« NOW AVAILABLE: First Amendment Institutions | Main | The Deference Question, Part I: Why Care? »

Tuesday, November 27, 2012

Two Worlds of Software Patents

I recently participated in Santa Clara Law School's great conference on "Solutions to the Software Patent Problem." The presentations were interesting and thoughtful, and...short! A total of 34 presentations in one day, including some Q&A from the audience. Op-Eds from the conference are continuing to appear at Wired Magazine's blog, and Groklaw has a fairly thorough article summarizing the presentations.

I want to focus this post on an epiphany I had at the conference, one that is alluded to at the end of the Groklaw article. In short, there appear to be at least two world views of software patenting (there is probably a third view, relating to natural rights and property, but I'm going to put that one to the side). More after the jump.

On the one hand, you have the utilitarians, who believe that the costs of patenting might be worth the benefits of patenting. Or maybe they aren't, but that's the important question to them: to what extent does allowing software patent drive innovation? The Groklaw article implies that this group is primarily large corporate interests, but I think that's too restrictive. For example, I'm unabashadly a member of this world view, and my affinity is toward start-ups.

On the other hand, you have what I'll call the friends of free software (more fully called FOSS - Free and Open-source Software). These individuals believe that software is thought, and math, and that no one can own it. I've found that some take this view to the extreme - they have no problem with a circuit that performs the same thing as software, so long as it is performed in hardware. Members of this group believe that software patents should be unpatentable as a matter of principle, and that by allowing any kind of software patenting bad things will happen to individual programmers, to free software, and in the world generally. As further evidence that the divide is not just about large corporate interests, there are plenty of people who subscribe to this world view that started large successful companies.

Now, here is the epiphany - I belive that bridging these two worlds is possible if one believes that any software patent should issue. (If you agree that software patents can never satisfy utilitarian ends, then you can bridge the worlds. Benson Revisited  by Pamela Samuelson is a great example of such a bridge.)

Believe me, I tried to make the leap. I wrote a lengthy post at Groklaw that garnered more than 1300 comments where I tried to better understand the free software view and they tried to understand mine.

Surely, I thought, they might see that there are some lines that can be drawn that would allow for inventive software innovations. Surely, I thought, we can discuss some tweaks that would help alleviate the deleterious effects of low quality patents but save the system for one good software patent.

Surely, they thought, I would see how software patents are a bane to society, and must just go. Surely, they thought, I would see that there is no such thing as a good software patent.

The problem is that the goals of each world view are just too different. The following exchange from the Santa Clara conference between John Duffy and Richard Stallman drives the point home. I'm paraphrasing the statements, of course:

[Stallman's keynote]: Companies don't need software patents to innovate - just look at the rise of Google. [later] My proposal is that we can enforce software patents in standalone devices but not in general purpose computers.

[Duffy's talk]: I'm glad Stallman points out that software companies don't need patents - I think we agree on a solution. My proposal is that if an inventor is not induced to invent because of the prospect of a patent, then the invention is obvious and no patent should issue.[later]Stallman's proposal, though, is a kludge - a patch on the system rather than an elegant solution like redefining obviousness.

[Stallman in response to Duffy]:It doesn't matter if the patent induced the invention, it is still a bad patent. It may actually be worse, because now it can't be invalidated. My solution is not a kludge, because it handles the very real problem of software patents and eliminates it.

[Duffy]: But you have to look at the ex ante incentive to invent. If we don't allow patent enforcement, inventions might not happen that would have happened with the patent system.

[Stallman]: It's OK if we don't get those inventions. Maybe they will be developed, maybe they won't, maybe they will take longer, but the harm to any future software programmer/company is never justified by encouraging that investment with a patent.

And there you have the core of the problem. Utilitarians like Duffy (and me) believe that it is worth driving the ex ante incentive to innovate, but trying to hone the system to minimize collateral damage. Free software folks like Stallman (and probably 99% of Groklaw readers) believe that the collateral damage never justifies the ex ante incentive in a practical way.

You can see the core of these arguments in the debate about whose invention is elegant and whose is a kludge. Duffy believes that tweaking inducement to invent is elegant because that's what utilitarianism is all about. Just barring patents on general purpose computers is a patch, because there might be valuable innovations in the use of general purpose computers that are worth encouraging. Investment in standalone software might decline if there is not general purpose application at the end of the rainbow, especially in the age of smartphones.

On the other hand, Stallman believes that barring enforcement on general purpose computers is elegant, because it eliminates the most harmful effects to programmers. He believes that changing obviousness is a kludge, because it refuses to acknowledge that even the patents that come from the new rules will be bad for society. As Stallman commented to me after the conference: "There may be weak patents, and there may be strong patents, but they are all bad patents."

So, where does that leave us? I don't know, but I have to think it is helpful to understand why we can't seem to understand each other. I'm not sure where it leaves the utilitarians. They seem to be winning in policy circles, as this recent speech by PTO director David Kappos shows, but utilitarians can't even seem to agree among themselves the best course of action with software patents. Perhaps this recognition will aid those with the free software view to hone their arguments in a way that will get more policy traction - by making their same important points, but somehow framing them in a langauge utilitarians will hear. Samuelson's Benson Revisited article is a good example.

UPDATE: Thank you all for the thoughtful comments. It's really the only way to know anyone is reading at all. Because there are some common themes in the comments, I thought I would respond here rather than a long comment. 

Theme 1: Software is just ideas and math, the debate isn't utilitarian because you never get to patenting in the first place. I would submit that a) this is evidence of a separate world view (and one widely shared - by calling it separate, I don't mean to disparage it). However, it also reveals an important definitional divide -one I thought about putting into the main post, but then decided against as it ran too far afield of my point. Maybe I was wrong about that. The question is what is software. One comment below essentially says, "Well, of course circuits are patentable and software isn't. Software is just abstract math." The problem is that most patents don't claim just the abstract math part. They claim "The steps of making A happen by doing X, Y, and Z." Once you view a patent that way, a circuit and software are equivalently infringing if they do X, Y, and Z - they are the same - a means to some other patented end. This is another bridging difficulty.  FOSS folks may think utilitarians are not hearing their points about math, but reading some of these comments (asking me to "wake up," for example) sure makes me feel like my points about how process patents work are not being understood.

Theme 2: Software patents harm innovation. This might be true, but you have to look at the benefits on the front end. It's an ex post v. ex ante thing. Utilitarians will agree that software patents harm innovation on the backend if they can get the benefits on the front end.  This leads to...

Theme 3: Software patents don't help innovation on the front end. This is facially a utilitarian argument, I will admit. Commentors ask, "show me the evidence of such benefits." And that is what utilitarians debate about - whether the evidence is there. 

