About This Blog
Automating Invention is Robert Plotkin's blog on the impact of computer-automated inventing on the future of invention and patent law.
- Artificial Invention
- Design & Engineering
- Evolutionary Computation
- Genie in the Machine
- History of Computing
- Human Creativity
- Intellectual Property Law
- Philosophy of Computing
- Software Patents
- Technology Industry
- Bilski, Business Methods, Software Patents, and the Supreme Court
- The Case for Software, but not Literary, Patents
- Turing Tarpits and Nonobviousness
- Implications of the Bilski Case for Software Patents
- Guest Blogging on IP Watchdog
- In re Bilski: Ten Reasons Why
- Patent System at a Crossroad
- Are Quality Software Patents the Answer?
- National Magazine on Business Method Patents
- The Software Patent Cultural Gulf
- U.S. Ruling on Business Method Patents
- European Patent Office and Software Patents
- December 2009
- November 2009
- October 2009
- September 2009
- August 2009
- July 2009
- June 2009
- May 2009
- April 2009
- March 2009
- February 2009
- January 2009
- December 2008
- November 2008
- October 2008
- September 2008
- August 2008
- July 2008
- February 2008
- June 2006
- May 2006
- February 2006
- January 2006
- December 2005
- November 2005
- October 2005
- September 2005
- August 2005
- July 2005
- June 2005
November 11, 2009
Bilski, Business Methods, Software Patents, and the Supreme Court
In late 2008 the U.S. Court of Appeals for the Federal Circuit (CAFC) issued an opinion in the Bilski case holding that inventor Bilski's patent claims, which covered a business method for hedging risk in commodities trading, was not patentable subject matter. Ever since the CAFC's decision in Bilski, I have been asked by clients, other attorneys, and reporters about whether the CAFC's Bilski decision means that software is no longer patentable in the U.S. (For a few articles quoting me on the Bilski case, see these in National Magazine, CBS MarketWatch, and Software Development Times.)
In fact, many people have told me that other attorneys have informed them that software is no longer patentable in the U.S. as a result of the CAFC's Bilski decision. This is not true. I have continued to routinely obtain software patents for my clients since the Bilski decision was rendered. As I explain in more detail below, the CAFC did not hold in Bilski that software is not patentable. The CAFC did not even hold that all business methods are not patentable. Unfortunately, much of the misinformation that has been spreading about the Bilski case seems to stem from second- and third-hand reporting about the case by people who have not read the CAFC's opinion or who are not familiar enough with patent law to understand how to interpret the opinion.
Understanding the CAFC's Bilski decision requires first understanding the claims in Bilski's patent application. Although it is long, it is worthwhile considering claim 1 in Bilski's patent application:
1. A method for managing the consumption risk costs of a commodity sold by a commodity provider at a fixed price comprising the steps of:
(a) initiating a series of transactions between said commodity provider and consumers of said commodity wherein said consumers purchase said commodity at a fixed rate based upon historical averages, said fixed rate corresponding to a risk position of said consumer;
(b) identifying market participants for said commodity having a counter-risk position to said consumers; and
(c) initiating a series of transactions between said commodity provider and said market participants at a second fixed rate such that said series of market participant transactions balances the risk position of said series of consumer transactions.
The first, and arguably most important, feature of this claim is that it merely claims a "method," and does not state that the method is performed using a computer or any other machine. As a result, if this claim had been granted in a patent, such a patent could conceivably be enforced against people who performed Bilski's method, in whole or in part, using their minds, pencil and paper, oral conversations, or otherwise without the use of machine.
The CAFC could have issued a narrow ruling that such "pure" business method claims are not patentable subject matter. Such a decision, whether correct or not, would have had limited impact on the patent system as a whole. It would even have had a limited impact on business method patents whose claims require the use of computer software, hardware, or other machinery to carry out the claimed methods. A narrow ruling that pure business methods are not patentable would almost certainly have had little impact on software patents in general, and the press would not have been riddled with reports that software had been deemed unpatentable in the U.S.
It seems that the CAFC tried to achieve this result by announcing, in its Bilski opinion, that:
• "we decline to adopt a broad exclusion over software or any other such category of subject matter beyond the exclusion of claims drawn to fundamental principles set forth by the Supreme Court"; and
• "the process claim at issue in this appeal is not, in any event, a software claim. Thus, the facts here would be largely unhelpful in illuminating the distinctions between those software claims that are patent-eligible and those that are not."
Such statements should make clear that the CAFC's Bilski opinion should not be interpreted to mean that software cannot be patented. Instead, it seems that the CAFC intended to leave the state of the law with respect to software patents unchanged.
So why all the fuss about Bilski and software patents? The reason is that the CAFC, instead of issuing a narrow opinion about the patentability of pure business methods, decided to use the case as an opportunity to announce a new legal test for determining whether processes are patentable. In particular, the CAFC held that a process only qualifies as patentable subject matter if it: (1) is tied to a particular machine or apparatus; or (2) transforms a particular article into a different state or thing.
Unfortunately, this "machine or transformation" test has raised questions about software patents which the CAFC could have avoided by writing a more carefully-worded opinion. For example, under the "machine or transformation" test, does a patent claim which merely states that a method is "computer-implemented" or "performed by a computer" mean that the method is tied to a "particular machine"? If not, then can adding additional details about the computer's hardware into the claim cause it to satisfy the "tied to a particular machine" test?
As I mentioned above, I have continued to obtain software patents for my clients even in the face of these questions. I have been finding that when a software patent application is well-written, patent examiners either do not raise objections to it under Bilski, or the objections they raise can be overcome using arguments about the meaning of the Bilski decision and possibly some amendments to the wording of the claims, if appropriate.
I have found, however, much variation in how patent examiners interpret the Bilski decision. As a result, I have found it to be more important than ever to talk to examiners in individual cases to understand the precise nature of their concerns so that I can address them, rather than applying a cookie-cutter approach every time I receive a rejection under Bilski.
