fifteen minutes time at their disposal?
Courgette meets a dangerous (Red) Bend Posted Nov 3, 2009 13:40 UTC (Tue) by gbutler69 (subscriber, #54063) [Link]
Ask 100 RANDOM software developers that have a college degree and 5 years of experience for
a solution to this problem. I highly doubt more than 3 or 4 would come up with this idea in a reasonable amount of time. No, it is not OBVIOUS. If you think it is, you are either lying to yourself, or you are the 3 or 4 % exceptional of the 100 who would come up with this idea.
Courgette meets a dangerous (Red) Bend Posted Nov 3, 2009 14:49 UTC (Tue) by michaeljt (subscriber, #39183) [Link]
>Ask 100 RANDOM software developers that have a college degree and 5 years of experience for a solution to this problem. I highly doubt more than 3 or 4 would come up with this idea in a reasonable amount of time.
Wouldn't that be an answer as well? I can't say that I'm a fan of patents in general (software or otherwise), but at least it would be a more honest test of whether it fits within the current rules.
Courgette meets a dangerous (Red) Bend Posted Nov 4, 2009 11:47 UTC (Wed) by ebiederm (subscriber, #35028) [Link]
My reaction when I read about Courgette the first time was interesting someone got around to implementing that.
The usage in Courgette also seems to be an independent reinvention.
Perhaps I have missed some detail but this seems like a battle over
Courgette meets a dangerous (Red) Bend Posted Nov 5, 2009 6:51 UTC (Thu) by nevyn (subscriber, #33129) [Link]
Maybe true, but a better question is if only 3-4 people came up with the idea (out of your "random" 100) ... does that qualify it for being patentable? Is the bar of "obvious" really so low that you could realistically expect 1000s of people to be able to independently invent it at google/MS/IBM/etc.
Yes. Posted Nov 5, 2009 13:16 UTC (Thu) by gbutler69 (subscriber, #54063) [Link]
Obvious to me would be at least 30% - 50% of those polled skilled in the art would come up
with the idea and method in a reasonable amount of time. Everything is Obvious in hind-sight. Think about even simple things like the derivation of the quadratic equation. Once you know it, it is obvious. For *most* people though, it is not obvious until shown. Now, admittedly, this is not the kind of thing that is patentable (mathematical algorithms being specifically excluded from patent protection) but, it does demonstrate the point I'm trying to make.
Frankly, I think the only problem with patents is that they are awarded for too long (at least in
Patent Law, like all other laws, is nothing but an agreement between all the members of the
Honestly, we all must learn to take more control over the laws of our society instead of just
Sticking your fingers in your ears and yelling at the top of your lungs, "NA NA NA NA" won't
I know that I often find myself frustrate, angry, and even bordering on wanting to lash out
Besides, in 5 or 10 years we're going to run out of oil and then the whole world-wide economy
Re: Yes. Posted Nov 6, 2009 5:15 UTC (Fri) by nevyn (subscriber, #33129) [Link] Obvious to me would be at least 30% - 50% of those polled skilled in the art would come up with the idea and method in a reasonable amount of time. Well, personally, I would disagree ... if 3-4% of a random sample in an "art" would come up with the idea, when asked to solve a problem, then I fail to see why 1 person/entity should be able to stop upto 9 million people from using it (assuming every .us person could be taught said "art", that'd be 9 million). Patents aren't supposed to be a lottery, they are supposed to solve the problem of sharing when only a very small number of people would ever be able to solve a problem. Thus it's worth giving you a monopoly as an incentive to share. When you are anywhere close to 1%, you are saying at least 1 person in any public company could find the solution ... that's just not rare, IMO. Of course atm. patents seem more like if less than 99% of a random sample would come up with it, then it's patentable. So it's all academic. Honestly, we all must learn to take more control over the laws of our society instead of just bitching about them. If you can convince enough people you are right, you can win. Do you have any examples of that working?
