David Kappos and Aaron Cooper recently co-authored a paper arguing that patents on software-implemented inventions are accomplishing the constitutional imperative to promote progress. Indeed, the paper makes the case that these so-called software patents have led to tremendous growth in the software industry, have fostered open innovation and interoperability, and are incentivizing solutions to unique problems.
I caught up with Kappos for an on the record interview on Monday, September 14, 2015. What follows is the second and final segment of our conversation. To start reading from the beginning please see The Case for Software Patentability, An Interview with David Kappos. In part 2 below we discuss how bizarre it is that an algorithm can be patented when it is implemented in a hardware, but patent ineligible when the same algorithm is implemented in software.
QUINN: Yes, probably true, but let me back up for a second. As you were just giving that answer there was one thing that you said that we really need to further discuss, because I can just hear the critics with their condescending laugh already. You said that back in the day of Jefferson and Madison there were patents granted that you and I would probably call software patents. I can just hear the critics, “Doesn’t he know that computers didn’t exist.” And what it really signifies is a problem in understanding what it is that you’re claiming.
Martin Goetz is the inventor who the New York Times and others have credited with receiving the first software patent issued back in the 60s. He filed his patent application in ’65 and it was issued in about 1968. Martin has done some writing for us over the years and he is recently saying that we really need to stop calling these things software patents. Just stop. It’s not a software patent, it’s an invention. It’s an innovation. And so when you said that they issued these patents that you look back on throughout time and they would be software patents it’s because of the way that you describe the innovation. And we use those same techniques to describe the innovation today because it is in many regards abstract.
KAPPOS: Right, yes.
QUINN: And how do you describe that? You know, you have to do the best that you can. And I really wish it was the best that you can was the law through the lens of the person who’s skilled in that area, which up until the last few years had always been the whole point of patent law. And I wonder whether that still is true, I’m specifically thinking about the Algorithm cases here. It seems we are caught between a rock and a hard place because you can’t talk about this issue without using the word “software” or “software patents.” But Martin has an excellent point. It does no good to talk about this as software. It’s you’ve got a windmill or wind turbine that has been improved. Who cares how it has been improved. It has been vastly improved.
KAPPOS: We tried to hit on this point in the paper and towards the beginning made the point that look, it really is a bit of a misnomer to be talking about software patents. What we’re really talking about is algorithms that actually may or may not be implemented in software. It’s actually common, and you’re seeing it now in the smartphone area. Back when I was an engineer we saw it in mainframe computers where you’d make an invention and frequently initially the software wasn’t fast enough to be able to run the algorithm. So the algorithm would first be built in silicon, really expensive, but you’d wind up then fabbing up chips to be special purpose chips to run the algorithm. And then later as the software got faster the underlying computer systems got faster you’d reimplement the same algorithm in software, same algorithm, same invention but just reimplement it in software and then even later after that when the ASIC density got good enough you’d reimplement yet again in an application-specific integrated circuit, an ASIC. And so you’d have a little bit of a hybrid, if you will, but more on the hardware side, it’s an IC. It’s again putting the algorithm in a chip. And so what you’d see by looking at that is that it made no sense to say that an algorithm was patentable if it was implemented in a hardware chip. But the same algorithm implemented in software was unpatentable. Just didn’t make sense to say that.
Looking at the nature of the innovation that way brings home the fact that the invention isn’t software. The implementation of the invention is in software maybe, but the invention is the algorithm. Another way that I sometimes point this out, Gene, is that when we make an invention we frequently will file a patent application in various countries. Maybe we’ll file one in France, maybe we’ll file one in Germany, maybe we’ll file one in Japan, the U.S., China, et cetera. Nobody would seriously say that we’ll accept a patent application in some of those languages but not in others of those languages. Nobody would seriously say that. Well, when you talk about submitting a patent application that involves an invention that’s implemented in software you really are talking about the same thing because software is just a medium of expression like a language. French, German, Spanish, and C#, they’re all languages. It’s just another language. So it has always seemed to me that it made no sense to say we’re going to discriminate against some medium of expression, software being one medium of expression just like any other language is a medium of expression. We’re going to permit patenting for some medium of expression but we’re not going to permit it for others. It’s like saying we’ll take that patent application if it’s in French but not if it’s in German. What sense does that make, right? It makes no sense at all.
QUINN: I completely agree with you. And hopefully this will eventually get through to the people who are the decision makers and to the judges. I hope that over years as we start getting more judges that grew up not playing Pong but grew up really taking some coding classes in high school or college, or who have some familiarity with computers, we’ll get different decisions because it really is intellectually dishonest to treat software differently when it should be all about a particular innovation and making something better. And the frustrating thing from the practitioner’s side, and it has to be even more frustrating from the client side, is just tell us how you want us to describe this and we can describe it that way. But it seems right now what we have a Supreme Court that just says “NO.” No, not that way, no, not that way, no, not that way either. And then Congress is completely absent on this issue. And the patent office is left to try and reconcile all this because they actually get it. They understand that this is about innovation. But how do we be true to this law and be true to what we know is correct and giving patents on innovations? We’re in this hodgepodge space where we are left to try and claim it however we possibly can. And I think to some extent you’re better off going back to the way software was claimed in the 70s, you know, after the Supreme Court essentially said software is not patent eligible. You just claimed a machine that did certain things. And that just bothers me because that’s so much form over substance.