My experience is consistent with the examples described by the AIPPI in its excellent amicus brief, which documents inconsistencies in decisions issued by the USPTO's Board of Patent Appeals and Interferences (BPAI) over the last year as it has attempted to apply the "machine or transformation" test. For example, the BPAI has not been able to reach internal agreement about whether reciting a "computer system" or "processor" in a patent claim is sufficient to satisfy the "machine" prong of the Bilski test.
Now the case is before the U.S. Supreme Court, which one can only hope will resolve such inconsistencies and issue an opinion which is clearer than the one issued last year by the CAFC. In particular, the Supreme Court should issue a decision which relates to the "pure business method" type of claim for which Bilski sought a patent, and leave the law with respect to claims for software and other computer-related technology unchanged. Although it is impossible to predict how the Supreme Court will rule, the significant interest in the case, including the the large number of amicus briefs which have been filed, leads me to believe that it is unlikely that the Supreme Court's opinion will have any significantly negative direct impact on software patents, particularly in light of the long line of cases from both the Supreme Court and the CAFC which support the patentability of software. The outcome that I consider more likely is one in which the Supreme Court's opinion does not explicitly affect software patents, but instead inadvertently contains language which is subject to interpretation. In such a case, it will be up to the lower courts and the USPTO to hash out the meaning of the decision over time.
The Supreme Court will likely issue its decision in the spring of 2010. In the meantime, I plan to continue filing and obtaining software patents and other computer-related patents for my clients. Neither the spirit nor the letter of the CAFC's Bilski opinion, which is currently the law of the land, rule out patents on software, provided that such patents are written correctly, based on a solid understanding of patent law as it applies to software. Although the confusion within the USPTO about the CAFC's Bilski opinion now sometimes causes some extra care and effort to be required to obtain software patents, including more frequent and detailed conversations with patent examiners to explain the state of the law to them, the death of software patents that has been reported by some in the press is greatly exaggerated. Stay tuned for further developments later this year and once the Supreme Court issues its much-awaited decision.
September 8, 2009
The Case for Software, but not Literary, Patents
Timothy B. Lee recently wrote an essay for the CATO Institute entitled, "The Case against Literary (and Software) Patents," in which he argues that software patents should be prohibited for the same reasons that we do not allow patents to be granted for literary works. I agree that literary patents should not be allowed, and that computer programs share some important features with the novels, poems, and legal briefs that Lee discusses. Yet I think software should be patentable because it also has an important feature which Lee's essay ignores.
Software can guide a missile, control the flow of water in a pipe, and detect tumors in an x-ray. Just try doing that with a copy of War and Peace.
Although both computer source code and Tolstoy's prose express ideas, software has an additional property essential to patentability -- the ability to perform practical functions such as those just described, by controlling computer hardware and/or other machinery. I have argued on this blog in the past (see here and here) that there is no principled basis to impose a per se ban on patents for software that performs these kinds of functions, since non-software-based machines that perform such functions (and which are novel and nonobvious) are entitled to patent protection. I still see no reason for changing this conclusion.
Not all software performs the kind of practical function which patent law is intended to protect, just as not all physical objects perform such functions. Therefore not all software should necessarily qualify as what we patent lawyers call patentable (or statutory) subject matter. For example, although a new song stored in a digital format might satisfy the definition of software" because it constitutes instructions to a computer to play a sequence of sounds, such a song almost certainly does not qualify as patentable subject matter because its function is purely aesthetic, not practical. (I leave for another day an explanation of why it is at least conceivable that such a song could perform a "practical" purpose under the right circumstances, however unlikely such circumstances might be.)
The fact that not all software performs a practical function does not mean that all software does not perform a practical function. Patent law should apply to that class of software which performs a practical function for the same reasons that patent law applies to machines and other human-made products and processes that perform such functions. For example, if traditional thermostats are patentable subject matter because of the function they perform (namely, controlling a furnace), then software which performs the same function should be patentable subject matter for the same reasons.
This is not to downplay the literary qualities of computer programs, or to dismiss the significance of software's dual literary-practical nature (what we in the intellectual property field sometimes call the "expressive" and "functional" aspects of software). In fact, I find this unique combination of features so fascinating and important that I chose to describe it right at the beginning of Chapter 1 of The Genie in the Machine. I have also written about the need to interpret the First Amendment carefully to ensure that distribution of software is not regulated in a way that stifles programmers' freedom of speech.
The problem, however, with Lee's case "Case against Literary (and Software) Patents" is that it makes a logical leap from the premise that literary works and software share something in common (namely, their expressiveness) to the conclusion that intellectual property law should therefore treat literary works and software identically, namely in the way that intellectual property law currently treats literary works. The flaw in this logic is that it does not take into account an additional property of a significant fraction of software -- namely its ability to perform the same kinds of practical functions performed by traditionally-patentable machines and processes -- that renders the argument's conclusion invalid.
August 10, 2009
Turing Tarpits and Nonobviousness
Alan Perlis coined the term "Turing tarpit" in a 1982 article entitled, "Epigrams on Programming," in an epigram which stated, "Beware of the Turing tar-pit in which everything is possible but nothing of interest is easy."
Turing tarpits tell us something about nonobviousness in patent law which may seem trivial, but which is often missed in debates about software patents: the mere fact that computers make the creation of a particular piece of software possible does not render that software obvious. Computers may facilitate the creation of software, and thereby raise the bar of nonobviousness for software, but they don't raise the bar infinitely. Yet it continues to be common to hear the argument that computers render all software trivial to create, and therefore obvious and unpatentable.
I presume that the "Turing tarpit" idea was inspired both by Turing's conception of the computer as a universal machine, capable of mimicking any other computing machine, and also by Turing's response to "Lady Lovelace's Objection" in his paper, "Computing Machinery and Intelligence." Mere possibility does not imply predictability and therefore should not be treated within patent law as sufficient proof of obviousness.
July 8, 2009
Implications of the Bilski Case for Software Patents
Although the facts of the Bilski case that the U.S. Supreme Court will hear this fall only involve a business method patent, the case could have significant implications for software patents and patents on computer-generated inventions, as I explain in this interview on the "Create the Dream" web site.