Courgette meets a dangerous (Red) Bend Posted Nov 12, 2009 17:54 UTC (Thu) by tbrownaw (guest, #45457) [Link] Ask 100 RANDOM software developers that have a college degree and 5 years of experience for a solution to this problem. I highly doubt more than 3 or 4 would come up with this idea in a reasonable amount of time. No, it is not OBVIOUS. Would the solution to FizzBuzz be obvious under this test? Or how to expand (from=1/1/2000; to=1/3/2000) into [(when=1/1/2000), (when=1/2/2000), (when=1/3/2000)] with SQL? This second one is actually an interview question here, I'd guess the candidates we get (with degrees and mostly 5+ (some supposedly 15+) years experience) have a (probably low) single-digit percent hit rate. Does this mean that a solution should be patentable?
Mathematical algorithms Posted Nov 2, 2009 23:37 UTC (Mon) by man_ls (subscriber, #15091) [Link] I think an argument can be made that any invention becomes obvious if you can ask the right question to prompt it and understand the basic prerequisite knowledge.When I hear this argument about obvious patents I always think about the FFT algorithm, which seems to me the most unintuitive algorithm of all times. But it is after all a mathematical algorithm, and these should not be patentable -- not any more than physical laws. Today I learned on Wikipedia that the prior art for the FFT goes back to Gauss in 1805. Go figure.
Mathematical algorithms Posted Nov 3, 2009 4:13 UTC (Tue) by JoeBuck (subscriber, #2330) [Link] Gauss discovered the FFT in 1805, so I think any patent would have expired by now. :-)
Mathematical algorithms Posted Nov 3, 2009 7:31 UTC (Tue) by man_ls (subscriber, #15091) [Link] OTOH I think the prior art would still be valid today :D You might as well want to patent the Eratosthenes sieve to find out prime numbers.
Mathematical algorithms Posted Nov 3, 2009 18:30 UTC (Tue) by Trelane (subscriber, #56877) [Link]
No, no, no. Did he discover it _on_a_computer_?! If not, then it's clearly novel!
Mathematical algorithms Posted Nov 4, 2009 10:00 UTC (Wed) by Kluge (subscriber, #2881) [Link]
Actually, according to that link, Gauss didn't publish the algorithm (like so much of what he did), so by my understanding of patent law, it doesn't count as prior art.
Mathematical algorithms Posted Nov 9, 2009 15:48 UTC (Mon) by gmaxwell (subscriber, #30048) [Link] I see your hope and raise you 6,859,816, claim 1 of which appears to read on odd-radix Cooley—Tukey.Of course— if that is what this patent is actually doing, it is patently invalid without a shred of hope at being enforceable. But it does show that the presence of a clear description of your algorithm in the patent database doesn't mean much without a fair amount of costly analysis.
Courgette meets a dangerous (Red) Bend Posted Nov 3, 2009 1:03 UTC (Tue) by ikm (subscriber, #493) [Link]
Well, the thing with chemistry (and drugs in particular) is that they do require a lot of tests and clinical trials (and hence a lot of money and time) to prove they do actually work with the side effects all being known and well-studied. As you can see, this isn't true for software. Software patents is total cheating. Like, *total*. They are ALL obvious.
Courgette meets a dangerous (Red) Bend Posted Nov 3, 2009 3:34 UTC (Tue) by drag (subscriber, #31333) [Link]
You can't copyright drugs, unlike software. So the laws that restrict the
use of drugs only lasts a few years and the trade off of the patent is that it becomes public domain invention.
While the patent system is probably abused in the case of drugs, having the
Courgette meets a dangerous (Red) Bend Posted Nov 3, 2009 3:40 UTC (Tue) by ikm (subscriber, #493) [Link]
Well, that actually was my point, too. Patents clearly work for drugs. But not for software (bummer!:)
Courgette meets a dangerous (Red) Bend Posted Nov 3, 2009 11:45 UTC (Tue) by niner (subscriber, #26151) [Link]
No, they don't even work for drugs. And they are not needed there either.
The pharmaceutical industry spends a multitude of it's R&D investments on advertising while basic research nowadays is nearly completely state funded and happens at universities.
It's just a fairy tale that pharma companies would stop doing the little
Courgette meets a dangerous (Red) Bend Posted Nov 3, 2009 11:57 UTC (Tue) by ikm (subscriber, #493) [Link]
But if basic research is really state-funded and done at universities, I don't see where patents come to play at all.
