I do think that the distinction between software or an algorithm implemented on a computer is quite different from the software or algorithm existing in someone’s head, or even printed on a T-shirt (remember PGP?)

Then you have a lot of balls calling others technologically illiterate. Algorithms are mathematical objects.

Since software is nothing but source code and perhaps some graphical/sound elements (icons, sounds, images, and whatnot) it could be considered that the source code is speech. Because it is speech it is protected under the 1st amendment. Thus; any enforcement of a software patent violation would be a violation of the 1st amendment.

This isn’t the same kind of limits on speech that copyright entails since two implementations of the same patented process could use completely different source code and even written in completely different languages for completely different platforms/devices. So essentially what this means is that by patenting a specific aspect of software you’re effectively barring anyone from communicating it without your permission or licensing.

-Riskable
“The Supreme Court ruled that political contributions are to be considered free speech. Apparently the more money you have the more free speech you get!”

JUSTICE STEVENS: I don’t understand how that can be a patent on a machine if the only thing novel is the process that the machine is using. Isn’t — isn’t the question — really, the question there was whether the new process was patentable.
(p. 42)

JUSTICE KENNEDY: That’s — that’s a problem I have.
(p. 44)

JUSTICE BREYER: But then all we do is every example that I just gave, that I thought were examples that certainly would not be patented, you simply patent them. All you do is just have a set of instructions for saying how to set a computer to do it. Anyone can do that. Now, it’s a machine. So all the business patents are all right back in…all you do is you get somebody who knows computers, and you turn every business patent into a setting of switches on the machine because there are no businesses that don’t use those machines.
(p. 46)

This argument is the same one I made in this paper that would invalidate business method and software-only patents regardless of whether “computer” or “CPU” or “memory” or similar was added on as a Christmas tree ornament:

“Elemental Subject Matter”
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=916963

“A long line of United States Supreme Court cases have followed the methodology of Neilson v. Harford in requiring novelty, and also often non-obviousness, to come entirely from the parts of the claim that are statutory under what is now Section 101 of the 1952 Patent Act. This paper traces the development of subject matter law in the area of software patents, where loss of the Neilson test led to contradictory holdings and de facto legitimization of supposedly nonstatutory subject matter. It proposes to restore stability and substance to subject matter law with a return to the Neilson doctrine, in a modern form consistent with the 1952 Patent Act. ”

I suspect Bilski can be decided without deciding this Christmas-ornament issue, but if or when the philosophy of at least these three justices is implemented, even your cleverly drafted software patents will be toast.

If you want software to be patentable, you should get Congress to pass legislation to make it patentable. What the Federal Circuit did in allowing software patents that use a machine element only as a fig-leaf, as you have cleverly done, grossly violated a long line of Supreme Court precedent in the way that Stevens, Kennedy, and Breyer were complaining about here.

Hence a (novel, non-obvious) algorithm which changes the duration of a digital audio signal of the human voice, without changing the pitch (aka “constant pitch playback speed modification), becomes clearly patentable when the claim includes “wherein the so-modified digital signal is converted to acoustic waves by a transducer for playback to a listener …” (or something like that).

So, I am advocating in favor of algorithms that result in GUI impacts (provided the GUI is part of the purpose, which it usually is), but not the “material transformation” involved in the hardware circuits that execute the code, since that’s just implementation, not purpose.

IANAL, but I have written some software patents which I’m proud of (they really are novel), so I’d be interested to hear what lawyers have to say about this.

More importantly, software developers shouldn’t be required to perform a patent search before writing a program.

Also, one of the greatest aspects of software is that there can be many applications available that perform the same task. This means that the consumer can choose from a multitude of options and it also means that any software developer can make a better version of an existing application. This is the reason why the computer software industry (and the Internet in general) are the fastest-innovating industry of all time.

-Riskable

So, should all those patents on their CISC architecture processors now be invalid because they are implemented in software instead of hardware?

95% of the time when a piece of hardware is being “emulated” the whole chip isn’t actually being emulated. The front end instruction set (the buttons and knobs on the chip) is and then on the back end the machine code is dynamically re-compiled for host system. Most patents on IC’s are all on the back end, and aren’t even implemented when virtualized or emulated. Did Parallels obtain an x86 license from Intel when it was making its PC emulation software on old ppc macs?

At least claims like RSA mirror traditional patents in that respect.

I do have problems however in distinguishing between RSA and Newton’s method. How are we to draw that line for algorithm purposes?

Or perhaps the claim should have attempted to be on asymmetric encryption itself, with RSA as just one possible implementation? (Ignoring the potential that such encryption was already known. I’ve seen claims that Diffie-Hellman was prior art.) Such a claim would mirror previous patents even more.