July 1, 2009
Guest Blogging on IP Watchdog
IPWatchdog.com published a guest blog entry from me yesterday in which I argue that the "wishes" I talk about in The Genie in the Machine should be patentable. This kind of discussion is particularly important to have right now, in light of the U.S. Supreme Court's decision to hear the Bilski case. A flawed decision in that case could set back patent protection for software and for computer-generated inventions for years, if not decades.
June 7, 2009
In re Bilski: Ten Reasons Why
In re Bilski is an en banc decision of the United States Court of Appeals for the Federal Circuit (CAFC) on the patenting of method claims, particularly business methods. The Bilski petition for certiorari has been discussed here in an earlier post. Erick Arner, co-author of the petition, has written an interesting guest post on the Patently-O blog which lists ten reasons why the Supreme Court should hear the Bilski case.
April 25, 2009
Patent System at a Crossroad
At the 11th annual Sughrue Symposium on Intellectual Property Law and Policy, Judge Michel called for net improvement to the patent system. In his keynote address, Judge Michel describes the current situation as unstable and at a crossroad. Citing problems with the Intellectual Property Act of 2009 and problems at the Patent and Trademark Office, Judge Michel asked Symposium participants to get involved in the effort to improve the situation. Read more about the Intellectual Property Act on the Promote the Progress website.
For my own thoughts on how to improve the patent system, see The Genie in the Machine when it is published in a few days. . .
April 2, 2009
Are Quality Software Patents the Answer?
The subject of software patents is polarizing, with businesses making use of them at the same time that many software developers oppose them. Part of the solution may lie in a way to ensure fewer but better-quality software patents. Writing in the U.K. edition of ZDNet, intellectual property consultant Jeremy Phillips looks into the possibility of ensuring more quality patents through heightened patent application examination.
March 5, 2009
National Magazine on Business Method Patents
National Magazine, published by the Canadian Bar Association, has a good article on the implications of the U.S. Court of Appeals for the Federal Circuit's decision in In re Bilski on business method patents in Canada and elsewhere. (Shameless self-promotion: I am quoted at the end of the article as stating that although some are interpreting the Bilski decision to mean that software is not patentable, the decision itself expressly states that it is not intended to address the patentability of software.)
February 25, 2009
The Software Patent Cultural Gulf
At a recent Brookings Institute conference entitled, "The Limits of Abstract Patents in an Intangible Economy," lawyers and technologists met to discuss issues related to patents in the software industry. According to an article on techdirt.com, there was a 'cultural gulf' between the participants, with patent attorneys and law professors supporting the system of patents on software, and techies speaking out against it. For some hypotheses about why this gulf exists, see the article. For my own views on the topic, just wait a while longer until The Genie in the Machine is published.
January 21, 2009
U.S. Ruling on Business Method Patents
In October, the U.S. Court of Appeals for the Federal Circuit (CAFC) ruled that business methods which are not tied to a specific machine or which do not perform a physical transformation cannot be patented. The case was closely watched and has been heralded by some as the death knell for software patents.
Those who interpret Bilski
to render software unpatentable must have read a different opinion than
I did. Does the following sound like a court that has decided that
software is not patentable?
- "[A]lthough invited to do so by several amici, we decline to adopt a broad exclusion over software or any other such category of subject matter beyond the exclusion of claims drawn to fundamental principles set forth by the Supreme Court."
process claim at issue in this appeal is not, in any event, a software
claim. Thus, the facts here would be largely unhelpful in illuminating
the distinctions between those software claims that are patent-eligible
and those that are not."
There is substantial precedent from both the CAFC and the U.S. Supreme Court which is consistent with the conclusion that a wide variety of software patent claims satisfy this "machine or transformation" test. That is not to say that all software patent claims qualify as patentable subject matter or that the Bilski opinion will not be interpreted within the CAFC, in the district courts, or within the U.S. Patent and Trademark Office to modify the standard of patentability for software. Such changes, however, will likely be subtle and play out over time in a wide variety of cases, despite what recent press reports to the contrary would lead you to believe.
January 15, 2009
European Patent Office and Software Patents
The European Patent Office (EPO) has referred the issue of the patentability of software to its Enlarged Board of Appeal. The EPO is looking for guidance on uniform application of the European Patent Convention (EPC), a multilateral treaty which provides an autonomous legal system to govern the granting of European patents. According to EPO President Alison Brimelow, there is currently confusion about the patentability of software and inconsistencies in how EPC guidelines are interpreted in different countries.
Efforts to produce uniformity are to be applauded. Just as important, however, is that the resulting uniform rules are the correct rules. Such a result will not be achieved until the law comes to grips with the genie nature of computers.
June 28, 2006
What is "obvious" is obviously not obvious
One criticism that has long been leveled against software patents is that such patents are often granted on "trivial" inventions, i.e., inventions that are only a trifle different from existing technology. (Look here for one such argument.) The "nonobviousness" requirement of patent law is intended to prevent "trivial" patents from being granted. (For an article on the difficulty applying the nonobviousness standard to software, look here.)
The U.S. Supreme Court has just granted review in a case (KSR v. Teleflex) that will re-examine how the "nonobviousness" standard should be applied in patent cases. Although the case does not directly involve a software patent, a ruling that makes obviousness easier to prove could make "trivial" software patents more difficult to obtain. Stay tuned for further developments.
June 22, 2006
U.S. Supreme Court Declines to Rule on Metabolite Case
The case of LabCorp v. Metabolite had the potential to clarify and either expand or contract the reach of patent law. In layperson's terms, the case involved a patent on a method for detecting a vitamin B deficiency in a person by: (1) analyzing the person's blood to determine whether the person has an elevated level of a certain amino acid; and (2) "correlating" the elevated level with the vitamin B deficiency. The claim was written and interpreted so broadly that it could cover a doctor reading the blood test results and performing the "correlating" step of the process in his or her mind. Imagine a doctor reading the blood test results and thinking, "Aha! These results indicate a vitamin B deficiency."
