2012 Kastenmeier Lecture


– Alright, good afternoon. My name is Margaret Raymond, and I’m the Dean of the University of Wisconsin Law School. It is my great pleasure to welcome you to what is, by my count, Peter, you can correct me if I’m wrong, the 19th Kastenmeier lecture. Thank you so much for joining us today. Let me start out by
thanking the staff members who’ve done such a beautiful job preparing for today’s lecture. If you’d give Teresa Evans and Kim Frank a round of
applause for all the work that goes into an effort like this. I’d appreciate it. They make it look really easy, but we know it’s not. I’d also like to thank my
colleague Peter Carstensen for his behind the scenes leadership that makes this very impressive,
annual event possible. This lecture is supported
by the fund established to honor Robert W. Kastenmeier, an outstanding graduate of the University of Wisconsin Law School, who served with great distinction in the United States
Congress from 1958 to 1990. During his tenure, Representative Kastenmeier
made special contributions to the improvement of the judiciary, and to the field of
intellectual property law. He drafted the rules
for the House Committee on the judiciary that were
used for the impeachment of Richard M. Nixon, and drafted the articles of impeachment against Judge Harry Claiborne. In 1985, Representative Kastenmeier received the Warren E. Burger award, presented by the Institute
for Court Management, and the Service Award of the National Center for State Courts. In 1988, he was honored by the
American Judicature Society with its Justice Award
for his contributions to approving the
Administration of Justice. I am thrilled that Bob and
Dorothy are with us today, and I hope you’ll join
me in welcoming them to the Law School this year.
(applause) The Kastenmeier Fund was created to recognize Bob’s contributions by fostering important legal scholarship in fields which were significant to his career as a legislator. Intellectual property, corrections, the Administration of Justice, and Civil Liberties, and to do that, we have a dedicated planning committee on which Bob participates, which each year identifies
a speaker whose work is timely, relevant, and exciting. My thanks to the planning committee for all their efforts, and for the truly rewarding
program they’ve put together for us today. Michael J. Remington is a member
of that planning committee, and he’ll introduce our speaker today. Mike has over 30 years of experience in intellectual property law, legal reform, government relations, and lobbying. His work in private
practice and in government, and its impact nationally
and globally could, quite literally fill volumes. He’s currently a partner at the firm of Drinker Biddle in Washington D.C., more important, from our perspective, he is a 1973 graduate of this law school, and continues to be a member of the bar in the state of Wisconsin. Please join me in
welcoming Mike Remington. (applause) – Thank you, Dean Raymond. I told Professor Lemley last
night that I would give him a short introduction, and I asked him what the
shortest introduction that he had ever given a speaker was, and he indicated that
it was Pamela Samuelson. Professor Samuelson, who participated parenthetically in the second Kastenmeier lecture in 1994, and he said, “You all know Pam Samuelson, Pam,” so Mark Lemley, Mark! No, no, not so, not so quick. The… First, a story about Henry Kissinger, who was introduced that
way in Washington D.C. They said, “You all know Henry Kissinger, Henry,” and a voice came from the audience, “But I want to hear the introduction!” (audience member laughs) So, here you’re gonna get a
deserving introduction. Mark is the William H. Neukom, Professor of Law at Stanford
University Law School, and director of the Stanford LLM program in law, science, and technology. In addition to being the author of seven or more books, he’s written well over
100 seminal articles, has taught intellectual property law to federal and state judges, including the Federal
Judicial Center programs, which are hosted by Stanford, and has testified numerous
times before Congress. Mark is the founding
partner of Duri Tangrie LLP, and in his storied career, his varied career, he’s represented clients that we all know. Comcast, Genentech, Google, Grokster, Netflix, Palm, TiVo, and the University
of Colorado Foundation. He’s received too many
accolades for me not to embarrass him, but I’ll mention two. He was named one of the
25 Most Influential People in IP by The American Lawyer, and was named one of the
10 Most Admired Attorneys in IP by IP 360. He is also the founder of, and board member of Lex Machina Inc., something that I’ve
used in my law practice, and I would recommend
it to all you lawyers and practitioners and academics, which is a start up company providing data and analytics about IP lawsuits, and as I said, it can be licensed, or you can sign up. I believe non-profits can get it for free, and law firms have to pay a little bit, which is a good thing. Mark is precisely the
type of law professor with a broad vision and a
sense of the public interest. Not only in IP law, but in competition policy law, that was frequently invited to testify before Bob
Kastenmeier’s IP subcommittee on the House Judiciary Committee. When Bob retired, Mark was just starting his
career as a law professor, so unfortunately, paths never crossed, but he is the paradigm
witness that Congress needs to look to and listen to when it considers legislative changes in law and intellectual property, and competition policy. I give you Mark Lemley. (applause) – Thank you so much. If I understood the Dean’s charge for the committee that
does these lectures, I’m supposed to be timely,
relevant, and exciting, and I can only hope that
you grade on a special curve for patent lawyers.