But here's the thing - and the reason why I put in the Duffy/Stallman exchange that was so eye-opening for me. Duffy was quite clear: If there is no software patent that would have been induced by the patent system, then fine, there should be no software patents. He thought there was agreement. But Stallman was quite clear that no, even if there was such a patent that withstood that test, that survived the evidence, it would still be bad and should still be unenforceable. That was the point of my post. For all those people who say there is no evidence, I ask you: what if that evidence came? What then? Would you change your mind? I suspect most would say no.

Posted by Michael Risch on November 27, 2012 at 10:28 PM in Intellectual Property | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference Two Worlds of Software Patents:


Hello, I think it is an Attractive section of content. I just stumbled upon your web site and in accession capital to assert that I get actually enjoyed account your blog posts.

Posted by: lisa | Sep 17, 2019 9:25:25 PM

"Further, I don't know what point you think I'm pushing. I know you are never ever going to agree with my position. "

I appear to be communicating badly. The sentence starting with "I believe that pushing a point..." is intended to refer to the Stallman approach. I could 100% agree with Stallman that software patents are inherently broken and should never be issued ever and that wouldn't make a lick of difference because I rate the chances of that ruling occuring as pretty low.

I DO agree with Duffy's position (that we need to enforce every legal angle to clear up the really bad patents) not because it's correct from a technical perspective but because, regardless of whether or not it's right technically, it's more accessible legally. The legal system believes, and has a vested interest in believing, that putting software on a generic computing platform that causes no hardware interactions is creating some new machine for practical purposes. We, as technical authors, have to work within the legal framework we have available.

From my perspective the perfect solution would be some way of opting out of the madness other than "have no presence in the US".

Posted by: JonCB | Nov 29, 2012 4:56:42 PM

PS no, I didn't hear your Santa Clara talk. It sounds like your position has evolved since you last posted on Groklaw, for which I commend you.

Posted by: nartreb | Nov 29, 2012 3:24:06 PM

Show me one empirical study that shows a benefit from software patents?

Actually, just show me one good software patent.

By "good patent" I mean actually deserving of a twenty-year monopoly, under practically any criteria, as long as the criteria are pragmatic - we can use yours. Mine would be: -would not have been independently invented and disclosed within twenty years, given the state of the art at the time the patent was filed.

Posted by: nartreb | Nov 29, 2012 3:22:28 PM

Question: Where is the utility to a field where the practitioners are not allowed to read patents because it would automatically induce willfull infringement, and thereby triple the damages?

As a practitioner in the field of software development I cannot read any software parents due to the legal problem of willfull infringement; therefore they have no utility to me.

The utility of a software patent is not for the software development field; but to the existing gorillas and elephants in the field that want to keep the newborns from becoming more gorillas and elephants fearing their own way may become that of the dinosaurs.

That is - the only utility is to the Microsofts of the tech industry in order to kill off competition that they fear may make them obsolete if allowed to remain in business. And, btw, the software developers at the Microsofts of the tech industry will also have zero utility of any patents as they too will not be allowed to read patents for the same reason I cannot.

Is that really a utility to society? Is that promoting the arts and sciences?

Please too keep in mind that much of what would account for prior art is hidden within the historical records companies (and companies that have long since gone out of business) due to the historical use of Copyright and Trade Secrets as sufficient legal protection. That is why the USPTO Prior Art searches are inadequate to find the requisite prior art to invalid patents as the patent system was designed. Even now, much of what would be prior art is still locked away and not entered into anything searchable by the USPTO for the same reasons; or (as is the case with FOSS) it is public but not where the USPTO is allowed to look or even able to find.

Posted by: Benjamen Meyer | Nov 29, 2012 3:20:47 PM

If you really want to bridge the divide you are going to have to find very good, well respected programmers to come out and say that they've filed such and such a patent, be willing to defend the specific patent, and say that it's helped them pay the bills and keep coding.

Even programmers from very large patent wielding companies (Apple, Google, Microsoft) I've met have been very reluctant to do so. That's not the case in, e.g., the hardware side of google where I've seen some proud bragging about power center patents.

As long as patents are being defended by: perceived patent trolls, those who have sold to said perceived patent trolls and left the industry, company PR reps or high level executives, or lawyers it just isn't going to be very convincing. That's both because of the (unfortunate) tendency of people to buy into the ad hominum fallacy, and (legitimately) because none of those people speak the right language.

Get a Ken Thompson, Jamie Zawinski, John Carmack, Eric Lippert, Jon Skeet, or the like to make the kind of appeals musicians have been making with respect to copyright infringement and you might crack the door open.

Of course you'd have to first convince them ...

Posted by: brad | Nov 29, 2012 2:17:51 PM

To add on to my already too long prior comment; the reason point #3 is so important to those of the "utilitarian anti-software patent" strain of thought, is that it would *appear* (to those of us who are software developers but not lawyers) to be an elegant way of limiting the (utilitarian) damages to society from allowing software patents without needing to change any laws or give software any special legal consideration. We'd just need to update legal reality to match with actual reality (e.g. software is just math). The "idealist anti-software-patent" folks (and the utilitarians with "more nuanced" utility functions) will additionally make the argument about infringement of liberty.

Posted by: nicholas a. evans | Nov 29, 2012 11:58:30 AM

First, I want to thank you for engaging with and attempting to communicate with the Groklaw (i.e. idealist anti-software-patent) community. I think you are on to something big with this "two worlds" communication barrier. I'm going to try using this framework to better understand your view and better communicate mine in the future.

Yes, I am (by this dichotomy) an idealist. When push comes to shove, I agree with Stallman that "Maybe they will be developed, maybe they won't, maybe they will take longer, but the harm to any future software programmer/company is never justified..." But I'd like to suggest that most of us idealists got to our views by way of utilitarianism, perhaps with a different utility function.

A few points (without much elaboration, lest this comment turn into an (even larger) essay):

1. We are skeptical that any software patents could ever fulfill *your* utility function (let alone others). Our experience and knowledge of CompSci history leads us to suspect that there is simply no such thing as a socially beneficial software patent.
2. We are *very* skeptical that any legal framework can ever be invented and sustained to weed out the weak software patents and leave only the "strong".

You may not share this skepticism, but these points will be debated on *utilitarian* grounds.

3. We learned that "Making A happen by doing X, Y, and Z" is not considered patentable when X, Y, and Z are all pure math... so we find it absurd that it should be any different when X and Y are either obvious or not novel (e.g. general purpose computing hardware), and Z is novel or non-obvious but pure math (e.g. software).

That point is creeping into idealist territory, but once you take the first two (utilitarian) points into account, suddenly it appears as it offers an elegant *utilitarian* approach to dealing with the problem. *EVEN IF* there exists some mythical software patent that would advance society with its existence, there is no way to sort only for it without also bringing along a far greater cost of searching through, litigating, and dealing with all of the others. It's far more utilitarian to doom that non-existent (or one in a trillion) software patent to never exist than to allow the other trillion to impede the promotion of science and useful arts.