The case was being closely watched because it raised the question whether a patent could cover a method that is performed at least partially by a human mind. But it had even broader implications for the patentability of medical techniques, software, business methods, and more
I'm referring to the case in the past tense because the U.S. Supreme Court has dismissed the appeal of LabCorp v. Metabolite as being improvidently granted. For non-lawyers, this means that the Court at first decided to review the case, but then changed its mind, deciding that it had been wrong to consider making a decision in the first place. The result is that the patent still stands -- at least for now -- and that the law remains unchanged, and as murky as ever at the edges.
January 16, 2006
European Software Patent Directive Rises From the Ashes?
January 11, 2006
U.S. Patent Office Takes Steps to Improve Software Patent Quality
The U.S. Patent and Trademark Office (USPTO) recently announced that it has partnered with members of the open source community to expand the access that patent examiners have to "prior art" (such as existing software patents and other documentation of existing software) in an effort to improve software patent quality. The partnership also agreed "to develop a system to alert the public when USPTO publishes certain software-related applications so that interested parties can submit related prior art in accordance with relevant rules and law; and, to explore developing additional criteria for measuring the quality of software patents."
January 4, 2006
Software Patent Application Filings on the Rise in India
December 21, 2005
Why the Solution to the Obviousness Problem in Software Patents Isn't Obvious
Techdirt says that there is a simple explanation for why software patents are dangerous. If I understand him correctly, there are two problems.
First, it takes too long for the U.S. Patent Office to figure out whether a patent is valid.
Second, "many programmers faced with similar problems will come up with similar solutions." Granting a patent to one of those programmers gives that programmer the ability to block the others from using the same solution, even if they developed it independently.
It might be true that programmers are more likely than other kinds of scientists and engineers to come up with similar solutions when faced with similar problems. People often make this claim, but I haven't seen any evidence for it. In evaluating whether any particular software patent claim is "obvious," the best that any patent office or court can do is to try to figure out what would have been obvious to an ordinary programmer in the relevant field at the time. This is a hard thing to do in any technological field, not just computer science. And if programmers want to help reduce the number of obvious patents being granted, they can contribute prior art to prior art databases (such as Gauss) or in a patent reexamination. Look here for another interesting proposal along these lines.
If it is true that "many programmers faced with similar problems will come up with similar solutions," the real problem with software patents might not be that the "obviousness" standard doesn't work correctly for software, but that there is no "independent creation" defense for patent infringement, as there is for copyright infringement. In other words, you can infringe a patent even if you independently invent the invention covered by the patent (i.e., without copying the invention covered by the patent). I raise this only to point out how many cans of worms can be opened even by a single seemingly simple observation about the nature of innovation in software.
November 30, 2005
Would the real invention please step forward?
Please excuse me while I blog on some long-overdue topics.
The patent examiner had rejected the claim for being directed solely to a "presentation of information," which is not patentable under Article 52(2)(d) of the European Patent Convention. More specifically, the examiner rejected the claim because the term "message" defined neither a physical entity (such as a product) nor a method. The examiner considered a "message" not to be a physical entity but rather merely a kind of information, "which was non-physical information of an essentially abstract character."
The Board reversed, stating that:
when considering the nature or category of a claimed invention attention must be paid to the substance of what is claimed, rather than only taking into account how the claimed subject-matter is designated, which can be deceptive. In the board's view the content of the information in the message, in the sense used by the examination division, is not claimed. It is commonplace that the same word may be used both for the physical realisation of some information and its content, in the sense of what is understood by its recipient (consider the word "film"). Thus whereas the term "message" may in some contexts refer merely to the information it is intended to convey ("The message of 'Don Quixote' is . . ."), when qualified by the term "electronic" the natural meaning refers to its physical realisation. An electronic message is an electrical, magnetic or electromagnetic signal or collection of signals and moreover clearly the product of an electronic process.
Exactly what I've been saying all along.
In other words, the fact that a software patent claim contains terms such as "message," "data," "record," and other terms that can refer to abstractions, does not mean that the claim necessarily claims an abstraction and therefore is not patentable. A record is a physical thing when it is stored in the memory of a physical computer.
The disconnect between the tangibility of products and the high level of abstraction at which such products are described and claimed in patents will only become larger as the automation of invention continues. One can imagine a patent claim to a (very tangible) toaster that only describes the toaster in terms of an abstract set of requirements that were provided to a genetic algorithm to produce the design for the toaster (or, say, just its heating elements). The fact that the claim is written in "abstract" terms does not mean that the claim doesn't cover something tangible and patentable.
October 27, 2005
Symposium on the Future of Intellectual Property Rights in Software (Part 2)
The symposium on the future of intellectual property rights in software at the Centrum voor Wiskunde en Informatica (CWI) in Amsterdam was a great success. There is now more information available on the symposium web site, including abstracts of the presentations and CVs of the speakers.
What struck me about the symposium was the diversity of the audience, which included, among others, in-house patent attorneys, proprietary software company owners, open source programmers, government officials, law professors, and computer science professors. This kind of interdisciplinary dialogue is exactly what is needed to advance the debate on IP protection for software. My hat goes off to Prof. Paul Klint, Prof. Jan Bergstra, and everyone else one who made this event a reality.
August 23, 2005
One way to limit the scope of software patents
I/P Updates reports on a recent decision by the U.S. Court of Appeals for the Federal Circuit in a software patent case, Harris Corporation v. Ericcson, Inc. Read the summary at I/P Updates if you are a patent lawyer. If you aren't, and at the risk of overgeneralizing, the gist of the opinion is that if you write a software patent that describes a particular algorithm for performing a function, and you write the claims in your patent in a way that attempts to broadly cover any algorithm that performs the same function, you won't succeed. Instead, your claim will be interpreted to cover only the specific algorithm that you described in the patent.