Courgette meets a dangerous (Red) Bend Posted Nov 13, 2009 14:14 UTC (Fri) by cowsandmilk (subscriber, #55475) [Link]
ever heard of Bayh-Dole? http://en.wikipedia.org/wiki/Bayh-Dole
Courgette meets a dangerous (Red) Bend Posted Nov 3, 2009 12:02 UTC (Tue) by csigler (subscriber, #1224) [Link]
> the little R&D that they have to do anyway
I'm sorry, but any credibility you may have had was trashed by these, your own words.
Courgette meets a dangerous (Red) Bend Posted Nov 3, 2009 13:43 UTC (Tue) by niner (subscriber, #26151) [Link]
Can you tell me how those words could do that? As a non-native speaker of
English I may just have used the wrong words.
Or maybe I shortened my argument too much. I meant, that those companies
The point is: yes, they collectively spend some 100 Million USD on R&D
Even if there were no patent protection anymore for their research
Courgette meets a dangerous (Red) Bend Posted Nov 3, 2009 16:30 UTC (Tue) by csigler (subscriber, #1224) [Link]
Your claim is grossly exaggerated. This publication:
http://www.sciencedaily.com/releases/2008/01/080105140107...
quotes a York University study. Their work shows pharmaceutical companies spend almost twice as much on advertising as R&D. Frankly, I would expect that ratio to be reversed, as high-tech R&D costs are, well, high. (I have first-hand experience, having worked in similar R&D and piloting operations in years past.)
Your extravagant claim is "$100 million" for R&D and "billions" for advertising, which is a ratio of 1:20 or less. The NYU study says the ratio is not even as low as 1:2. This is why your statements lack credibility. They don't match up with published facts.
Courgette meets a dangerous (Red) Bend Posted Nov 3, 2009 16:32 UTC (Tue) by csigler (subscriber, #1224) [Link]
Sorry, my error. For "NYU" above, substitute "York University."
Big Pharma Posted Nov 3, 2009 16:31 UTC (Tue) by tialaramex (subscriber, #21167) [Link]
Do you have a source for those two figures you've quoted? Because the last I remember it was suggested that R&D spend is about the same as advertising, still a long way from "poor us, we spend every cent on R&D and now you want to take away our patents" but a lot more than pretty much any other industry.
The idea that all the real work is done "at the universities" may be superficially correct, but it involves a bit of sleight of hand. You've gone from counting how much is spent /by/ the companies to whether work is done /at/ the universities. In practice drug research _at_ universities is heavily funded by these same big pharmaceutical companies. If they decided to cut that spending, you'd see big job losses.
In fact this is a major topic of discussion. If the researchers were funded by government (= higher taxes to pay for it) they'd have no reason to do some of the dubious things they do today to ensure they keep their funding from big pharmaceutical companies. For example, deciding not to publish results from an experiment which shows no difference between an old drug and a new drug. Or changing the measured outcome of a controlled trial after the data is collected, in order to have a positive result rather than an equivocal one.
Big Pharma Posted Nov 4, 2009 10:17 UTC (Wed) by Kluge (subscriber, #2881) [Link]
As far as I can tell, the vast majority of university research is funded by the government, especially basic research. The numbers might be different if you're talking about clinical trials on drugs.
I'm not sure about the specific abuses you're referring to; certainly universities have generally failed to establish sufficiently strict codes of conduct regarding grants and contracts with pharma. I believe that's changing, though.
Big Pharma never actually cures anyone Posted Nov 5, 2009 9:46 UTC (Thu) by pflugstad (subscriber, #224) [Link]
As an aside, my other impression with Big Pharma is that the tend to spend virtually ALL their R&D on products that don't actually cure anything, but rather, put the user on an continuous dose program (think Lipitor, etc). While drugs like Lipitor may actually help out quite a bit, they really seem to be treating the symptoms and not the underlying disease. So there is actually a negative incentive to actually do research into a drug that might cure the underlying disease (almost an inventors dilemma). For diseases that actually could be cured, there's zero non-govt/university research in it because there IS NO MONEY IN IT.