4. For some of us, utility is not simply defined as "whatever promotes the most progress of science and useful arts".

We have other factors in there, all jockeying for priority; e.g. life, liberty, and the pursuit of a good BLT. Some of those factors might have slightly higher weight than "promotion of science and useful arts". I think of myself as a utilitarian with high weights on "personal liberty" and "ability to create stuff with my mind and get paid for it without fear of being sued" in my utility function. And for some people (the true idealists?), these are not so much "a higher weighting" as non-negotiable trump cards.

5. From our point of view and training, "Everything should be patentable," and "Patents are the preferred tool for promotion of science and useful arts", and "Promotion of science and useful arts trumps individual liberty" are not empirically founded, at least in the realm of software.

Perhaps these are the idealist stances in the debate. ;-)

Of course, nothing I've written above contradicts your assertion that "there is a divide in how we see the world", but perhaps it can establish some bridges across the divide? Anyway, thanks for engaging so thoughtfully and graciously in the debate.

Posted by: nicholas a. evans | Nov 29, 2012 11:33:17 AM

Nartreb - did you see or hear my presentation? Please don't assume anything about my positions. By no means am I saying we should do nothing. It's just that the something I think we should doesn't carve out one type of subject matter.

As for patents and invention, believe what you will. There are plenty of economic studies that show the link (and plenty that go the other way). I've also represented start-ups trying to get funding for software business models. My point- my only point- is that there is a divide in how we see the world, even if we might come to the same conclusion in the end. Your comments bear that out.

Posted by: Michael Risch | Nov 29, 2012 10:22:35 AM

As to Elroy's latest point, this is a very good illustration:
What color are your bits?

For Prof Risch, I'd like you to answer why you feel justified in using the word "pragmatist" in describing your own position. It seems to me that you take a legal fiction (patents encourage innovation) as a starting axiom. There's nothing pragmatic about that. Honestly, it's hard to believe that you are debating sincerely. You seem to accept that (to some degree you aren't clear about) software patents are a problem, but your position is basically that we should do nothing about it. Why? Is it because you think that currently, good software patents outweigh bad software patents? Because you think that the courts will sort out any problems without outside help? Because you really believe that litigation is the best way to solve this problem?

Posted by: nartreb | Nov 29, 2012 10:04:00 AM

Michael - I'm glad the patent is narrower, since it will allow other ways to achieve the same and it doesn't block a very usefull way of inputting text for FOSS projects.
My main point is still valid though it is still a series of calculations performed on input represented by a code/number to transform it in another code/number that is the hopefully intended output.

But as you said we will likely never agree on it. I'm thinking background has a lot to do with it. As a graduate in operations research and actuarial sciences I've been educated to see everything in models and algorithms. I've taken CS classes out of interest whenever I had the opportunity and it fit in perfectly with the thinking in models and algorithms of econometrics.

Professionally I've had to follow several legal classes and I've had a hard time with some of it, because it does not always follow the rigid logic that my previous studies did. There is a definite disconnect there.

Noone in any of the fields I studied except the legal field would accept something like legal fictions, when there is hard evidence to the contrary. Yet during my legal courses (none of them anything to do with patents by the way) I've seen many of these fictions used as thruths to come to a legal conclusion. I've questioned the teachers about them and they all were completely comfortable with always accepting something as true, knowing that this was not always (or even never)the case.

I guess thats why law is considered an art and not a science, but a lot of the programmers out there are science people that you will likely never get to accept such things.

Posted by: Elroy | Nov 29, 2012 9:06:21 AM

JonCB - I didn't say shape matching was novel. I said it further limited this patent, because there are other ways to implement swipe keyboards without using shape matching. Every patent has aspects that are not novel - in fact all of them might not be novel, but the combination is. I'm not saying that's true of this patent. My only point is that the patent is much narrower than Elroy was even positing.

Further, I don't know what point you think I'm pushing. I know you are never ever going to agree with my position. That is the ENTIRE point of this blog post. We are speaking different languages. I am living in a different world than you are. It may well be that the swype patent is obvious, and otherwise shouldn't exist. But the way we get there is very different.

Posted by: Michael Risch | Nov 29, 2012 7:54:32 AM

" I believe that pushing a point that your target audience is never going to agree with is counterproductive, and you don't get points for being technically right in this case, regardless of how wrong their points are."
A cynic would say that's what you get with a highly political supreme court whos members come from the same profession that benefits the most from maintaining the status quo.

Good thing I'm not a cynic ... oh wait.

Posted by: Elroy | Nov 29, 2012 7:06:21 AM

"It's not all swiping, but shape matching coupled with estimation of differences from shapes to guess at words that might have been meant if the finger didn't swipe properly."

While i can understand how someone not skilled in the art would think that "estimating differences" is potentially patent worthy, I can tell you from firsthand experience that in the last 10 years that would be considered an exercise.

This highlights one of my issues with software patents. Not only is a patent based on principles directly out of a university textbook likely to issue(Does anyone really contest that?), it will mangle the description so badly that you'd be better off going to the textbook to work it out(i.e. it will "teach" me nothing) and if i'm a small to medium business, this fact doesn't help me. The bogus patent will still bankrupt me.

Whats more, the pace of development is working for the people abusing the system and against the hardworking inventors. Technology is too broad and the PTO too overloaded to ensure good patents, and there is no real disincentive to litigate. Even if you win you're still going to pay out large dollars in fees.

As a pragmatist, however, i have agreed with your point in general for a while now. I believe taking the hard line "Stallman" approach is only going to have an affect if there's some doctrine we can use to induce the supreme's to rule clearly in its favour (which i doubt is plausible) and every lawyer up and down is going to resist it because it doesn't resonate with them... And they already hold the high ground. I believe that pushing a point that your target audience is never going to agree with is counterproductive, and you don't get points for being technically right in this case, regardless of how wrong their points are.

Posted by: JonCB | Nov 29, 2012 6:48:35 AM

"As for the x,y,z - I think most would agree with you that if x,y,z are math and math only, then there should be no patent. Usually that's not the case - usually there is more, and software is only a piece of the process."

With the software patents I've seen, whenever there is something not math its either the input or output device. This will basically get you to the idiotic european situation: Software >as such< is not patentable. The original idea being that a larger thing using software for part of its claims should be patentable and not excluded because it relies on a step in software. The powers that be in europe degraded that to basically mean if it comes on a disc/stick/computer it's no longer software as such. Luckily whatever the EPO does is not community wide and we should keep it that way.

But what do the patents generally do? The non software steps generally describe common input, output and storage devices. The input for example being gestures on a touch screen. A storage device with information on what to do when certain gestures are detected and some software step that uses the input and stored instruction to display the desired output.