Those who oppose software patents on the grounds that they provide legal protection that is overly broad should pay attention to this decision. There are several ways to protect against the problems caused by overly broad patents. One that is often cited is to ensure that good sources of prior art exist and that the Patent Office and litigants have easy access to those sources. Another, often overlooked outside of the patent bar, is for courts to develop appropriate rules for limiting the scope of claims in individual patents. I advocated for the further development of such rules in software patent cases in this paper (see Section V.C.2), and for the reasons provided therein I think more decisions like the one in the Harris case would be a good thing.
August 9, 2005
UK judge has problem defining "technical problem"
Axel Horns reports on a recent decision by Judge Prescott of the Royal Courts of Justice in London effectively rejecting two patent applications (here and here) on what I think can fairly be characterized as computer-implemented business methods.
Judge Prescott expounded at length on how difficult it is to define the terms "technology" and "technical" -- definitions that are essential if "technical contribution" is to serve as a legal standard for distinguishing patentable from non-patentable subject matter. Judge Prescott rightly observes that "technology" is "a horribly imprecise concept" and that "trying to define the words 'technical' or 'technology' is a dead-end."
Although I agree with his observations, they beg important questions:
- Why are the terms "technology" and "technical" so hard to define?
- Why does the problem of defining these terms arise only in certain kinds of cases -- such as those involving software and business methods -- but not others?
Patent law seems to do a pretty good job of separating patentable from non-patentable subject matter in many areas despite the vagueness of these terms. More generally, the law deals with vague terms such as "reasonable" every day without breaking down.
As you might have guessed, I have an explanation: computers are making it possible to automate processes that previously only human beings engaged in the "liberal arts" (as opposed to the "useful" or "technical" arts) could perform. In the past, machines automated processes in the “useful arts,” such as manufacturing, construction, and transportation. Only human beings, exercising their creativity and judgment, could perform services in the “liberal arts.” Lawyers practiced law, doctors practiced medicine, businesspeople created and executed business plans, and authors wrote prose and poetry.
Intellectual property law reflected and respected this distinction. Copyright law protected works in the “liberal arts,” while patent law protected creative works in the “useful” (technical) arts.
But now computers are automating processes in every field of human endeavor, including those falling within the “liberal arts”: medicine, law, business, and the visual arts to name a few. Although machines have always been used in these fields, software is the first technology to enable end-to-end automation, as software that can recognize and translate speech demonstrates.
It is the widespread end-to-end automation of business methods and other methods previously classified within the "liberal arts" that is forcing the law to confront the ambiguity of terms such as "technology," when previously such ambiguity could be overlooked safely enough, except in extremely rare instances.
Although I don't have an answer to the ultimate question of how far patent law should extend its reach, I do think the story I've just told at least helps to understand both why we are seeing so much controversy over the definition of "technology" and how computers are blurring the line between the technical and non-technical.
July 25, 2005
Hold onto your seats, baseball viewers, or you might infringe a patent
Techdirt reports that a Microsoft patent application is directed to a method for identifying when a segment of a baseball game on television is exciting. Although most of the claims are for software, claim 66 is for a method that doesn't require the use of a computer. One reasonable interpretation of such a claim is that it could be infringed by the mind of a person watching TV at home.
Some qualifiers: the claims include limitations that might make them unlikely to be infringed by a human brain. Also, remember that this is a patent application, not an issued patent. The claims of the application might never be granted, or they might be modified before they are granted.
Patent claims like these raise interesting questions about whether methods that are performed without the use of machines should be patentable. The opinion in State Street Bank v. Signature Financial Group, the case in the U.S. that announced that business methods may be patented, did not make clear whether a business method must be implemented using a machine to be patentable. This has led many to file patent applications for so-called "non-machine implemented business methods," my favorite of which is a "Method of Shared Erotic Experience and Facilities for Same." And, unlike the Microsoft application, this one actually issued as a patent.
Does this mean that "safe sex" now requires an opinion from a patent lawyer?
July 19, 2005
Shift seen from human coding to machine learning in the face of uncertainty
ComputerWeekly.com reports on the use of machine learning, instead of human-written algorithms, by Microsoft Research Cambridge (MRC) to solve problems in domains such as handwriting recognition. The reason, according to Christopher Bishop, MRC's Assistant Director, is that:
It is impossible to write an algorithm for recognising handwriting; many have tried but all have failed. There are too many differences even in the writing of one person to be able to write a set of rules or algorithm.
. . .
With handwriting recognition, for example, we do not try to program a solution; we give the computer lots of examples of handwriting, along with the text, and the machine learns to match the electronic image to the text.
The preceding quotes are taken from Bishop's British Computer Society Lovelace Lecture. A report of and slides from the lecture may be found here.
Much of the legal scholarship on intellectual property protection for software has focused on protection for "algorithms." In fact, many scholarly treatments have used the terms "software" and "algorithms" interchangeably. Developments such as those described by Bishop make clear that the subject matter of computer science is not limited to algorithms. The legal profession needs to wake up to this difference between the law's conception of computer science and the way that computer science works in the real world.
July 15, 2005
Patent infringement by teleportation
Patently-O comments on a lawsuit in which AT&T has accused Microsoft of infringing a patent on speech codecs. Microsoft claimed that it didn't infringe the patent because "Microsoft generates its source code in the U.S. That source code is copied and shipped abroad to Foreign computer manufacturers who . . . generated 2nd generation copies of the software that are then installed and sold."
Although Section 271(f) of the U.S. patent statute attempts to prohibit shipping components of patented inventions overseas to be combined there, Microsoft claimed (among other things) that it wasn't "supplying" components because it was the foreign-made copies, not the software that Microsoft supplied, that were installed and sold overseas. The District Court disagreed, finding that "supplying" software necessarily implies making a copy: "Copying, therefore, is part and parcel of software distribution."
It seems that the same logic would apply to "supplying" physical objects by "virtual teleportation" in the future (see my previous posting). For example, imagine that I have a piston for use in a patented automobile engine. I take a 3D photograph (using my handy-dandy 3D camera) of the piston, and send the digital photo file over the Internet to an overseas manufacturer. The manufacturer uses a 3D printer to produce a physical piston from the CAD file, and incorporates the piston into an automobile engine that it sells overseas. Am I an infringer under 271(f)? If Microsoft "supplied" the software in the case described above, then there's a strong argument to be made that I have "supplied" the piston even though I didn't transport any atoms overseas.