(laughter) I will do my best to do all three, and certainly at least to the timely bit. The first… Before I start, I do wanna say how
delighted I am to be here, not only here at the University, but here for the Kastenmeier lecture. You heard many of the great things that Congressman Kastenmeier has done, and in a variety of fields. What you didn’t hear, but may know if you’re an
intellectual property lawyer is the really enormous shadow he casts in intellectual property law. There are a lot of people, scholars, lawyers, who have a lot of views in
intellectual property law, and we disagree with
each other all the time, but I would say the one thing on which there is unanimous consensus is that we miss Bob
Kastenmeier in the Congress. The IP world at least was a
better place when he was there, so I am delighted to be here. I wanna talk to you
about software patents. The problem of software patents, and maybe something we can do about them. So, why are software patents a problem? They’re a problem because there’s lots of evidence out there that suggests that they do less good and more harm than any
other kind of patents. On the less good side, we see, we have pretty good evidence that we got, and continue to get substantial innovation in the software world without, or perhaps even despite patents. We saw lots of innovation in software before patenting existed. Even once patents existed, a lot of innovators, if you ask them, if you ask software engineers, will say we hate patents, we wish software patents would go away. Some of them vote with their feet, either by taking an open source route, and choosing a non-proprietary approach, or by joining organizations, standard setting
organizations and the like that do their best to limit the impact of software patents. This has been true I think part because the economic structure
of the software industry, to oversimplify dramatically, isn’t the kind that we normally associate with a need for strong legal protection. In the pharmaceutical industry, you have to spend hundreds
of millions of dollars, and wait years to get
your product to market, and if you don’t have an opportunity to recoup those expenses, you’re not gonna lay the
money out in the first place. In the software industry, by contrast, the iconic invention is developed by two people working in a garage, and by one recent count, one third of Stanford undergraduates are in the process of
developing a software program for sale in the world, usually an app to run on an iPhone, or an Android operating system, but there’s a lot of room for development. Not because I had to get
a bunch of money to hire a bunch of people and work for years to produce this product, but because I had some time and interest. In fact, I’ve talked to some University
of Wisconsin law students who are also in the business of starting software companies. So, less evidence that we really
need software patents. On the flip side, there’s an extraordinary array of evidence that says software patents are creating problems out of proportion to what happens in other industries. There turn out to be a substantial number of bad software patents issued. Patents that the patent
office should not have issued. This is true in all areas of law, but it seems to, perhaps more true in software. We see example after example of things that when they go to litigation, the evidence says, well you know what, people did this before. People have been doing this for years before you filed your patent application. We see lots of over-claiming, and I’m gonna talk in some
detail about that in a minute. It’s not just that I wrote a program, and I claim to own this program, and I don’t want you to use it. It’s that I wrote a program
to solve this problem, and I claim that I’m the first person ever to solve this problem. So we’ve seen patent after patent asserted on things like global electronic commerce. Not a specific mechanism for performing an operation in global
electronic commerce, but the idea of doing
commerce internationally over the internet, or the idea of video-on-demand, or the idea of multi-media, and on, and on, an on. Many of these patents are in the hands of so-called “patent trolls”. The patent troll phenomenon is a phenomenon in which an individual, the most canonic example is somebody who doesn’t invent something themselves, but buys a patent from someone else, and hides under a bridge
they didn’t build, and pops up to surprise the
passerby and demand a toll. We see a substantial
number of toll demanders in primarily the software industry. There aren’t patent trolls really in the biotechnology or
the pharmaceutical space, but by most counts, somewhere between 40 and
60% of all lawsuits filed in the software patent world
are filed not by people who are actually making
products themselves, who want to keep a
competitor off the market, or demand licensing fees, but by people who are in the business of filing patent lawsuits
in exchange for money, and that’s what they make. They make lawyers, I guess. So we’ve got bad patents, we’ve got over-claimed patents, we have patents in the hands of people who many people at least
consider to be unsavory, or not necessarily socially
desirable characters. More to the point though, we have some structural problems with the software patent world. The first of which is we just have too many darn patents in software. Nearly 50% of all patents
issued in the last decade are by some counts, and it turns out to be
hard to know exactly what a software patent is, but by some counts are software patents. The general counsel of
Google estimated last year that all in a smart phone, including of course the hardware, the touch screen, the communications technology, and the apps and things you load on it, might implicate 250,000 patents. 250,000 legal rights in this
little device owned by hundreds or thousands of different people. The result is what Carl Shapiro has called a “patent thicket”. There’s so many legal rights interacting and interlacing with each other, that as a practical matter, it turns out to be impossible to go out and advance and license all of the patents you might
conceivably need to license in order to make these things work. That’s even more true because
some of the other problems that I think are endemic
to software patents, one of which is the difficulty in knowing what exactly it is that they cover. We have in chemistry and in biotechnology a pretty good scientific
language that allows us to define the scope of things. If I give a chemical to a chemist, and I asked them, “Is this chemical in this group described in written chemical notation?” they can answer that question, and every chemist will tell me yes it is or no it isn’t. We can’t do the same thing with software. We don’t have a structured language, at least that we use
in software patent law, maybe the computer scientists in the room will tell us
we’ve solved this problem, and we have a structured
language for software concepts, but we don’t use it. We use English words to