At which point, it becomes obvious that Big Pharma has zero incentive to actually cure anyone of anything. So at this point, it's pretty clear to me that anything Big Pharma says about patents and how much they spend on IR&D is totally irrelevant and frankly, disingenuous.
Courgette meets a dangerous (Red) Bend Posted Nov 3, 2009 18:17 UTC (Tue) by blitzkrieg3 (subscriber, #57873) [Link]
When you have a number of companies patenting the very chemicals that make up our DNA as they discover them just because they _might_ turn out to be drugs, I believe you have a failure of the patent system.
Courgette meets a dangerous (Red) Bend Posted Nov 3, 2009 10:03 UTC (Tue) by AlexHudson (subscriber, #41828) [Link]
I'm not sure I'm in total agreement with what you said, but I do agree that hindsight is a particularly poor judge of these matters. What I meant by "asking the right question", though, was that many software patents tend to be pre-empted by the requirements set forth by the inventors. In many ways, it ends up not causing huge issues because there tend to be many ways of skinning a cat, but I'm not sure it's easily dealt with in the current system.
As an aside, I can think of one patent which I think of as being quite clever: the selective quantization method used in MPEG encoding (e.g. MP3) to compress data without impacting the resulting sound. I think it's a nice tie-up between mathematical theory and biological reality, and it's a useful contribution to our knowledge of the physical world (the fact you pretty much have to do it in software is neither here nor there for me).
Courgette meets a dangerous (Red) Bend Posted Nov 5, 2009 18:48 UTC (Thu) by jrigg (subscriber, #30848) [Link]
> As an aside, I can think of one patent which I think of as being quite
> clever: the selective quantization method used in MPEG encoding (e.g. MP3) > to compress data without impacting the resulting sound.
It might be clever but it most definitely affects the sound. Only lossless compression algorithms (eg. FLAC) have no effect on sound quality. Sorry if this is OT, but as an audio engineer I find this misconception a little irritating.
Courgette meets a dangerous (Red) Bend Posted Nov 9, 2009 16:12 UTC (Mon) by gmaxwell (subscriber, #30048) [Link]
In an absolute sense digitization also "affects the sound".
But— you protest— that well executed digitization with sufficient bit-depth and sampling captures the information with precision exceeding the noise floor and band-pass of human hearing, and that under carefully conducted double-blind testing even the best listeners can not discern a difference.
Quite right.
But the same is true of lossy compression: At high enough rates with well well enough done methods it exceeds the limits of human perception and produces results which are not ABX-able. It's often not well done, it often is used at fairly low rates, and because it uses more sophisticated techniques it can fail in subtler ways...
This is relevant because your model of lossless=perfect, lossy=bad brings about unreasonable conclusions. Which would produce a more accurate experience: Stereo lossless audio or surround sound at the same (high) bitrate using lossy compression?
Pulling this back on topic, perceptually weighed quantization *is* obvious to a practitioner in the art, or at least it has a very long history of incremental development stemming back to the early vocoder speech crypto devices from the WWII era, the weighing filters used on analog telephone lines and for analog noise shaping (dolby a).
Like in many other areas the underlying technology needed for MP3 existed for a long time before computers became so stupidly fast that what would have seemed like a joke (executing 152 288point complex/complex FFT's per second for the MDCTs in MP3) became completely reasonable. The same is true for some 'recent' innovation in asymptotically optimal error correcting codes.
Courgette meets a dangerous (Red) Bend Posted Nov 12, 2009 19:06 UTC (Thu) by jrigg (subscriber, #30848) [Link]
> At high enough rates with well well enough done methods it exceeds the limits of human perception and produces results which are not ABX-able.
I have yet to hear such a thing from mp3, but I agree it would be possible using a good enough method.
> This is relevant because your model of lossless=perfect, lossy=bad brings about unreasonable conclusions.
Actually my model for compression is: audible=bad, inaudible=good.
It can be argued that the principle of perceptually weighted compression is obvious to a practitioner in the art, but then I think the same applies to many patents. In practice the criterion often seems to be whether or not it is obvious to the patent examiner.