I agree. Interoperability ought to be an absolute safe-haven from software patent claims.

Of course, that more or less defeats the purpose of software patents in the general market (though not, perhaps, in certain highly technical markets).

I got some other interesting answers, but haven’t heard from you. You seem to have experience filing patents (?) — do you have an example of a good software patent? Perhaps something you’ve filed or run into over the years?

With regards to some of the other responses (e.g. RSA and Reed-Solomon encoding), I do agree that there needs to be some kind of protection–I think these are good examples. However, what about CDU’s example of the solitaire/cryptonomicon encryption algorithm? Something can be patented when executed on computer but anybody can do it by hand? Or with a calculator? I don’t know how to draw a line here.

Readery: In repsonse to “Microsoft Windows … ought to be patentable.” What does this do for interoperably? Would open source projects such as Samba (http://www.samba.org/) be forced into some sort of patent licensing scheme merely to be able to “talk” to Windows networks? Talk about destroying innovation — the established companies could just patent their products making it completely impossible for competitors to do anything.

Therefore, a time-critical solution is more likely to be implemented in hardware, while a solution that requires the ability to add features easily will be implemented in software. As a result, to be intellectually consistent, those people against software patents also have to be against patents for electronic circuits.

Actually, it is a bit more complex than this. When you are dealing with some electronics, and in particular, working on the boundary between hardware and software, the line of what is implemented in what type of hardware and software, when, often changes over time.

So, let us assume a black box that provides novel and non-obvious (for our discussions here) functionality. The functionality may, and often does, start out as, possibly high level, code executed by a computer processor. Then, the code may be implemented on an embedded processor. At some point, it may be modeled as a software expression of circuitry (e.g. Verilog), which may in turn be executed almost as if it were a software program. Maybe then, it moves into an FPGA where much of the programming involves routing, with one and zero bits turning on and off paths through the circuitry. Then, maybe a custom IC is fabbed from that Verilog model. The decision of how and when to implement the invention is an engineering decision based on stability and volume.

So, does it really make sense to have a system where patentability depends on how an invention is implemented, and where that patentability may change dramatically over the life of a product? Or where someone can bypass infringement by implementing something in software, where the patent was limited to hardware?

And lest you think that the dynamic only goes one way, several of the old mainframe companies discovered that they could no longer afford to develop custom ICs for their CISC processors. What to do? The answer was to emulate these legacy processors on IA64 processors. And, for the most part, their customers cannot distinguish between when they had processors on custom ICs in the past, or when that processor is emulated in software. All they care about is that their code runs, either way. So, should all those patents on their CISC architecture processors now be invalid because they are implemented in software instead of hardware? I don’t think so, but, then again, I wrote some of those patents (as well as some of the emulation patents).

pc: And a really non-time critical feature would be implemented by hand. So we go back to algorithms. Knowing how to calculate the eigenvector centrality of something is not novel. Applying that calculation to x is?

But that isn’t what most software patents cover. Most of them are much more routine, and really don’t involves any math about maybe a 6th grade level.

Therefore, a time-critical solution is more likely to be implemented in hardware, while a solution that requires the ability to add features easily will be implemented in software. As a result, to be intellectually consistent, those people against software patents also have to be against patents for electronic circuits.

And a really non-time critical feature would be implemented by hand. So we go back to algorithms. Knowing how to calculate the eigenvector centrality of something is not novel. Applying that calculation to x is?

I think Microsoft Windows — a specific operating system for a specific kind of computer — ought to be patentable (although not necessarily through thousands of individual tiny-claim patents).

I would tend to agree that Congress, not the Supreme Court, ought to resolve this mess by determining what sort of intellectual property rights ought to apply to software and developing a set of rules and standards to provide guidance to the courts.

Now while there are probably some decent gems in the software patent universe that do “promote the progress of the art” they are far outnumbered by junk and I have yet to see a credible method that the patent office can use to reliably separate the wheat from the chaff.

What about something akin to the “essential nexus” test from land use regulation law? “Essence of function”, perhaps, to test if the link to a machine/process is essential to the totality of the claims, or if it’s just an “on the Internet!” tacked on…

Of course, software is not a machine. The question is whether it being loaded into a computer converts that computer to a specific purpose machine.

Interesting question. Would that depend on how close to the hardware the software gets? I’m thinking of the lawsuit by Paltalk and their recent lawsuit against some of the biggest names in the MMO industry (they settled with Microsoft in 2006). I’m not familiar with the law, but shouldn’t IP monopolies “promote the Progress of Science and useful Arts?”

If the patent is broad enough to cover MMO servers and clients, how would installing the World of Warcraft client on my PC make it a specific purpose machine?