My problem with this: I see no problem patenting a touchscreen that translates physical touches into data usable by a computer, provided it's novel and new. Innovation happens here a lot with the screens becoming thinner by new methods to integrate the different layers that were previously needed for example. But that is not what this is about in a software patent. The touchscreen is already there and is only claimed as an input method.

The storage device likewise may be patented, ways to store more and more data on optic or magnetic carriers require new techniques that are innovative. But that is not what the software patent is about it uses this storage device to store information. That information is often what to do when certain input comes from the touch screen. That means the stored information are mathematical algorithms, since computers know only how to do math. So what is stored on the storage medium should not be patentable.

Next you will find some method to select from storage the method appropriate for the input. This is then claimed as a software implemented claim. A computer executes this step. Computers can only do math, so this step can not be patented (or that should be the case since math is explicitly not patentable). After selection of the method the patent will go on to describe the effect of applying the method and how this will display on said touch screen, or vibrate the device, etc ... The display is basically the reverse of the touch part. The latter takes outside input and translates it to something computer readable, the display is there to make the outcome of the math the computer did visible to the user. Again, the screen technology is patentable, but this is never what a software patent claims.

What results as new/novel claims are the only the algorithms that are stored and the algorithms to select which stored algorithm to apply. Additionally there is the overall idea that you can bring a computer in a certain state and based on that show a certain state of the computer/data in a certain way to the user as a result.

So the only thing new in software patents that could possibly be the basis of the patent is either math or an idea, neither of which is patentable.

This is a lot different from claiming a patent where everything from start to finish is a novel thing, but one of the steps is implemented in software. You should be able to patent the whole thing. But not the software step. If someone finds a use for the software in a different machine he does not infringe the patent, since he is only using one step in the description of the patented invention that is by itself excluded for being math.

Every software patent I've seen so far is of this form. All the non software steps in the patent are always existing things, all the novelty is always math. Like x, y, z is claimed. x and z are not math, but also not new. y is new, but is pure math. But you neatly packaged your claims in x and z to pretend you are claiming more than just the math.

Posted by: Elroy | Nov 29, 2012 5:18:22 AM

Brad - I think it was Co-Op that mentioned sports moves, though I wrote about the same in Everything is Patentable. The interesting thing is that abstract processes have been patented since the beginning of the system. Though computers are new, the practice is not. That doesn't mean it's good or bad, just that the analysis shouldn't be based on newness. See my article America's First Patents for more discussion. http://papers.ssrn.com/abstract=2017275

Elroy - I didn't say that it was OK to patent a circuit that was the same as software. I only said it was odd to view them as different, but I've read many, many comments that imply these are different. As for the x,y,z - I think most would agree with you that if x,y,z are math and math only, then there should be no patent. Usually that's not the case - usually there is more, and software is only a piece of the process.

bugstomper - you are now seeing some of the definitional divide I mention in the update. The claim is not to the comparison only a very specific process of doing so - indeed more specific than you mention. It's not all swiping, but shape matching coupled with estimation of differences from shapes to guess at words that might have been meant if the finger didn't swipe properly. Still math, as I've noted before, but there is definitely more than math going on - there is touch input on a keyboard, display on a screen, touch input (again) on a keyboard to pick a matching word, display (again) of other words, and of course the finger itself. I am guessing Stallman would say that the phone is a general purpose computer and that his plan would ban enforcement of the swype patent.

Posted by: Michael Risch | Nov 28, 2012 9:41:58 PM

"As for the 'show me a beneficial, non-obvious software patent' comments, I had that debate at Groklaw - there's no room (and I don't have the time) to do it again here!"

I don't think I saw this point mentioned in the Groklaw comments: Your primary example of a possibly good software patent was Swype. I am firmly in the no software patents camp, yet that example does resonate with me. I thought about how one would apply Stallman's filter to that - Does it run on a general purpose computer? Then I realized what makes it seem like it could be a reasonable patent. Stallman's criteria would not allow patent protection to be used to prevent someone from writing software that recognizes finger swipes on a keyboard displayed on a touchscreen that is attached to a general purpose computer. But neither does Diamond v. Diehr preclude one from writing or even running software that calculates the Arrhenius equation for reaction time of curing rubber. Swype teaches a method of user input on a touch sensitive display device. Diamond v. Diehr's patent is infringed when someone uses it to make rubber. Swype should similarly only be infringed when someone uses the techniques to turn an ordinary touchscreen into a user input device that works with continuous finger swipes over displayed keys.

Which still leaves open the question of can anyone actually demonstrate the existence of a "good" software patent.

Posted by: bugstomper | Nov 28, 2012 9:16:51 PM

I'd like to comment on your folowing added remark: "One comment below essentially says, "Well, of course circuits are patentable and software isn't. Software is just abstract math." The problem is that most patents don't claim just the abstract math part. They claim "The steps of making A happen by doing X, Y, and Z." Once you view a patent that way, a circuit and software are equivalently infringing if they do X, Y, and Z - they are the same - a means to some other patented end."

The first part about the circuit doing the same thing as a program are not quite equivalent. A program by itself does nothing. It's just a set of instructions in some machine readable language. When you load a program on a computer it sets the circuits in a certain state. Assuming no self modifying code, you could do the same by making a circuit with that state hardcoded in the die.

To you the latter is obviously patentable and the former should therefore be as well. To me the former is clearly a mathematical algorithm that should not be patentable and it makes me seriously consider whether the latter should be patentable. Should an IC be patentable if the only 'novelty' is the way it is hardwired? Because of the equivalence with software we know that the only novelty is nothing else than a mathimatical operation the circuit will perform on whatever input you feed it. Thus while types of transistors used in making the IC are patentable. The state they are in should not be. It is irrelevant whether you program them after fabrication or put in the program at fabrication time. The only thing you gain by loading a program is flexibility. If you find a better algorithm you don;t need to fabricate new hardware. Functionaly both are performing math.

The math brings us to another one of your points. Patents claim "The steps of making A happen by doing X, Y, and Z." That is as close as you can get to the definition of a mathematical algorithm in plain english. You basically made the point that software is math here yourself. Many of the patents I read describe the algorithm at such a high level that there are many ways to accomplish it. Some only leave you less freedom. However they all build on other algorithms that are also math. When teh patents describe communicating over a network They refer to a defined activity which is also math. The patent is simplified in the same way a mathematician that applies pythagoras in a proof would not include the proof of pythagoras theorem but assumes the readers knowledge of that proof. Using such broad language hides the fact that software is math from a reader not familiar with the math. It does not make it 'non mathematical'

Posted by: Elroy | Nov 28, 2012 5:46:53 PM

@Bruce Boyden:

"Only math" is an alternate form of "only information", which is also un-patentable. To take my earlier analogy: why can't I merely patent court-briefs or, actually, any information at all? Why can't I patent this post and charge you license fees for reading it? Unless you really stretch the concept of process patents (this post caused a rearrangement of electrons and neurons in your brain, and I'm patenting the outcome of that particular rearrangement), it's indefensible. Life itself would grind to a halt because the commons (of freely exchangeable ideas) would vanish.