try to describe what it is that we’ve done in our computer programs, and then, being lawyers, we fight over the meaning
of those English words. We fight not only over the meaning of the complex, technical ones. We have federal circuit
disputes over the meaning of the words “a”, “or”, “when”, “through”, “to”, in fact, “a” has the subject of a
dozen different decisions in different directions saying
it has different meanings. The result of which is, not only are there
250,000 patents out there, that I have to try to
read through them all, but if I tried to read through them all, I wouldn’t know in advance of actually litigating the question, which ones might conceivably cover the product that I’m proposing to make. Even if I found all of those, well it turns out I might
still not be alright because it takes the patent office, on average, 4 years to get a software
patent out of the office, which means that if I’m
producing a product, it might be years down the
line of selling the product before the patent even issues, and I have the ability
to figure out whether or not it’s within the
scope of my technology. That time can be extended
by a number of practices that patent owners engage in to delay the application even further to modify the patent
claim during that process with the result that it turns out to be extremely hard even
if we could devote the time to know what patents are out there. Tim Lee and Christina
Mulligan have estimated that given the number of
software patents in the world, and given the number of
software companies in the world, if every software company
were to hire a patent lawyer to read through all of the patents and figure out which ones they
probably need to worry about, and even if that patent lawyer spent only 10 minutes per patent, we would need two million patent lawyers working full-time just in
software patent clearance. Now that might be good
news at a law school, but I’m not sure it’s
good news for society. In fact, what we’ve seen is the
development of a patent arms race in which software engineers
get patents not because in most circumstances they think, I need this patent to
protect my innovation, but because they think, I need this patent to protect
myself against other people who have patents who might sue me. We’ve seen the consequences
of that arms race in the telephone, wireless communications market
in the last couple of years in a really dramatic way. In the last two and a half years, smart phone companies
have spent between 15 and 20 billion dollars buying
other people’s patents, to then use as chips in this arms race. Leave aside the amount of money they’ve spent
patenting their own inventions, hiring their own lawyers, taking time out of their engineers’ work to have them write and
review patent disclosures. Then they’ve spent, by my estimate, probably three quarters of a
billion dollars paying lawyers to litigate those cases, and they’re not done yet. They’re an endless array, there are 60 or 70 patent lawsuits in the smart phone industry among the major players waiting
still to be resolved even as we’ve resolved a couple. The result is an estimate
by Bessen and Meurer that software patents are a
net a drag on innovation, not a benefit. How one estimates the
size of this drag is hard, but they estimate just from patent trolls that the social loss
is 500 billion dollars over the last 20 years. Well… If we’ve got a problem here, and I think we do have a problem, the question is what to do about it? One thing you might do about it, one proposal you hear on a regular basis, particularly from engineers is, let’s get rid of software patents. They’re not doing much good, they’re doing a lot of harm, let’s get rid of ’em. I have a couple of problems with that. Let me start with, well- First off, I think it’s a categorical problem. It’s not the case that all
software patents are bad, it’s the case that bad
software patents are bad, and it’s the case that some
of the ways we’ve structured the system create problems for society, but that doesn’t mean there
aren’t useful inventions in software out there that are in fact deserving of protection. I think if we simply say, You know what, throw all of these things out, we throw out some desirable patents. Patents that might be
encouraging innovation, as well as patents that
might be more problematic. Second, if we were to try to throw
out all software patents, we’d have to try to define
what a software patent is, and that turns out to
be surprisingly hard. In fact, of the various studies that
I’ve been talking about, each one of course has a
definition of software patents that helps them determine their numbers. When you place those studies side by side, the overlap rate between
them is about 33%. That is to say of the
patents they estimate to be software patents, two thirds of the time they disagree. What one calls a software patent, another one doesn’t
call a software patent. You can get some sense of the problem by asking… this question: The gasoline electric hybrid car. What is it that’s inventive about a gasoline electric hybrid car? It’s not a gasoline combustion engine, we’ve had those for a long time. It’s not an electric battery either, we’ve had those for a long time. What’s inventive is the
combination of the two, the controller that allows
you to switch between the two at the right time so that I switch from gasoline to battery when I can in order to save gasoline. That controller is a piece of software. In fact your car is basically a computer. If you drive one built
in the last 25 years. Is that a software patent? We say you can’t patent the
idea or the implementation of your hybrid gasoline electric engine because it’s controlled in software? That doesn’t seem like
what most people think of when they think of a software patent. Maybe it’s something
that runs on a computer, except of course that most
people don’t use software on a computer anymore. Most people use software on
some other mobile device, or a specific piece of
implemented hardware that happens to operate
like almost everything in our society right now
by having computer control in part. I’m not sure if we wanted
to abolish software patents that we’d have a very good way of figuring out which
ones we wanted to abolish, and of course once you say
we’re singling out this area of software patents for special treatment, lawyers are going to do
everything in their power to take mine from this
category and move it over here to this category. I’m not patenting software
that controls a pizza oven, I’m patenting a pizza
oven that just happens to have a computer
program that controls it. Finally and perhaps the most– the best objection of all to… let’s abolish software patents is it’s never gonna happen. There are hundreds of thousands, perhaps a million issued software patents. There are a lot of people
who have a vested interest in those patents. Even if we thought we could do it, even if we thought it
was a good idea to do it, it just ain’t gonna happen. If we can’t abolish software patents, maybe we can just get rid of the bad ones. Ramp up the PTO, spend a lot more time