Soronel Haetir: Sorry, gotta agree with CDU, software as such is not a machine.

Of course, software is not a machine. The question is whether it being loaded into a computer converts that computer to a specific purpose machine.

CDU: Software is just a set of instructions, it isn’t tied to any particular machine or physical object. It could be compiled for a different processor or operating system, translated into a different programming language, executed by hand, transformed into lambda calculus, etc.

Which is why it is pretty much unpatentable in that format.

Sorry, gotta agree with CDU, software as such is not a machine.

Bruce Hayden: A computer system containing the software is a machine (a programmed general purpose computer is a special purpose computer). And, a floppy disk (or CD/DVD) containing the software in a form executable by a computer processor is an article of manufacture.

Software is just a set of instructions, it isn’t tied to any particular machine or physical object. It could be compiled for a different processor or operating system, translated into a different programming language, executed by hand, transformed into lambda calculus, etc.

CDU: All software is simply a set of instructions. It obviously isn’t a state of matter, a machine, or a method of manufacture. The only possible place it could fit in under the statute is as process, and I think that is quite debatable and open to interpretation.

Properly done, software claims fall under three of the four statutory classes: process, machine, and article of manufacture. The method implemented in the software is a process. A computer system containing the software is a machine (a programmed general purpose computer is a special purpose computer). And, a floppy disk (or CD/DVD) containing the software in a form executable by a computer processor is an article of manufacture.

PatHMV: Let me ask you this. Do you deny that there’s any problem at all in this area? I realize you’re defending a position undergoing fairly stiff attack in this forum, and kudos to you for the valiant effort. But I think there’s pretty wide-spread agreement out there that stuff like “one-click shopping” is a serious abuse of the patent system and a hindrance to innovation that benefits the consumer. At the same time, your point about the inventiveness of the software hidden in the bowels of our cell phones and the like may have some validity. What I don’t see is any effort to eliminate at least some of the abuses, while preserving the parts that have some merit. Can you suggest a solution?

While patent trolling may be easier in software patents, it is not limited to that realm. Indeed, arguably the biggest abuser of the patent system was Jerome Lemelson (though there is no indication that he ever did anything anywhere close to illegal). He made close to a half a billion dollars on patent applications claiming technology invented 30+ years after his priority dates. And part of how he did this was to sue, and then settle before trial at a fraction of what he had been asking for. The invention that he leveraged off of? A very crude, electromechanical machine vision invention. His ultimate claims read on readers for bar code and other machine marking (such as are used for sorting good and bad ICs).

Cleanville Tziabatz: Have you read section 101 by any chance?

I have, and what it actually says is:

Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.

All software is simply a set of instructions. It obviously isn’t a state of matter, a machine, or a method of manufacture. The only possible place it could fit in under the statute is as process, and I think that is quite debatable and open to interpretation.

Einhverfr: Bruce Hayden:I disagree.I think software patents need to be curtailed though not eliminated.In general what often passes for a software patent is a patent on a vague method which usually lacks the sort of precision and boundaries I tend to see when reading plant patents, physical device patents etc.

Part of it is probably whether you are reasonably skilled in the relevant art or not. I think that I am, and therefore probably can understand what it would take to make disclosed inventions work better than someone who is not.

That said, there are some software patents (and a lot more patent applications) that lack sufficient disclosure. But, that is not a 101 problem, but rather, is a 35 USC 112 (esp. Paragraphs 1 and 2) problem.

CDU: In other words, it requires interpretation in order to stretch the statute to cover software. You seem to be objecting to the prospect of the courts adopting an interpretation other than the one you prefer. That’s a pretty loose definition of judicial activism.

Umm, no. The statute defines four broad categories of subject matter which are patent eligible subject matter. Like I said above, software invariably falls within two of those categories, and, in its commercial aspects, will invariably fall into a third. It does not fall into the fourth (at least until they can encode software into genetic type material). To be patent eligible subject matter, of course, software needs only fall into one category. However, it falls into three and none of the categories need stretching to accomplish this result.

Have you read section 101 by any chance?

Cleanville Tziabatz: Um, no. 35 USC section 101 is what is supposed to determine what is patent-eligible subject matter. This section does not make software patentable as such, but does make it patentable insofar as: (i) it is manufactured (which it generally is); (ii) it runs on a machine (which it invariably does); and/or (iii) it executes processes (which it invariably does). This is because that is what the statute says, and not because it is what some judge said (although the co-drafter of the statutory language went on to become a federal judge).