Software is a series of instructions and thus no more patentable than any instruction manual. If I can patent any instructive work, I can patent any work at all, because all works are instructive and transform their audience (for the purposes of patentability, just as a general purpose computer is transformed).

Posted by: Nick | Nov 28, 2012 5:07:45 PM

From the update
One comment below essentially says, "Well, of course circuits are patentable and software isn't. Software is just abstract math." The problem is that most patents don't claim just the abstract math part. They claim "The steps of making A happen by doing X, Y, and Z." Once you view a patent that way, a circuit and software are equivalently infringing if they do X, Y, and Z - they are the same - a means to some other patented end.

I'm not sure this really helps your case. Rather it makes me wonder if process patents in general need a second look.

I think in the mind of the public the classic patent is a tangible invention - bifocals, the windshield wiper, aspirin. Something like the Bessemer process does not immediately come to mind, but people can certainly be made to see the logic. But when you get into State Street, you are going to face a very skeptical audience.

The FOSS world is where you are currently getting pushback because people in that industry are coming into regular contact with 'modern' patent doctrines, but I imagine that you'd see the same thing across many sectors of society, based on the intuitions I outlined above, if patents started becoming a day to day matter.

I remember a blog post (here? concurring opinions?) which had a thought experiment on patenting sports moves. If that were to happen the FOSS people would seem like sane, rational, and easy to reach in comparison.

Posted by: brad | Nov 28, 2012 4:54:29 PM

"Software is only math." This argument, repeated several times above, proves too much. Everything in existence is simply information stored in various forms. Thus it is all "only math." But that doesn't categorically rule out patents on physical objects either. Processes, patentable since 1790, are algorithms by definition. There may be good arguments against patenting software, but the "only math" or "only algorithms" argument isn't one of them, at least not without further elaboration.

Posted by: Bruce Boyden | Nov 28, 2012 4:10:05 PM

Fair enough, nartreb - though note that I did say "friends of free software" was my definition. That said, plenty of pragmatists have posted comments above that tend to show the divided worlds, as I note in my update.

Posted by: Michael Risch | Nov 28, 2012 4:04:56 PM

The difference you saw between Stallman and Duffy is real, and should not have been surprising. It's nothing to do with Utilitarian vs FOSS. It's more to do with the difference between Free/Libre and Open Source.

Open Source folks tend to be quite pragmatic, Free Software folks can be downright dogmatic.
In most cases, there is little practical difference. In this case, both are opposed to software patents.
To assume that an Open Source supporter who opposes software patents is dogmatic is a major mistake.
Most of the folks I know who oppose software patents (and that's everybody I know who actually writes software for a living) do so on entirely pragmatic grounds: they simply don't encourage innovation. The argument in favor of patenting software is entirely theoretical and unsupported by any evidence, ignores the vast differences between software and other fields where patents *might* be useful (for example, that software is already covered by copyright, and that there is not regulatory hurdle such as FDA approval that might justify extra incentives for inventors)... I could go on.
True, the "on ethical grounds, software should never be patented or otherwise restricted" folks are vocal, and it's easy to focus on them. They're interesting because they seem alien to folks like you who have never even considered the idea that patents or copyrights might be harmful.
They might even be right - but I'm not really interested in ethics, and I think you aren't either. My point in this comment is simply that large numbers of opponents of software patents do so on entirely pragmatic grounds, and ignoring their existence is frankly stupid.
I know that when you get an avalanche of comments like this, the ones that stand out are the most strident, and anyone who doesn't explicitly distance himself seems to be joining the chorus, so nuances are hard to hear. But this isn't your first trip into this ring. By now you should know better than to focus on Stallman's peculiar sense of ethics to such an extent that you can't hear any other arguments.

Posted by: nartreb | Nov 28, 2012 4:02:49 PM

Software is only math. All software can be done with paper and pencil, it just might take a lifetime to do so. The general purpose computers only have hardware circuits that can preform math very fast. These are just the facts. No amount of Lawyer speak can change these simple facts. The US constitution prohibits the patenting of math. Why does this simple fact cause so much confusion among those who are just unwilling to understand the basic concepts of what a computer is and what it is not.

Posted by: Jeff Jacobs | Nov 28, 2012 1:56:44 PM

Am I correct in assuming that you, and most of your colleagues, believe that math is not patentable? If so, please examine the list of patented encryption algorithms at rsalabs.com and explain the rationale for granting patents for them. Encryption algorithms are pure math, and only math. Since the USPTO thinks this sort of thing is OK, it obvious that just about anything in software is patentable. It seems to me the problem is with the USPTO, and not patent law.

See www.uspto.gov for the USPTO current fee schedule:

Filing fees are somewhat reasonable(mostly < $1000), but a few others got my attention:

Request for ex parte reexamination - $17,750
Petition to institute an inter partes review - $27,200

It looks like bad patents make money, especially when litigation comes around (and it will).

According to the USPTO figures, they had "net position of $602.3 million" end of FY 2011. I'm not an accountant. Perhaps someone skilled in the art might look at the USPTO financials at www.uspto.gov.

Perhaps I'm too cynical (the USPTO director is an IP lawyer), but isn't this patent mess really just about money? Just for fun, what do you think Congress would do if the USPTO unilaterally stopped issuing s/w patents?

Posted by: albert | Nov 28, 2012 1:18:27 PM

All right, how about this way of bridging the two: The benefit of software patents is very slight, if not zero. Software got written anyway, lots of it, before software was patentable. Is more software being written now? Yes, because devices are cheaper, more capable, and more ubiquitous. More software is not being written due to software being patentable.

But software patents discourage the next piece of software from being written. They make it more difficult and dangerous to do so.

Therefore, from a utilitarian standpoint, there should be no software patents. They harm, and they do not help.

Posted by: Mike Stimpson | Nov 28, 2012 12:14:45 PM

Software as we know it is useless without hardware. One can run a program in ones head; indeed, I have seen people write programs in 68000 assembly code on a white-board and they worked properly on entry. The individual did not have a 68000 CPU in their head, but they understood it so well that they could run the program in its absence. They grokked.

There is little or no value of learning how to do such a feat outside of the profession of programming, i.e. a businessperson will not read an XLS file and create the spreadsheet, but it underscores the point that software is communication, math, and even the human brain can interpret it; albeit very slowly.

Patent hardware, but only if you have a working example. Patenting the automobile is useless if the machine, when constructed according to the patent, is non functional (George Seldon vs Henry Ford). If the task can be performed just as well (or better) by a general purpose computer then there is no sense in creating the hardware to do a specialized task, unless the specialized hardware has an advantage.