evaluating these things, dig further for prior art, and get rid of the bad software patents, and then we’ll be in a
much better position. I’ve got two problems with that. The first is that… I think most of the time when we do that, we’d be wasting our money. It turns out that most software patents, like most or all kinds of patents, issue from the patent office at the result of you spending 20 or 30 thousand dollars, and are never heard from again. They go sit on a shelf and
nothing happens ever again. We could weed all of those patents out, just like we could weed
out the silly patents that people encounter from time to time. The patents on a method
of twisting a swing that issued to an 8 year old, or the patent on a method of exercising a cat with a laser pointer, there are actually trying
to be several of those, how they distinguish
each other I’m not sure, but it’s silly to waste our
time weeding those things out because nobody’s really
gonna care about those. We might want to try to
weed out the ones that are both bad and causing harm to society that are in fact being asserted. We could do that in court, we could ratchet up the standard and say software patents, we presume they’re invalid or you’re gonna have to
show something special to allow us to justify
this software patent. I think doing so still leaves
both the troll problem– The business model that’s based
on asserting these patents– and settling without going to
court in exchange for money; and it still leaves a
lot of valid patents. Let’s say that 90% of
the patents out there in the software world are invalid. The Google general counsel says there are 250,000 patents on a smart phone. We can change the system and revamp it, and we’ll come back to them and say, good news, now there
are only 25,000 patents on a smart phone, your problems are over. Well not exactly. Third, maybe we might try to
solve the boundary problem. I said we can’t tell exactly
what a software patent covers, and that is a problem. Bessen and Meurer suggest
that what we need to do is have clear, defined boundaries. Their syllogism is if you
can’t tell the boundaries, it ain’t property, and if we want to have
software patents work, we need to make them
more like real property. Draw lines, put in a recorder’s office, and we’ll all know where they are. I don’t think that’s
actually possible under the legal system we have created, under the peripheral
claim construction system as long as we don’t have
a structured vocabulary that everyone agrees on for
defining software inventions and software scope, we are always gonna fight as lawyers about the meaning of these
words and what they cover. As long as we’ve got
peripheral claim construction, we’re always gonna have lawyers trying to draft their patent claims
as broadly as possible, trying to reach for the borders
and push the dubious cases. If we did… manage somehow to perfectly
define the scope of all of these patents, it might just tell us
exactly how badly off we are. What we might discover is, yeah you know what, we really do have 50 patents that claim to be the only way to use video-on-demand. We really do have, as of years ago I’m sure
the number has gone up now, 7,000 patents that are essential to the practice of 3G wireless technology. All of which, at least claim to be things
you have to use if you want to implement 3G wireless technology. What’s going on here? I think a large part of the problem comes from the broad language
of software patent claims. We claim software. We let software patentees get
away with claiming software. Not by reference to the computer
program they actually wrote to solve a problem, but by reference to the problem itself. We let people claim software
in broad functional terms. My patent is not on the system for video-on-demand I implemented, it’s on video-on-demand. It’s on solving this particular problem. It’s on global electronic commerce. Not the program I actually wrote, not even the sort of
approach I actually took to solving that problem. If you look at this from
anywhere other than software, it’s bizarre. We would never do this in
any other area of technology, at least not to the extent we do here. Imagine that I’m a pharmaceutical company, and I have come up with a new… chemical which I think is
useful in treating cancer, but instead of claiming that chemical, or a group of chemicals that
looks structurally like it that perform that function, I’ve decided to claim atoms configured in a way that cures cancer. That’s my invention. I have here an example of an atom, a configuration of
atoms that cures cancer. So I have done it, but what I claim is
not this configuration, not this solution to the problem, but idea of solving the problem itself. We’d say that’s crazy, we’d never allow it, but that’s precisely what
we do allow in software. Understanding why and
what we might do about it I think requires a little bit
of a step back into history. 200 years ago, people patented inventions. They patented the things
that they actually created, and when I filed a patent, what I did was I sent in a
description of my invention, the machine I had actually built, and that is what I patented. If you copied my machine exactly, you were an infringer. If you didn’t copy my machine exactly, but you made one that
looked pretty darn similar, you might or might not
have been an infringer. To figure out whether you were an infringer under this
central claiming system, we looked at your invention, the machine you actually built, we looked at what the defendant did, and we asked how similar are these things? Are they close enough that
we think this is within the scope that we wanna
grant the patent owner? We had some rules and
heuristics for trying to figure out what that scope was. If you were a broad, pioneering inventor, if you opened up a new field, you got broader protection. If you’re an improver you
got narrower protection, but by and large, the question, is this thing close enough to the patentee’s invention
to be an infringement? was a question we just left
case-by-case to the courts. Patentees didn’t like this, so they started writing
in the 19th century so-called “patent claims”. Along with my patent
describing my invention, I might lodge in… a statement that says, you know what? I’m not claiming just this machine. I don’t think it’s limited in this, I think other machines that work in these other ways would be included. I’m claiming this broader principle or this particular piece, this is what’s new, and this is what I think people shouldn’t be able to copy. As that claiming practice developed, we actually moved from a world of what I call… sign posts. A flag in the ground that says, “I invented this right here,” to a world that looks
more like fence posts, to peripheral claiming in
which the goal of the claim was to define the outer
boundaries of my invention like the goal of a
property deed is to define the outer boundaries of property. In doing so, we replaced a focus on the thing itself, what did I build, with a focus on what did my lawyers write? What did the lawyers write down as the scope of my invention, and how does that compare