My boss clerked for that judge, and one of his favorite stories of his clerkship is some litigant trying to explain the meaning of the patent statutes. The judge retorted that he thought that he did know the meaning of the patent statutes. It was only later that the litigant figured out how foolish he had looked.

Kuzbad: Can we see an example of what software patents OUGHT to be like?

Google just got itself into a suit with a small mobile phone software developer over a method that dramatically reduces the size of binary patches to a piece of software. Google uses this method in a program they wrote called courgette to reduce the size of the patches it sends out for Chrome, but Red Bend (the mobile phone software company) has a patent on this method and has filed suit against Google. At LWN.net (a Linux/Open Source site, generally opposed to software patents) there is an article (temporary link that LWN allows subscribers to on occasion post, please don’t abuse) that explains the situation and several commenters to the article indicated that Red Bend’s patent might fall into the category of a “Good” one.

Now while there are probably some decent gems in the software patent universe that do “promote the progress of the art” they are far outnumbered by junk and I have yet to see a credible method that the patent office can use to reliably separate the wheat from the chaff.

CDU: In other words, it requires interpretation in order to stretch the statute to cover software. You seem to be objecting to the prospect of the courts adopting an interpretation other than the one you prefer. That’s a pretty loose definition of judicial activism.

Actually, I would suggest that it takes interpretation to shrink it so that it does not cover software.

Soronel Haetir: I would argue that if any pure algorithm should qualify then RSA would be the perfect exemplar.

As long as we’re in the realm of encryption algorithms, would you say that Bruce Schneier’s Solitaire cipher is patentable? It seems quite a stretch to say that a set of instructions for manipulating a deck of cards is patentable. Yet it’s rather difficult to draw a sensible line between Solitaire and RSA.

CDU: Given that it was judge made law that make software patentable in the first place, how exactly is it “judicial activism” to overturn those decisions?

And judge made law that made it unpatentable before that. The problem is that the patent statute (notably 35 USC 101) is very broad when it comes to statutory subject matter. As some court pointed out, anything under the sun is patentable if it falls into one of the four statutory classes and doesn’t run afoul of other parts of the Patent Act (35 USC). But, then judges have carved out some exceptions over the last 200 years, notably the laws of nature exception discussed above. Plus, maybe, a written matter exception. In the early 1970s, software was piggy backed onto the laws of nature judge made exception by analogizing that software involved algorithms, algorithms were mathematical equations, and they, in turn were akin to laws of nature. From the point of view of someone who has been programming now for almost 40 years, that analogy seems pretty bogus for most of the code I have written and most of the software patents I have written and/or prosecuted over the last almost 20 years.

I will agree with the above poster who pointed out that the RSA algorithm is the sort of algorithm or equation that is close to being factual, fairly close to the Newton approximation method mentioned above. But that is rare in software. As noted above, the only reason that we consider Windows a law of nature is Microsoft’s monopoly position in the market.

Cleanville Tziabatz: Um, no.35 USC section 101 is what is supposed to determine what is patent-eligible subject matter.This section does not make software patentable as such

In other words, it requires interpretation in order to stretch the statute to cover software. You seem to be objecting to the prospect of the courts adopting an interpretation other than the one you prefer. That’s a pretty loose definition of judicial activism.

Soronel Haetir:
I would argue that if any pure algorithm should qualify then RSA would be the perfect exemplar.

Kuzbad, Soronel Haetir:

I’d add Reed-Solomon Coding to that list. Very close to pure math, but incredibly important for things like Satellite communications and CDs. Granted, when first developed the general view was that it was cool but no one would *ever* have enough processing power to make it useful. It was really ahead of its time. I don’t know if a patent was ever issued (or applied for).

Kuzbad: Bruce Hayden / any others,Full disclosure–I’m a computer guy who is walking into this conversation very skeptical of computer patents. Here’s my question–can you (or anybody else) provide an example of a software patent which clearly OUGHT to be patentable? That is to say, it’s incredibly easy to find myriad examples of reprehensible software patents and to point out examples of companies trying to patent everything under the sun in the name of mutually assured patent destruction. I don’t think there is any disagreement that there are many bad patents, but what about the good patents? Can we see an example of what software patents OUGHT to be like?

I would argue that if any pure algorithm should qualify then RSA would be the perfect exemplar.

Full disclosure–I’m a computer guy who is walking into this conversation very skeptical of computer patents. Here’s my question–can you (or anybody else) provide an example of a software patent which clearly OUGHT to be patentable?

That is to say, it’s incredibly easy to find myriad examples of reprehensible software patents and to point out examples of companies trying to patent everything under the sun in the name of mutually assured patent destruction. I don’t think there is any disagreement that there are many bad patents, but what about the good patents? Can we see an example of what software patents OUGHT to be like?