Stay tuned for developments in this area as 3D printing and desktop manufacturing become more widespread. If anyone knows of any cases specifically interpreting 271(f) in this context, please let me know.
July 13, 2005
Microsoft obtains patent on training people to analyze music
News.com reports that Microsoft has obtained a patent on techniques for "training a trainee to analyze media, such as music, in order to recognize and assess the fundamental properties of any piece of media, such as a song or a segment of a song." As News.com summarizes, "that roughly means the company has a recommendation tool for music that is run by real people, and it needs to make sure that people are rating songs the same way."
Rather than comment on the merits of the patent, let me call your attention to the fact that the training process covered by the patent "includes an initial tutorial and a double grooving process."
Double grooving process?
The idea is that music experts are used to rate (classify) songs according to various criteria. The trainee is then asked to rate the same songs, and if he or she is able to match the experts' ratings sufficiently closely, the trainee is deemed "a groover" and is then allowed to rate new songs. Or, as the patent puts it: "When a high enough degree of cross-listening consensus is reached, the new listener becomes a groover and can classify new songs or segments of songs."
I can see the letter that Microsoft will send when it believes someone is infringing the patent: "We hereby demand that you immediately cease all grooving using our patented double grooving process. If you continue to groove, swing, get down, or otherwise funkifize, we shall be required to take legal action against you."
July 12, 2005
Artificial inventions in science fiction
IlliGAL Blogging links to worldchanging's coverage of a science fiction novel in which the protagonist has "patented using genetic algorithms to patent everything they can permutate from an initial description of a problem domain – not just a better mousetrap, but the set of all possible better mousetraps. Roughly a third of his inventions are legal, a third are illegal, and the remainder are legal but will become illegal as soon as the legislatosaurus wakes up, smells the coffee, and panics." I don't know how prominently these inventions figure in the novel, but it looks like an interesting read, and it captures quite nicely an issue that the law needs to address soon.
July 7, 2005
Software Patents Can't Be Banned
News.com reports that the European Parliament has rejected the proposed "software patent directive." The Foundation for a Free Information Infrastructure (FFII), which strongly opposed the Directive, announced that "[t]his is a great victory for those who have campaigned to ensure that European innovation and competitiveness is protected from monopolisation of software functionalities and business methods." For news and views from other blogs on the directive's rejection, see Axel Horns, Information Policy, and A Polytrope's Musings.
The success or failure of the Directive will have little practical effect because software patents cannot be banned. I'm not saying that software patents should not be banned. I'm saying that they cannot be banned, at least not without also banning patents on hardware. And the reason isn't political or economic. It's that there is no justification for granting patents on hardware that does not also justify granting patents on software.
Imagine that the European Parliament adopted a directive tomorrow proclaiming that "software shall be excluded from patentability." The next day, coincidentally, two inventors at opposite ends of Europe independently build two separate machines for clarifying x-rays. As described previously, the two machines are identical in external appearance and they both perform the exact same new and useful process for clarifying x-rays. One box contains circuitry custom-designed by an expert electrical engineer, while the other box contains a computer running software written by a programmer.
The electrical engineer submits a patent application for the circuitry-based x-ray box and obtains a patent on it. The programmer submits a patent application for the software-based x-ray box and is denied a patent, based on the new ban on software patents.
The hypothetical ban on software patents will only drive our hypothetical programmer to write his patent applications differently. Instead of describing his invention as a computer program in the patent application, he will describe the physical implementation of the program in a physical computer, thereby reframing his "software patent" as a "hardware patent" without any dishonesty or inaccuracy. In an extreme case, he could describe (perhaps using a binary listing) the state (on or off) of each and every one of the transistors in the computer's memory when the program is stored in it. Such a description would describe, admittedly inelegantly, a new and useful physical state of a computer that could have been obtained by hardwiring rather than by writing a computer program.
There would be no basis to reject such a patent application under the ban on "software patents." Such a patent application would be just as much a "hardware patent" as the patent application submitted by the electrical engineer.
The effect of the ban on software patents, therefore, would be to drive up the cost of writing and prosecuting patents, and make patents on inventions created using computer programs more difficult to understand. This is exactly the opposite of the effect intended by groups such as FFII.
And such a scenario is scarcely science fiction. Indeed, it is exactly what has happened in the U.S., Europe, and elsewhere beginning at least as early as the 1970s in the face of uncertainty about whether "software patents" would be granted and upheld. The trend towards "hardwarification" of patents on computer programs has declined somewhat in the U.S. as the status of such patents has become clearer, but developments such as those taking place in the E.U. may reverse the trend again, once again increasing the costs of obtaining patents and making them more obscure. I have proposed a solution to this problem for those who are interested.
A software patent puzzle (part 4)
John Koza claims that genetic programming is an automated invention machine. I think this is a reasonable way to think about genetic programming. But it is also accurate and helpful to think of genetic algorithms as tools that assist humans in the process of inventing. If you imagine a continuum of invention automation tools, ranging from a sharp stone on one end to a "genie in a box" on the other end, genetic algorithms are much closer to the genie than the stone. But they still require human input and manipulation to generate inventions.
As a result, genetic algorithms effectively boost the "inventing power" of the people who use them. Something that many people at last week's GECCO and NASA conferences told me was that genetic algorithms are making it possible for people with less and less knowledge and skill in a particular field to produce new inventions in that field. For example, an engineer with a bachelor's degree and a genetic algorithm might be able to match the inventing prowess of an engineer with a Ph.D. without a genetic algorithm. The Ph.D. engineer could, of course, use the same genetic algorithm to boost his own inventive performance. A genetic algorithm on everyone's desktop would, in effect, raise the level of skill of everyone in the field, if "skill" is measured by the ability to produce inventions.