to what the defendant does? Today it’s that that we ask when we ask about patent infringement. We don’t ask, does the defendant’s product look anything like the
plaintiff’s product? We don’t care. We ask, does the defendant’s product look like what the plaintiff’s patent lawyer wrote as the legal description of
the bounds of their invention? Once we started allowing that
sort of peripheral claiming, patent lawyers got
creative in writing broader and broader boundaries. One of the ways they got
creative in writing broader and broader boundaries was to start claiming their
inventions in functional terms. It’s not that I invented this machine, it’s that I invented a machine
that solves this problem, and any machine, however it looks, that solves this problem is
within the scope of my patent. When the Wright brothers filed
their patent on the airplane, the Wright brothers invention
in the airplane was actually a particular means of
controlling the relationship between the warping of a wing
and the movement of a rudder. We already had planes, we had wings, we even knew how to
turn planes in a theory. You tug on the wing, you bend its shape, and then there’s a different
air pressure coming on one wing and then the
other and the plane turns. The problem was of course, the plane turned and then
it flipped over and crashed. What the Wright brothers did, which was a very clever
and useful invention, is they tied the rope that they used to warp the wing on the one side to a rudder that would pull
and turn on the other side, slowing the air foil there so that the plane didn’t
swing around and crash, but instead gradually turned into a… into a movement. That was a useful invention. The Wright brothers’ patent
lawyer didn’t claim… tying a rope between
the wing and the rudder so that when you tug on one, it necessarily tugs on the other. They wrote a much broader claim. A functional claim to the ability to control
the warping of the wing. When Glenn Curtiss came
up with a different, and frankly better technology, the aileron, which we still use in some form today that allows the movement of
particular parts of the wing, and independently allowed the
pilot to control a rudder, but didn’t connect them
in a mechanical way. The Wright brothers sued
him for patent infringement, and they won. They won because… the court said… this is within the function. You’re doing the same thing, you’re just doing it in a different way, but it’s within the scope
of your patent claim. A frustrated Glenn Curtiss
is reported to have said, “The Wright brothers think
their patent is so broad, that if I jump and down and flap my arms, I’ve infringed it,” but that patent was upheld and
enforced for over a decade, basically stopping airplane innovation in the United States until the eve of World War II and the US
government stepped in and said, “Alright, everyone is going
to cross-license their patents and you’re actually
going to build us planes because we need them.” That sort of functional
claiming was quite popular in the early part of the 20th century, but it became less and
less popular with courts because it led to some really
remarkably broad claims. The courts started increasingly
to crack down on it, until in 1946 in the Halliburton case, the United States Supreme Court said, “You can’t write a patent claim that uses broad functional language at the very point of novelty. You can’t say my invention is different than everything that came before in this respect, and then say this respect is
I’ve solved this problem.” You’ve gotta be limited to the actual way in which you solved the problem, not to the mere concept of
solving the problem itself. Patent lawyers weren’t happy with that. They went to Congress
and when Congress passed the 1952 Patent Act, P.J. Frederico’s term I think is actually that they did
not reverse Halliburton, they rendered obsolete
the Halliburton decision. They rendered the
Halliburton decision obsolete by passing a new provision
of the patent statute in what’s now called Section 112F. Section 112F said, “You can write your patent
claim in functional terms, but if you write your patent
claim in functional terms, that’s a signal for the courts not to give you every possible function that might perform those things, but to go read the patent specification, read the text of the patent, and limit you to the
actual machine you built, and equivalents of that machine.” If my patent claim element says, “means for processing data,” a literal interpretation of that language, any means for processing data. Computers, is a means for processing data. A cell phone is a means
for processing data, but so is a pencil and paper, or an abacus, or the human brain. So a literal interpretation of that functional
language would say anything that processes data, that counts as within
the scope of the claim. What Section 112F say is, If you use that term, if you claim in functional language, we don’t give you every possible means for processing data. We go read your patent, we figure out which one you used, and we limit your claim to that product, and equivalents thereof.” Maybe it’s an iPad and
things like the iPad, because that’s what I use to process data in the actual invention. Patent lawyers today will tell you, means plus function
claiming is undesirable. You’re limiting your claims, you don’t wanna do that
if you can avoid it. What’s happened in software
in a way that hasn’t happened in any other area is
that we’ve figured out through some combination of history, happenstance and I think mistakes by the federal circuit, how to claim in functional terms without triggering this
language Section 112F. What the federal circuit has said is, “How do we know if something
is functional claiming? We think it’s functional
claiming if it uses the words ‘means for doing x’, and if it doesn’t use the
words ‘means for doing x’, it’s probably not functional claiming.” It’s a rebuttable presumption, but the federal circuit really, really, really doesn’t
like to have it rebutted. As a practical matter, what the patent lawyers
tell their clients is, don’t use the words “means for doing x”, that’ll be a signal to
trigger this narrowing rule. What could I do instead? It turns out that what I could do instead is not say “means for processing data,” but to say instead, “A computer for processing data,” or “A computer for implementing this word processing program.” Practically speaking, I haven’t done much to
limit the scope of my claim when I’ve introduced the
concept of a computer or a processor, or a… computer system. In most of these technologies, you gotta have a computer in
order to make the thing work, at least for some definition- sufficiently broad definition
of a computer or a processor. What I’ve done is I’ve
gotten out of the scope of Section 112F by the
simple artifice of saying, this isn’t a functional claim, this a claim to a thing. The thing is a computer. It happens to be programmed