Copyright software. Software is communication. If a general purpose computer implements the same functionality and usefullness without relying on specialized hardware then the hardware that has been patented is obsolete, move along.

Posted by: BitDr | Nov 28, 2012 11:35:55 AM

I would have to agree with most of the posters here, this is a poorly written and highly skewed post.

The author's position is distorted to the point of distraction, "I've found that some take this view to the extreme - they have no problem with a circuit that performs the same thing as software, so long as it is performed in hardware." That I am aware, what is being referred to is part of the current test for patents. This is not an extreme view, this is the normal process.

Entirely left out of this article, pointed out by many other comments, is a central argument: is software math? Without addressing, or at least acknowledging this argument, the entire thing is irrelevant on it's face.

Posted by: Valentine S | Nov 28, 2012 11:21:28 AM

The following is not intended as trolling, but as a semi-serious joke expressing the ridiculousness of the situation:

Since software is nothing but series of instructions for operators (circuits) to perform, can I start patenting case-law?

Posted by: Nick | Nov 28, 2012 11:15:31 AM

Make me believer.

Name SINGLE patent used to disclose any useful info about any software innovation.

Than give reasons on why any good (as in able) programmer could not come up with same solution upon stumbling upon same problem (or different but with same solution).

Than prove that cost of independent discovery IS bigger than patent search.

Here I ask about ONLY reason for patents as I do understand US CONSTITUTION. To make knowledge public good.

Posted by: Przemysławw Lib | Nov 28, 2012 10:55:27 AM

As a software developer of nearly 30 years experience I just don't see the inventive leap that creates software patents, to me its an every day job to computerize whatever my client wants. The law on patents is therefore not being upheld as these are supposed to be understandable by an expert in the field and non obvious. Every software patent I've read bears little resemblance to a real world application and is written in lawyerspeak and or blindingly obvious to someone versed in the field. I therefore conclude that as were not being consulted on their validity and when we are being ignored there is another reason for allowing these through. Weather you think software patents are good or bad is irrelevant when the law as written is being manipulated by those not versed in the field and imposed on those of us who are

Posted by: Greg | Nov 28, 2012 10:38:38 AM

There is a lot of talk about good and bad patents. But there is no clear definition of what a good patent is. In a recent Canadian Supreme Court ruling, Pfizer's Viagra patent was invalidated because the patent did not disclose the active ingredient.

Doesn't it seem odd that patent holders want protection for their invention without full disclosure of the invention? This is what I see as so wrong with software patents. Software patents consist primarily of functional claiming that offers no insight into how the "invention" actually works. They claim an idea without claiming the method or technique for making it work.

Couple that to an evidence standard that is almost impossible to overcome in court and you have an unchecked agency in the executive branch, the USPTO.

I'm willing to grant software patents so long as the code for the "invention" is disclosed. That will limit the scope of the patent to one software language using one example of code. These conditions would effectively reduce the protection to copyright, but for 20 years, not for 70 years plus life, as Jerome Lemelson would like.

Posted by: Scott Dunn | Nov 28, 2012 10:00:44 AM

Per your quote... about how FOSS and other free software folks who don't believe software should not be patented? You say the that their only reasoning is based on the belief that software is "unpatentable as a matter of principle"...?

You have not been listening have you..., you can't because you believe something else. Where your belief is only a belief. Because you "want" it to be real, does not mean it is reality.

It's math. Digital is the "myth in the middle" that is being sold as re al when it is not. See:
Part of the transcript is as follows:
1st - Let me add this quote found at 1:08:00 point (here is an example of where he gets into the answer to an interesting question - related to patents for sure, as then we ask where is the real invention):
(not exact, but close to exact quote):
"I have been searching for what is special about computers...
I will announce this...


You have been misled."

(go to the timeline points highlighted from the video and the pay attention to the excerpt points in the above link and the video itself, for a very interesting insight of this wonderful talk).
All JUDGES and LAWYERS should view this video, before they even think any more about what is a computer is, or is not... as, it will present them with more questions that they must ask, where they will end up at one final resting place called MATH.
HERE below are some very interesting transcript excerpts that were posted in former related Groklaw comment)
The talk was interesting - would love to see a written transcript of it to refer

14:45 What about computers? Hmmm? How are we fooled?

"People think computers are conceptually digital, conceptually

(same section when talking about applications):

things that we do with computers:

"are Not rendered discrete, just because carried by a discrete

"Can encode things that are not discrete, on top of digital or discrete

"That fact, that in some sense that you can represent the non digital on
top of digital, is probably as important to the computer revolution and to the
future of our society as is the digital miracle itself"

Then, there is a slide that shows up as part of the segment (if you will) that
starts at around the 17:20 and/or 17:45 time point. There is a slide then that
is discussed and shows up that shows the 3 layers of the conceptual levels.
Where the middle, the digital layer is only a Myth.

The 3 levels:
Top (concrete) - not digital level
Intermediate (abstraction) - a digital level
Bottom (physical) - a not digital level

(hmmm, Remember that the digital level is A MYTH.

A Myth is a story.

18:10 The misunderstanding about digital (being a myth)?

At conceptual level "it's just not true that computational is digital"

"It's a major conceptual falsehood"

"It's a myth, that is expensive".

"Getting over this myth, that the computational realm, is in fact digital,
is going to cost us most of what we learned in school, a great deal of the
modern intellectual tradition, and my favorite project all of

(hmmm, could software patents be part of the expensive confusion over this
digital is real "myth" belief by some?)

24:00 - The Mind Body problem for machines paper that he can not publish
because it is so messy.

"The distinction between software and hardware, but anyone who has built a
system knows that the distinction between hardware and software is itself not a
conceptually a discrete system".

Representations of computers:
"a subject object divide"
"difference between a system and it's environment"

"If you actually have experience building computational systems, you
realize that in the case of what I call computation in the wild, the actual
systems that get built,
the distinctions are not sharp".

"when you build computational systems you discover that you are messing
around in complex territories"

"possibly on the edge of chaos"

"the abstractions"

"all of the notions are not discrete"

"messy and confused and disruptive"

Need to understand not what is digital, but what is not digital...!

--- from that last quotes, then moves on, he then goes into the subject of
"formality", and more, that you have to view to get the full picture.

(Hmmm is formality then what some folks are trying to think they can patent in

32:00 (before and after)

"world of blogs is not a digital future"

(hmmm, again talking about applications?)

32:30 THE REAL PROJECT segment

Leave the discrete behind...

"Digital notions are a mistake"

we need to "Escape from digital notions"

Main talk ends here!

Lots of good questions and responses (2/3rd of the talk are responses to the
wonderful questions).


"There is wide spread idea that there is something special about

"Computer does not mean digital"

"What is a machine?"