What does this mean for our hypothetical "person having ordinary skill in the art" (PHOSITA)? From the perspective of public policy, I think the right outcome is to increasingly assume that PHOSITA has access to genetic algorithms and the knowledge of how to use them, as such access and knowledge become increasingly common in the real world. PHOSITA's exact degree of access and knowledge will vary from case to case.
The logical consequence of this is that whether a particular invention is nonobvious should be determined by asking whether PHOSITA would have been able to generate the invention by applying ordinary skill to a genetic algorithm, even if the resulting invention would have been surprising or unexpected to PHOSITA. This is consistent with the underlying purpose of the novelty and nonobviousness requirements, namely to issue patents "only for those literally new solutions that are beyond the grasp of the ordinary artisan who had a full understanding of the pertinent prior art." Chisum on Patents, 5.01 (emphasis added). What interest does the public have in granting patents on inventions that could be produced by anyone with ordinary skill and access to a generally-known genetic algorithm? As artificial creativity software continues to effectively increase the level of ordinary skill in various arts, the law should assume an increased level of skill on the part of PHOSITA for purposes of determining whether an invention is nonobvious.
What is interesting is that genetic algorithms and other forms of artificial creativity raise difficult questions about what is meant by phrases such as "beyond the grasp"? Does this mean "beyond the mental grasp" in the sense that PHOSITA would not have conceived of the invention based on his knowledge and skill, or "beyond the practical grasp" in the sense that PHOSITA would not have generated the invention by applying his knowledge and skill to the tools (including genetic algorithsm) at his disposal?
I can think of precedent in the U.S. that supports both of these interpretations. Therefore I am only arguing here for what I believe is the right outcome from a public policy perspective, not what I think the current state of the law actually is. I plan to research the law on this point in more detail and to publish my conclusions at a later date.
A software patent puzzle (part 3)
I had an interesting discussion last week at the NASA Evolvable Hardware Conference with John Koza, Martin Keane, Matthew Streeter, Sameer Al-Sakran, and Lee Jones. We talked about the work they have been doing using genetic algorithms to re-create previously-patented inventions and to generate new patentable inventions. Keane, Koza, and Streeter were granted a U.S. Patent earlier this year on improved proportional, integrative, and derivative (PID) controllers (a kind of circuitry) that were generated using genetic algorithms.
Martin Keane posed a question that is essentially the next piece to parts 1 and 2 of this software patent puzzle. In those parts, I argued that whether an invention satisfies patent law's novelty or utility requirements should not depend on whether the invention is embodied in hardware or software, or on whether the invention was invented with the assistance of a computer.
Martin Keane's hypothetical addressed the third requirement for patentability: nonobviousness (or "inventive step"). If a human and a genetic algorithm independently generate the same invention at the same time, is it possible for the human-generated invention to be nonobvious and the computer-generated invention to be obvious?
My short answer is the same as for novelty and utility: no. Whether an invention is nonobvious shouldn't depend on how the invention was created. So either the human-generated and computer-generated invention are both nonobvious, or they are both obvious.
One reason for this (at least in the U.S.) is that the section of the patent statute that defines nonobviousness explicitly states that "[p]atentability shall not be negatived by the manner in which the invention was made." 35 U.S.C. 103(a).
But even if this sentence weren't in the statute, whether a particular invention (and I'm using the term "invention" here with its colloquial, rather than legal, meaning) is nonobvious at a particular point in time depends on whether the invention "would have been obvious at the time the invention was made to a person having ordinary skill in the art to which said subject matter pertains." Id.. In other words, what is important is whether the invention would have been obvious to the hypothetical "person having ordinary skill in the art" (PHOSITA), not how the invention was actually invented in the real, non-hypothetical, world.
In summary, whether an invention is nonobvious shouldn't vary depending merely on whether the invention was invented using a genetic algorithm or some other kind of computer automation. The tougher question, in my view is: should computer-automated inventing have any impact on how the nonobviousness requirement is interpreted for all inventions, regardless of how they are actually invented in the real world? I will propose an answer to this question in the next entry in this series.
June 28, 2005
Patent Law and Layers of Abstraction
In his talk yesterday (see my previous posting), Drew Endy talked about how some of his work is shifing towards designing biological systems at higher layers of abstraction. The idea of "layers" (or "levels") of abstraction is deeply engrained in the way engineers, and computer scientists in particular, think about systems. I've argued elsewhere that some of the problems with software patents could be fixed if patent law recognized the distinction between different layers of abstraction and focused its attention on the layer(s) where innovation is actually occurring in a particular field.
If you're not familiar with the concept of "layers of abstraction," imagine that you're designing a toaster. You might design the toaster as follows:
- Start with the very high-level (i.e., very abstract) goal of designing a machine to toast bread.
- Then you might identify the functions that the toaster needs to be able to perform, such as holding bread, heating bread, detecting when the bread has been toasted to the desired darkness, and ejecting the bread. Identifying these functions is called "functional design."
- Then you might design the physical components for performing the functions described above, such as heating elements for heating the bread and springs for ejecting the bread. Designing these components is called "physical design" and is often what we think of as "inventing."
Once you've designed and built your toaster, there is of course only a single physical toaster. But you can still think about and describe the toaster in terms of its "physical layer" (its physical components) and its "functional layer" (the functions it performs). Machines and other systems are often designed and described in terms of many such layers in addition to the physical and functional.
It is possible to innovate in any of these layers. Mechanical engineers traditionally have innovated in the physical layer by designing new physical components or new combinations of existing physical components. Computer programmers traditionally have innovated in the functional layer by writing programs consisting of new functions or new combinations of existing functions.
But physical innovation holds a special place in patent law, as the result of assumptions that are no longer valid. One example of patent law's attachment to the physical layer is section 112, paragraph 6 of the U.S. patent statute, which says that patent claims written using functional language are to be limited in scope to the specific physical structure of the invention as described elsewhere in the patent. This section hard-wires the physical layer as the upper limit on the layer of abstraction at which an invention may be claimed.