to achieve this new function, but any computer programmed in any way to achieve this new
function is within the scope of my claim. What I wanna suggest is that that’s wrong, and that it’s a pretty
simple thing to fix. If we treated software inventions… as… software, not as hardware machines that happen to have some unspecified
computation programmed into them. If we said, You know what, your invention is software, then what we do is we’d ask the question, alright do I actually give you some sort of structural limit in my patent claim in the computer program itself, or am I merely claiming the function? By merely claiming the function, we ought to invoke 112F, and we ought to go limit you
to what you actually built. There’s been a move in the
federal circuit case law in the means plus function
claim language to say, you’ve got to actually
give me an algorithm, you’ve gotta give me at
some level of detail, that’s an issue that’s
gonna have to be worked out, but at some level of detail, you’ve gotta tell me
how you actually propose to solve this problem. If you don’t, then your patent is invalid, it’s not definite, it doesn’t tell me what it
is that you’re claiming. If you do, under Section 112F as I think it should properly be interpreted, your invention ought to be
limited to what you actually did, not to the problem you solved. I think if we do that, we get rid of not all of the problems of software patents, but we’d get rid of a lot of them. The overlap problem, the thicket problem comes from a bunch of people claiming their inventions in broad functional terms
and not limiting them to what they actually built. If that goes away, a lot of the… overlap problem goes away, a lot of the bad patents, invalid patents problem frankly goes away because most of the
reasons that the patents on video-on-demand are invalid
is because I’m claiming to own all video-on-demand, not what I actually built, what I actually designed. We may solve another problem as well. We are currently fighting a… battle over whether or not software inventions are patentable subject matter. In the courts, federal circuit just
this week decided to take a case en banc in the probably futile hope of resolving once and for all the question of when is
my software invention patentable subject matter? The problems that I think that the courts are confronting in these
patentable subject matter cases aren’t really that a computer program is an abstract idea or a
mathematical algorithm. Usually it isn’t. The real problem they’re
confronting is that it’s claimed in such broad functional terms that the program itself
is not being invented, it’s not being patented, it’s the concept, it’s the solution to the problem
that I’m trying to patent. We could end up with narrower, but valid software patents that did a lot less harm in society. They were a lot less likely to be passed down to patent
trolls 10 years later who asserted them against
different technologies that happen to fit within
these broad functional claims. They are a lot less likely to create this overarching patent thicket that prevents anyone in
the software industry from even being willing to
look and see what’s out there for fear of discovering
250,000 software patents. It requires us to not
even change the statute, requires us in my view just
to interpret the statute the way it was meant to be interpreted, to limit us to functional claims. Requires us to do some thinking about how broad that right ought to be. The “and equivalents thereof” language has a lot of magic in it, and that magic is gonna
have to be played out, and that’s not a panacea solution. We’re gonna have to decide, we don’t want you to presumably be limited to only the exact piece of code I wrote, that’s gonna be awfully narrow protection. We don’t want you to
hold the entire concept of the function, that’s too broad. We’re gonna have to find that happy medium by deciding how broad a range of equivalents we will allow. I think that’s a solvable problem. It’s a problem we have to solve
case by case in the courts, but I think it can be done. That leads me I think to the final issue, and maybe the objection that if I were a software patent writer, I would raise on this issue, which is, isn’t it unfair to prevent me
from controlling functions? Software functions, it’s what it does. I wrote software that
performs this function that solves this problem and I should be entitled to control that. I think the answer here is that you’re no more
entitled in software than you would be anywhere else to control not just the
actual thing you invented, but the concept or the problem, the market that it solved. The point of code is to function. That’s what software does, but that’s also true of jackhammers, and cardiac ablation catheters. You don’t get to claim- if you invent the jackhammer, you don’t get to claim
means for breaking rocks, and say you know what? Any way you use to break rocks, I own because here I’ve built one. If you invent the cardiac
ablation catheter, you don’t get to claim
means for clearing arteries, and then assert them against
entirely unrelated ways of solving the same problem. Somebody who uses a drug
instead of a catheter, and similarly in software, the problem is we’ve allowed, through this historical accident, people to say, to let the computer stand
in for the structure, even though it does no work, and give you control
over the market itself. Doesn’t mean there won’t
be functional claiming, I think there will always
be functional claiming. Instead what it means is
your functional claiming is gonna be recognized
as functional claiming. It’s gonna be limited to things
that you actually invented. It does mean, I think, that most of the time, patent… patent owners will not own markets. They’ll own inventions, but those inventions may face competition from other different
solutions to the same problem. My answer to that is deal with it. That’s what happens everywhere
else in the economy. You come up with a great
idea for a restaurant and you face competition. It’s even what happens everywhere else in the patent system. You come up with a
great idea for a statin, that’s great. You can have a patent on it. You can prevent generic companies
from making that statin, but you don’t own statins. You don’t own the idea of any drug, however structured that treats
this particular disease. It is true that on occasion, the identification of the problem is itself the invention, and it’s so fundamental that
maybe we should give it to ’em. Some people would put the Wright brothers
invention in that category. On balance, I think it does a heck of a lot of harm for an awfully small number
of patents that we really, really do think ought to be that broad. Patents sometimes insulate
their inventor from competition, but that’s not their goal. Their goal is to prevent imitation, not to prevent competition. Indeed it’s competition itself, I think in many industries
including software that’s the greatest driver of innovation. We stifle that competition at our peril, as I think the experience is
what software patents proves. Thanks.