"I have been searching for what's special about computers since 1967...
looking for something special"

"There is nothing special about computers"

"You have been misled".

So, I have missed a lot of notes, a transcript would be handy to refer to...!

The myth of the digital middle! The myth of the word digital in the first place
is something that many believe in, when it is not what is real.

The math, that is real.

* Brian Cantwell Smith has a book "On the origin of objects"

And, So... dear believer, please understand, that with software patents, it is only a myth to believe in, only because you want to believe in it (that you can patent with a software patent as an invention). But, the reality, due to math, and logic, and yes, the total history of the modern computer back over a thousand years, where you use the socratic method... is, that it what you want to belive in, IT is proven to be a myth, to be believed by utilitarians who "WANT IT TO BE REAL" - but, have deluded themselves totally because of their "religious belief in that it must be real", when it clearly is a myth. For an example of proof that software is just math - and the logic you are not using in your arguement then please see this url:


Quote - "...experts have argued back and forth since then about the details of Cohen's Theorem. None so far have proven him wrong. His proof is so diabolically straightforward that contradicting it comes down to arguing that "True" equals "False."

Cohen's discussion rests on work done in the 1930's by Goedel and other philosophers in the field of logic. Goedel used his Incompleteness Theorem to demonstrate there will always be complexities which cannot be handled by mathematics, logic, or any other "language" defined by our limited imaginations. Alan Turing and other mathematicians quickly realized that those "languages" include computer languages".

If you want to know who Alan Turing is, then read this whole link:
...and see where he fits into the computer picture... AND, they you can more fully understand where your view of the digital myth (a false god, but real in your eyes), is taking you off track. Computers, the methods and concepts, have been around for a long time, and predates the transfer of binary from analog computers to digital (the invention that can be patented is the design of the the chips that allow the math to happen - as that is the only method or concept that is new... and, due to time is not really that new TODAY).

If, software is not just math, then the whole article is false.

Copyright is all that is needed to protect software, and really, it should be a class of software with reduced protection of about 7 years for machine instructions or SOFTWARE CODE.

Does Programming a Computer Make A New Machine? ~ By PolR [Article as PDF]

Please wake up.

Posted by: Anon - don't use email address | Nov 28, 2012 9:54:06 AM

A few comments here, and I plan to put an update in the main body of the post:
TJ - Of course I'm painting with too broad a brush - it is a blog post, after all, not a journal article. But I don't think I'm far off the mark. I do make clear that a) bridging is easier if you would bar software patents, and b) utilitarians can't agree on the right outcome. It's all still within the utilitarian framework.

As for the "show me a beneficial, non-obvious software patent" comments, I had that debate at Groklaw - there's no room (and I don't have the time) to do it again here!

More in the main body.

Posted by: Michael Risch | Nov 28, 2012 9:50:19 AM

You said "..I've found that some take this view to the extreme - they have no problem with a circuit that performs the same thing as software, so long as it is performed in hardware..."

But software doesn't perform any task. Software is instructions for a different set of circuits (a general-purpose computer). The circuits are the patentable thing, not the software.

When you say that you've had an epiphany on how to bridge the two worlds of thought, you absolutely have to understand how the people from each world understand the issue, and I don't think you do yet.

You wrote "...Surely, I thought, they might see that there are some lines that can be drawn that would allow for inventive software innovations. Surely, I thought, we can discuss some tweaks that would help alleviate the deleterious effects of low quality patents but save the system for one good software patent.

Surely, they thought, I would see how software patents are a bane to society, and must just go. Surely, they thought, I would see that there is no such thing as a good software patent..."

Your thinking is clearly explained here. The thinking of the other camp is not. It would be better written as:

"Surely, they thought, I would see how software patents are a barrier to innovation. Surely, they thought, I would see that patenting software is patenting ideas."

It's now about whether a given software patent is "good"...that's a meaningless statement from the other perspective. Software isn't patentable stuff any more than an equivalent set of equations written on stone tablets is.

Software is useful because of the [patented] computing hardware it runs on, not because "while x do y" is an innovation, however craftily strung together with other mathematical utterances.

If the problem is a disincentive to innovate, then you're really talking about a problem of theft. Why isn't copyright still sufficient to protect software? It's handled that problem admirably well for a long time. Why can't closed-source software continue to have a monetary incentive to innovate in a world with no software patents?

Posted by: Adrian Stovall | Nov 28, 2012 9:19:37 AM

Mr. Risch, thank you both for this post and your contribution to Groklaw, both have been very informative and helpful in understanding your views.

To name the second camp, I'd call them the makers: folks who don't need patents to create (particularly those with a lifespan of beyond 5 years, an eon in the technical world). That worldview can be much in line with the recent Republican-Copyleft report: http://www.volokh.com/2012/11/19/republicans-going-copyleft/

Patents are not required to create inventions generally: people make and innovate, it's what we do. Patents are required to encourage people, who demand private ownership of an invention's expression, to create. The former camp views the patenting of software as a loss of math itself, removing particular tools from everyone's development toolboxes for 20 years, hampering the overall pace of innovation. Uncontrolled competition means that everyone's always competing, requiring constant innovation from all parties to just keep up with the disruptors, preventing anyone from resting on their laurels (and surviving, anyway).

But, really, laws will always reflect the world we wish to create. Folks like Stallman, and me, will continue to mourn the existence of software patents: since software patents cover mathematical concepts, re-engineering is impossible: a re-implementation of the process that produces the same results can't help but infringe the patent (unlike with physical objects, where minor changes to the physical process can produce equivalent results through a non-infringing method). Only copyright protections, and copyright alone, can provide the equivalent counter-engineering permissions that exist with physical patents.

Posted by: Nick | Nov 28, 2012 9:19:14 AM

You SAY that patents are worthwhile because they drive "innovation" but where is your EVIDENCE and what is the net benefit in dollars, again, based on EVIDENCE. The costs of patents can be quantified, but you expect us to accept amorphous claims of "value" based on a religious belief that patents increase "innovation" with no basis in fact and no dollar value that can be attached to it.

Posted by: Renee Marie Jones | Nov 28, 2012 8:46:33 AM

Investors are today investing billions of dollars into work only protected by copyright. Films. music books etc are all financed by others and are seen as sufficient protection for the investments.

Why cant investors in software companies invest purely based on copyright ?

There is really no need for patents on software as long as there is copyright for software. I think this is another issu - investers are used to normal tech companies where they protect their products with patents that they cant grasp a world where products are protected by copyright and demand patents for software too.

Tech investors need to be educated on how to make money based on copyrighted products.

Posted by: Anders S Lindbäck | Nov 28, 2012 8:44:20 AM

Maybe i was not clear enough....
1) What i'm trying to say is that before Copernicus the people BELIEVED in geocentric model of universe without any PROOF of correctness. It was question of BELIEF. Copernicus managed to PROVE that heliocentric model is correct. It is the same with patents (btw not only software patents). Some people (including you) BELEIVE that patents foster innovation. There are economic studies showing that this is not the case.