KAPPOS: Right, right. I agree, Gene, I’ve always had a problem with claiming software as a special purpose computer or special purpose machine. I’m of the view that you should just call it like it is, it’s a computer program. It’s an algorithm for doing X or substantiated as a program. And the steps of the program are either patentable or they’re not. That’s where I think we fortunately got some much better tools now to help us in that regard than we did in the past. We’ve got the Nautilus decision, we’ve got KSR that I mentioned before, we’ve got a whole slate of decisions covering what we used to call 112 6th paragraph, which is means plus function language and functional language. I wrote a paper, published now in Stanford Journal of Law and Technology, on the developmental law in that area and it was a hard paper it turned out to get published. Not that Stanford SJLT didn’t want to publish it, they did. But what kept happening is each time it went into the publication cycle a new court decision came out and we had to pull the paper back and add the new court decision to it because the law is developing so much in that area. I think it’s a good thing though and it’s finally bringing us to a point where we’ve got the analytical tools available to us to make sure that algorithmic claims of any kind, whether they’re software implemented or otherwise, are properly patentable if they meet the right requirements.
This brings me to the other point, Gene, which is that I have a lot of respect for the system in the U.S. at the end of the day. It’s not at a good place right now, as you and I know, coming off the back of CLS Bank. And there’s unfortunately blatant discrimination going on now disadvantaging software based innovation, software implemented innovation from patent coverage. But hopefully that will come back and we’ll have chances to correct it. The bigger picture though is I’m glad that software has not in recent era by statute or by judge-made law been declared unpatentable per se as it has been in some other places. The U.S. has instead done the right thing, which is to struggle with an admittedly hard problem. We’re struggling with it, but that’s the way you eventually find a way and that’s the way leaders deal with leadership is to struggle with hard problems. You don’t have any headlights to follow because you’re in the lead. You’re groping around, you’re looking for a solution. We’re still doing that, but I think we are making progress and we’re getting closer.
QUINN: As hard as these decisions really are, and 101 decisions no matter how easy they seem on the surface are extremely difficult, but the issues are only going to get harder as we get through the next wave of litigation that’s going on in the district courts and working its way up to the Federal Circuit. Some of those cases deal with much more substantive technologies. One of the ones that’s kicking around relates to pausing live TV and then going and watching it in another room where you left off. That’s revolutionary. All of the cable companies, Direct TV, Dish TV they all run ads about how wonderful that is and how you can’t live without it and if something like that isn’t patent eligible then I think we’re going to really have to have a lot of difficult conversations. Anybody who’s ever done anything with software has to know that is not trivial. Cable and satellite TV companies were so ecstatic when it happened and it is so useful to consumers. So I think we have had a little bit of bad cases or bad claims make bad law. And even with those bad claims the cases were difficult, with numerous difficult issues. So we’ve have more trials and tribulations in store before we get out to the other side.
KAPPOS: Well, as you say the technology is getting more complex all the time, and certainly there are examples of both kinds: important and difficult innovations embodied in software, and trivial, non-inventive implementation details embodied in software. But I’m for keeping going and keeping going on the struggle to sort the inventive from the non-inventive. I think it is a mistake to give up by adopting these litmus tests, as in, well if it’s abstract then just throw up your hands and declare the whole thing to be unpatentable. That’s a bad phase we’re in right now. But that’s not the way we usually do things in the U.S. As we’re seeing right now it’s not very principled.
QUINN: Right. And I fundamentally just can’t believe that when push comes to shove any judge on any level would find that IBM’s Watson computer and the processes that it enables are not patent eligible.
QUINN: Watson is really the first generation of the Star Trek computer that we all grew up looking at and saying wouldn’t it be great to have this computer you could just talk to and it would give you any answer on any subject. That’s what IBM created.
KAPPOS: Right, yeah. Yeah that’s exactly right. I’ve used that example, too, before to say how in the world can you look at that and say that it’s not innovation meriting patent protection?
QUINN: Exactly. All right. Well, I really appreciate you taking the time and I think we went longer than we were supposed to so thanks for staying long on the conversation. It’s always good to talk to you.
KAPPOS: Likewise, Gene. It’s a pleasure to speak. You know, I continue to marvel at your insightful and piercing analysis of these very tough issues so just keep going you’re doing great work.
QUINN: Well, thank you, thank you very much.