(applause) Sure. – [Woman] : Take questions… – [Man] : Obviously other countries are less than sympathetic to the idea of software patenting. Do you know, is there a country, national system of courts that does a better job handling this issue? – [Mark] Most of the other countries, including Europe try some version of the abolish software patents route. They usually– I think they frankly fail at it. What happened when during
the period in the 1970s and 1980s when we did that was people wrote their
software patent claims as something other than
software patent claims, and that’s what we see
in Europe right now. It’s not that IBM doesn’t
patent any software inventions in Europe, it’s that it doesn’t call
them software patents. Then we have a whole series
of cases in Europe dealing with the “as such” limitation. What we said is you can’t
patent software “as such”, but you can patent a new machine that has software embedded in it. My new machine, gosh, it’s a laptop, and it sure looks exactly
like the old laptop, except that it’s got new software in it. We went through this in the United States. The United States courts actually said, “Well that makes it a new machine.” It’s physically different. I’ve loaded the program into memory, and so the 1s and 0s, the microscopic electronic switches are differently configured
in this computer than they were before. It’s a different thing. That’s all a fiction I think. There are a few countries
that have basically followed the US lead and allowed software patents, although they haven’t tended to be countries sufficiently
large that we’ve seen a lot of influx of software patents. That might have as much to
do with our litigation system as it does with the patenting system. Other countries don’t seem
to have this same problem, but generally I think
that’s because they’ve tried to get off the train at the
abolish software patent stage. That’s led to other problems, I don’t think it’s worked, but it’s at least made
their experience different. (mumbles) You have a microphone coming to you. – One way that you see
in the European Union is to talk about a technical art. That comes up in business methods. If I were to push you a little bit about what types of means would work, it seems like you might have some overlap with some notion of the technical art. I don’t know if you
want to comment on that or if I have the right conception of your solution. – [Mark] : That’s a fair point. What does it mean to not require- not to allow patenting
of software as such. Europe says you’ve gotta
have a technical requirement. I think what Europe
means by that mostly is a physical hardware requirement, rather than a- I’m sorry? – [Man] : Doesn’t have to be that… – Exactly. That’s where I might get off the train, I think that’s the question. If we focus on hardware, I think we’re focusing on the wrong thing. I think that’s what led us
astray in the United States, we focused on the question of, is this software making a new machine? and the machine’s got less and less real, more and more virtual
until we threw up our hands in State Street Bank in 1998 and said, this is all a fiction. Some of that on the hardware side is what’s going on in Europe. If in fact what they mean
is something different, which is, you’ve gotta show me working code. Then I actually think I am
more in sympathy with that, although I also think that
they’ve gotta stop pretending they’re not patenting
software in that case. If a “technical” element
of a software invention is a computer program that operates, and can be loaded on a substrate, you are patenting software. I think you should be
able to patent software. Let me be clear. What I’d like to see people patenting is actually the software. Not the high level solution, the concept of solving the problem, but the way you solved the problem. The actual implementation. I think it probably is technical in the sense I would define it. I’m more skeptical that’s
it’s technical in the sense that the European Board would define it. – [Man] Clearly there are many examples
of that software practice that… Can you give some examples of what would be good software
patents under your criteria? Like Page Rank or RSI, or (inaudible) – [Mark] I think there are real inventions in software. I haven’t read the Page Rank claim, so I don’t know whether
the claim is over-broad, but I think the Page Rank invention is at some level of
implementation a valid invention. I think that some of
the stuff that’s going into touch screen technology… Some of it is clearly material science, but some of it is good computer processing that allows us
to recognize certain things, and treat them in certain ways. I think, there the hard
question is gonna be, how broad should that claim be? I think Apple– if Apple solved a
difficult technical problem to get us some of those scrolling bounce patents, they’re entitled to a patent. I’m less persuaded that
they’re entitled to a patent that prevents anyone else
from solving that same problem in a different way. To me, the fights in the smart phone industry, in the Apple-Samsung case ought to be not about
did you actually have this scroll down feature that
allowed it to bounce back, but did your scroll down
feature in fact work in a similar way to Apple’s, or did you actually solve the
problem in a different way? – [Man] It seems to me that your argument about software patents would
have to apply with equal or even greater force to
business methods patents, and I’m wondering if there’s any room left for what you would consider to be a good business method patent. – [Mark] That’s an interesting question. I’m not… It’s a fair point, and it’s not one I’ve
thought through in detail. Off the top of my head, I wouldn’t want to exclude the concept of a business method patent altogether, although I do think it
is gonna be the details of implementation that are gonna matter, and not the concept itself. I wouldn’t give FedEx the patent on guaranteed overnight
package delivery 35 years ago when they come up with this idea. I might give them a patent
on a particular system for optimizing the moving of the packages, even if that turns out
to be key to their idea. Part of the problem I think is that it’s actually hard today in the modern world to
think of new business ideas that aren’t also
implemented in a computer, so it turns out to be hard to
separate the software piece and the business method piece. That’s why I think the court allowed business method patents. Once it allowed software patents, it couldn’t actually come
up with a structured reason to say, well if we let you patent a computer that’s programmed with code, we won’t let you do that if
the code you’ve programmed is code that happens to
control financial instruments as opposed to something more technical. So I wouldn’t get rid of
business method patents, although probably what I
said of software patents in terms of harm and good is, if anything even more true
with business method patents. That is there are a lot more bad ones and probably fewer good ones out there, and my guess is that
application of this principle is gonna hit them harder than software patents, if anything because
there are fewer of them that have a real– could go down and find
an algorithm and say, yeah here’s an actual technical, to use Shuba’s word,
problem that I’m solving that’s at the base of my invention. – [Man] Where do data formats and interoperability fall into all this? For example, mp3 and then what if we want to use our open source software for mp3? – [Mark] Right. Right now under the law, the answer is you’re in a world of hurt. Both on the interoperability problem, that is I can patent the format itself. I sometimes can patent the connection, the actual interoperability code, and while open source