2) You believe that forced open sourcing will not work for everyone but it is simple belief not supported by the facts. The facts (for example rate of innovation in open source) prove the contrary.
3) Btw the same reasoning apply to copyrights it is pretty evident, i think, that if we abolish copyright the and the artists will be free to reuse and enhance output of other artists without restrictions the volume and quality of artistic production will soar - as again open source software model proves.

4) BTW to rationalize our discussion i have a question for you:
Can you imagine an argument which you accept as proof that you point of view is incorrect?
For example i think i'd change my POV if i'll see couple economic studies showing that
stricter IP enforcement caused more innovation and product price reduction

Posted by: Vadim Lebedev | Nov 28, 2012 8:44:07 AM

One place I worked wrote LOTS of software.

The same algorithms used in every GPS navigator was used back in the early to mid 1970s. Everything from LORAN-C (on PDP-8 computers no less), to GPS recievers.

We depended on copyright, and trade secrets - though there weren't that many secrets. Navigation systems have used the same algorithms for hundreds of years.

Yet, this same field is now covered by hundreds of patents... nearly all of them "on a computer" or "on a handheld".

It is nothing but mathematics. That is all you have - a computer is nothing but an engineering approximation (patentable no less) that when operated within specific limits (voltage, temperature, clock speed) performs only mathematical transforms.

And many of those transforms date back thousands of years, others go back to Newton/Leibniz.

All we have are mathematical models of structures - and transforms of those models performed by a patented machine, that reads, manipulates, and writes the results.

Reading and writing are already covered by copyright and/or trade secrets.

Posted by: jesse | Nov 28, 2012 7:15:52 AM

To get this discussion to a practical rather than theoretical level, it would be useful if someone could point to an example of a "good" software patent ("good" meaning that the invention was novel, useful, non-obvious, understandable by most software developers and has been widely licensed and used after disclosure).

Posted by: Charles Williams | Nov 28, 2012 6:21:05 AM

> "Utilitarians like Duffy (and me) believe that it is worth driving the ex ante incentive to innovate"

Could you please give one or more examples of innovations that would not have been made if the patent system was not available? You seem to assume that such exists, and I think this is where the "free software folks" disagree.

Posted by: Jesper | Nov 28, 2012 5:55:47 AM

Forced open sourcing will not work for everything. Companies need to recover the cost of programming their product. I agree they don't need patents, but copyrights work fine.

The competition can use your ideas, but has to make the investment in actually implementing it.

For me this is also the biggest issue with software patents. Computers are basically combinations of hardware that can do a limited number of things.

Take a Turing complete computer with the following ways to interact: Keyboard and monochrome screen. (to keep the example managable)

What can this system do? It can take input from the keyboard by reading any combination of keys the keyboard can provide. It can provide output by setting each individual pixel to black or white.
Anything you can think of that takes only keyboard presses as input, and produces output as a matrix of pixels, this computer can do. (Practically only limited by time and memory constraints.)

If you can think of it, all you have to do is write it down in a language the computer understands. The language the computer understands is ... math! Programming is nothing other than the art (copyrights) of finding mathematical (not patentable) formulas that transform your input into the desired output.

The reasons people do new things with computers or now do things on handheld computers that they didn't do before is not because of inventions in software. It's because of innovations in input and output technology and increasing speed and memory sizes (re)moving barriers. People may have thought about it years ago, but dismissed it them because if wasn't workable on then current hardware. Or people may not even have considered it because they were aware of the limits. 'Innovation' in software is driven by real innovation in hardware. It's ideas that have become feasible by progress in other fields.

This even goes for something like MP3, this only became feasible when processors got fast enough and was driven by the constraint 'storage' space. The result was a mathematical algorithm that traded storage space for processing time.

(MP3 uses a psycho acoustic model to decide what sounds it can drop. This model is something that could be copyrighted separately from the source code implementing the algorithm.)

Posted by: Elroy | Nov 28, 2012 5:55:36 AM

Apologies for commenting anonymously, but there's evidence that publicity of all kinds attracts patent trolls to startups, which I don't need right now.

I've followed this issue for many years, and I must have seen a hundred software patents, but I have never seen a good one. By which I mean one where the core idea is an important invention, not bleeding obvious, and where the patent says how to put the idea into practice. Actually, just the "not bleeding obvious" test gets rid of all software patents that I can think of.

People who are pro-patent seem to have seen lots of good patents. Or maybe they just assume patents are good, or they are astroturfing for trolls. To settle this, please could someone post half a dozen examples of good software patents, published in the last year. Thanks.

Posted by: Anonymous | Nov 28, 2012 5:30:27 AM

The premise of utilitarians is that patents in general and software patents in particular are promoting innovation. But you must recognize that there is no proof of such fact.
It is simply a belief. There is NO scientific study that shows that patents are actually promoting innovation, however there are studies showing that are have no positive effects on innovation or even hindering innovation.
BTW try to do a little thought experiment and imagine a system where all vendors are obliged to 'Open Source' all their product on the market let's say a 6 month after introduction. It is obvious that innovation rate will become exponential in such a setup... Because competitors will have a possibility to incrementally enhance the products and the original vendor will have possibility to incremenatlly enhance the competitor products... The best example of such sytem working is Open source software. The society will clearly benefit for such a system

Posted by: Vadim Lebedev | Nov 28, 2012 4:46:42 AM

I think you are painting "utilitarians" with far too broad a brush, though you make nods towards Pam Samuelson. I agree with you that there is a strain of moralistic absolutism in the anti-software-patent movement, but there are plenty of utilitarian software-patent-skeptics (e.g. Bessen & Meurer). There is a chasm between you and them--and, to a lesser extent, me--is not because of a utilitarianism-versus-morals divide but because of something else: fundamentally, you are very optimistic that we can tweak substantive patent doctrine to filter patents effectively, and you therefore equate a "strong" patent under existing (or slightly-tweaked) doctrine with a socially-beneficial patent under utilitarian philosophy.

We've discussed this before, but I am far more skeptical this is the case. If one believes that substantive patent doctrine cannot filter--or that it is administrative too costly to filter--the 1 socially beneficial software patent needle from the 100,000 software patent haystack, then I think one ends up largely where Stallman is at without any non-utilitarian moralistic overtones at all. That is there may be strong software patents (under doctrine), or weak patents (under doctrine), but they are all bad software patents (under utilitarianism) because the doctrine is fundamentally broken. Stated another way, if (as you seem to concede) the number of good software patents is vanishingly small, I wonder how you justify the administrative expense of finding that needle in the haystack.

Posted by: TJ | Nov 27, 2012 11:31:56 PM

The comments to this entry are closed.