Courgette meets a dangerous (Red) Bend Posted Nov 3, 2009 16:16 UTC (Tue) by NAR (subscriber, #1313) [Link] I think an argument can be made that any invention becomes obvious if you can ask the right question to prompt it and understand the basic prerequisite knowledge.My experience with "software research" was that the emphasis was on asking the right questions, that was the hard part - the actual implementation is mostly straight forward. The problem with patents is that in some cases they cover the actual question, not just the implementation, so they rule out any other implementation for the same problem. On the other hand, why does Google bother with this hacking? Wouldn't it be easier to split the exectuable into many DLLs, so if there's a bugfix, only the changed DLL should be delivered? For example the JavaScript engine could be separated, the image displaying libraries (libpng, etc.) could be separated, etc.
granting of patents for obvious solutions to non-obvious problems Posted Nov 5, 2009 12:10 UTC (Thu) by pjm (subscriber, #2080) [Link]
> Sometimes the interesting part of an invention isn't the process used (the content of the patent) but simply the problem it's trying to solve. The solution may be obvious, but realizing the fact that the problem existed in the first place is an insight in and of itself.
Such a realization may be “an insight”, and the result may in some cases be a useful contribution to society, but I believe the pertinent question is rather to characterize whether granting such patents is likely to advance or hinder the art (or whether it's likely to improve or worsen society).
In the case of the patent discussed in the article, it appears that the idea was independently conceived of, put into use, and publicized, without the motivation or other help of patents. The only argument I can think of that patents might have helped advance the art in this case would be to argue that Red Bend Software employees were funded in part by the hope that they would be granted this patent, and that they would not have been funded if the threshold for granting patents were such that patents were not granted for obvious solutions to non-obvious problems, and thus that the idea would have been delayed by a couple of years in its application. I don't find this argument very convincing. Perhaps someone more knowledgable can offer testament or evidence of this, or perhaps someone else can provide a better argument (though likely this isn't a good forum to seek one).
Similarly, in the case of the previous patent where I heard this argument made (viz. Amazon's “one-click” patent), my impression as an ignorant outsider is that Amazon would have conceived of, implemented and popularized the idea whether or not it was patentable.
(Two obvious arguments that granting such patents hinders the industry are that doing so incurs significant legal costs, and often hinders rather than promotes application of the idea.)
granting of patents for obvious solutions to non-obvious problems Posted Nov 5, 2009 19:02 UTC (Thu) by dark (subscriber, #8483) [Link] The original argument for the patent system is that it promotes publication. Solutions are made public rather than kept as trade secrets. Obviously, obvious solutions to non-obvious problems don't need this mechanism at all, since it will not be possible to keep the solution secret once the problem is known.
It's borderline-obvious Posted Nov 12, 2009 10:34 UTC (Thu) by edmundo (guest, #616) [Link]
I would expect at least some interview candidates to come up with this solution to the problem.
The general principle of "transform your data into a different form which standard compression algorithms can handle better" is very well known. For example, if you want to compress text that contains a lot capitalisation you might want to first transform it like this:
some lower case AND SOME UPPER CASE lower again
Then apply some standard compression algorithm, which can now spot the repeated words "some" and "case". I don't know whether this transformation works in practice, but it's an obvious thing to try, and I would claim that the Courgette transformation is only slightly less obvious.
Courgette meets a dangerous (Red) Bend Posted Nov 3, 2009 11:02 UTC (Tue) by epa (subscriber, #39769) [Link]
Dividing software patents into 'good' and 'bad' is not really a game worth playing. The legal system or the patent office cannot make such value judgements. They can only deal with clear rules such as 'programs for computers are not patentable', and sometimes manage to mess up even that (see the EPO).
What is a good patent Posted Nov 4, 2009 21:39 UTC (Wed) by rvfh (subscriber, #31018) [Link]
The best patent is not one that solves an obscure problem in a complicated way, quite the opposite! The best patent is simple, and solves a very common problem in the best ever way, so that everybody that has the problem (and that's a lot of people, as I said it is a very common problem) really wants to license the patent rather than using a free, inadequate solution.
|
|||||||||||