is an effort to get out from under the perceived overly broad strength of intellectual property, it doesn’t work so well for patents. It works quite well for copyright because copyright requires that people be in a relationship, that I get the code from you. There are lots of companies who get sued for patent infringement for
their open source software. Google’s Android is open source. It’s been the basis of a lot
of these smart phone lawsuits. My solution doesn’t make
that problem go away, that is it doesn’t– there’s nothing about it that
says you couldn’t ever patent a data format. If in fact that data format
has to be adopted exactly because it’s an industry standard, I want it to work with mp3, and so I have to use even the narrow one, I think it doesn’t solve that problem. To me the two ways to solve that problem are you could solve it in a
standard setting organization by saying these are
standards essential patents, and you have to commit to license them on certain specified terms. You can’t get an injunction. We’ve seen a lot of standard
setting organizations do that, and we’re seeing right now a lot of litigation for the first time really enforcing those requirements. In the Microsoft Motorola case, and in some other cases saying yeah when you made that commitment, you actually made a commitment, and you may be able to collect
some money in a royalty, but you can’t stop people
from using this standard. If we didn’t do it that way, you could solve it in courts or in legislation by basically creating an interoperability defense. We did that in copyright law in the courts with the fair use defense. It might require more
statutory action in patents, but you could certainly say if the only reason I’m using
your patented invention is to make my product, which is not based on
the patented invention, work with this other one, that’s okay, but under the law right now, and under my proposal I think
that problem still exists. – [Man] You mentioned that
one of the problems of the Patent Office is issuing patents without consideration
of the full prior art. It’s true of other
areas, not just software. Are there mechanisms you
see who are introducing a better quality of prior
art in the prosecution stage. – [Mark] Maybe. It’s a hard problem and it’s one I’ve struggled with. There are a couple of systemic problems. Patent office gets 550,000
applications a year now. They do not have a lot
of time per examiner to devote to these things. Having them go out and
search for prior art more than they do now just might not be feasible, at least unless we’re willing to throw a lot more money at the problem, raise the fees. Even if they do that, I think in software a
lot of the art is not art that’s easily found. It’s, I did this 10 years
ago with a media lab at MIT, and we were practicing that. That’s not documented in some clear way in a publication. It’s not itself patented. I’m not sure that doubling the amount of search time we give to examiners is gonna solve that problem. There have been proposals
to get at this problem in some other way by
finding better sources of information. There’s this peer to patent idea. We’ll ask a bunch of
engineers over the web, and they’ll send stuff in. We now allow under the
new statute pre-issue and submissions so a competitor could mail in prior art. Maybe that will work, although Chris Catropi,
Bhaven Sampat and I have an interesting new study that suggests that patent
examiners focus their attention almost entirely on art
they find themselves, and not art that comes from outside. Even art that comes from outside from seemingly reliable sources like the European Patent Office, which rejected a counterpart application. We don’t know what the mechanism is, why that’s true. Is it a kind of rational
time-saving device? Do I think this art was all
sent in by the applicant, and it’s all horrible? Depending on the answer to that question, finding better people to
mail in better art might or might not be a solution. Then we’ve got administrative revocation. We’ve got post-grant
opposition coming up next year. We’ve got inter parties reexam– there may be ways to get
that art in front examiners in a subset of patents that we care about more
cheaply than doing it in litigation and that might
be a reasonable compromise. – [Man] Can you focus on the solution of changing the patent term? Lots of schools of thought on this, but one would be chemical
entities deserve a longer term because that respects that industry, and software patents
deserve a shorter term, and that would respect the industry. – [Mark] I think it’s
certainly a true statement as an economic matter. The scope of the pace of change in software works differently than the pace of change in chemistry, and then we layer on to the ladder the FDA approval process. We’ve done some things in
the pharmaceutical side to try to lengthen the patent process, and also to create substitutes for lengthening
the patent process. The new generic biologic, or so-called “bio-similars bill” creates a quite long period of data exclusivity over
and above the patent that may run coextensively with it, but that gives you in some sense a guaranteed minimum floor of protection. A lot of people have
suggested what we need are short software patent terms. Maybe. I guess there are a couple
things to note about that though. One is as long as it’s
still taking four years to get software patents
out of the patent office, you’re still not really
focusing on the actual time that the invention’s most relevant. One of the reasons we see so
much patent troll litigation in software is by the
time my patent gets out in the patent office, it’s almost always directed at a prior generation of technology. I’d be okay with a shorter
software patent term if it were coupled with some mechanism that actually got it faster review. Maybe what we ought to do is say no pharmaceutical
company needs their patent in three years, they’d be perfectly happy to
sit for a couple more years, but software companies
really need an answer, so we’re gonna get you
faster protection that may expire a little more quickly. Brian Love at Santa Clara does have a study suggesting that most suits filed by
practicing entities are filed in the first five years
of a patent’s life, and most suits filed by
patent trolls are filed in the last five years of a patent’s life. I don’t think you can then say if we cut off the last five years, no more trolls. They might just move to
an earlier five years, but it is suggestive that maybe most of the harm is coming at the end of a software patent’s life, and not at the beginning. Are we… (applause)
Thank you. Alright. – Now sorry, I told Mark he can’t go anywhere because we have a present for him. Sometimes when you go someplace and you do something really important, they give you a trivial gift like a box of chocolates. Sometimes they give you a lovely, but insignificant gift, something made of crystal. Here we give you an iconic gift, which is another word for really, really ugly. This is a gargoyle. – Is it made of chocolate? – No, you wish.
(laughs) This is a replica of the
gargoyles that came off of our no longer in
existence law building. It actually means that
you’re a tremendous friend of the University of Wisconsin Law School. This is what we give only
to our very best friends. I promise we’ll send it to you, you don’t have to carry it around. – I was going to say, I might have to check luggage now. – Thanks so much. – Thank you.
(applause) – I’ll take care of your gargoyle. If you’ll all join us in the atrium. (audience talking)

Leave a Reply

Leave a Reply

Your email address will not be published. Required fields are marked *