This legal distinction between the physical and functional layers made sense when the focus of most innovation was the physical layer, and when there were no computers or other means available to automatically produce physical implementations of innovations described at higher levels of abstraction. But these assumptions are no longer valid, and patent law needs to become more flexible in response by focusing protection on the layer(s) of abstraction at which innovation is actually occurring in the real world. In future postings on this site I will explore ways in which patent law could be reformed to achieve this goal. I will also consider potential problems with such reforms.
June 21, 2005
European Parliament Renames the "Software Patent Directive"
ZDNet UK has an article reporting that the European Parliament has proposed changing the name of the so-called "software patent directive" to use the term "computer-aided invention" in place of the old term, "computer-implemented invention." For those of you who can't see what difference this could possibly make, the stated reason for the proposed change is to make it clear that software is not patentable per se, but rather only as part of an innovation that uses software "to aid the performance of the invention."
These and other attempts over the years to find just the right term to define patents on software are, in my view, doomed to fail. One motivation for my "software patent puzzle" (see parts 1 and 2) is to demonstrate that there is no principled way to use the hardware/software distinction as a basis for distinguishing patentable inventions from unpatentable ones.
Instead, I propose that we focus our attention on the patentability of "computer-generated inventions." Examples of computer-generated inventions include software (a computer generates software when you program the computer) and any device whose design is generated by a genetic algorithm or other artificial creativity software.
It is the "computer-generated" feature of software that keeps causing problems for patent law, and that will continue to cause problems for patent law as computers automate the invention of things other than software. I can't justify that claim in a single blog posting, but I tried to make the basic argument here, and will continue to extend the argument in future postings.
Richard Stallman on Software Patents
Richard Stallman of the Free Software Foundation has written (another) article criticizing software patents. Stallman makes several arguments against software patents: (1) patents cover "ideas," and therefore are broader than copyrights and more easily used to stifle innovation than copyrights; (2) software patents are frequently granted on software that is at best a trivial improvement over existing software; and (3) the ability to obtain a software patent without writing any actual software makes it easy for patent trolls (he calls them "patent parasite companies") to extort money from true innovators. Stallman's solution: ban software patents.
Stallman's arguments strike me primarily as arguments against patents in general, although they may have some particular empirical force in the software context.
I wonder what Stallman would think about a patent on a mechanical device that was designed by software? Or a patent on software whose only function is to design hardware? Or a patent on an electrical machine whose only function is to write software? Would such patents be "software patents" and should they be allowed?
I agree with Stallman that software raises some difficult problems for patent law, but for different reasons. And attempting to address these problems by banning "software patents" outright will only cause as many problems as it solves.
June 17, 2005
Inventors Work Hard to be Lazy
I don't think necessity is the mother of invention - invention, in my opinion, arises directly from idleness, possibly also from laziness. To save oneself trouble.
The purpose of many inventions is to make life easier for the inventions' users. New toasters, cars, and lawn mowers make it easier for the people who use them to make toast, travel, and mow the lawn.
But the purpose of many other inventions is to make it easier for inventors to invent. An engineer might write a computer program to simulate new designs for automobile frames, thereby saving the time, money, and effort needed to build and test physical prototypes of the frames. Similarly, engineers have long invented measurement tools, ranging from calipers to electronic calculators, to make it easier to build and test their inventions more accurately.
Inventors invent such new devices to make their lives as inventors easier -- "[t]o save oneself trouble," in Christie's words. The use of such "invention-facilitating inventions" is usually transparent to the end user, who has no way to know whether his new toaster was designed by pure human ingenuity or by an automated computer program.
Invention-facilitating inventions have long been patentable. But such inventions arguably "promote the progress of useful arts" (the ultimate purpose of U.S. patent law) only indirectly, by reducing the resources (e.g., time, money, raw materials) required to invent. If this is a sufficient basis for patentability, then why not allow improvements in pure mathematics, or at least improvements in "pure software" that performs calculations more efficiently? Surely such improvements may be applied to facilitate the process of inventing.
Although I won't attempt to provide any answers to these questions here, I think that the debate over software patents is in part a debate over how direct the connection needs to be between the function performed by a computer program and some real-world ("practical" or "industrial") use for patent protection to be justified. Future improvements in automated inventing will only make resolution of this question more pressing.
June 16, 2005
A software patent puzzle (part 2)
Glen Secor posted a thoughtful comment in response to part 1 of this entry. Now, as a result of my posting and Glen's additions, we have four x-ray clarification boxes that have the exact same externally-observable behavior, but that differ in the following ways:
(1) the first is powered by human-designed circuitry;
(2) the second is powered by a laptop running software written by a human programmer;
(3) the third is powered by the same software as the second, except that this software was "written" by other software (e.g., a genetic algorithm) rather than a human; and
(4) the fourth is powered by x-ray clarification gnomes.
Again, I'll ask: is there any basis for granting patent protection to some of these boxes but not all of them? Patent law protects (1) products and processes that are (2) new and (3) useful.
(1) All of the boxes are products (with the possible exception of the gnomes, who I'll ignore for now).
(2) All of the boxes are new (based on the way I've set up the hypothetical).
(3) Are the boxes useful? (Do they satisfy patent law's "utility" requirement?) If clarifying x-rays is "useful," then all of the boxes are useful. If clarifying x-rays is not "useful," then none of the boxes is useful.
This hypothetical appears to demonstrate that whether a device is a new and useful machine for purposes of patent law can't depend on either: (1) what's inside the device (e.g., hardware or software); or (2) how the device was designed (e.g., by a human or by software). Therefore, if there's a reason for objecting to software patents and not to hardware patents, or for objecting to patents on computer-designed machines and not to patents on human-designed machines, it must lie elsewhere.
In other words, what is the controversy about software patents really about?
June 13, 2005
What's in a name? Computer "science" vs. "engineering"
Illigal Blogging has a nice post criticizing the initial choice to use the word "science" (as in "computer science") to describe what computer people do. (I'll use the term "computer people" in this post to avoid any bias toward "science" or "engineering.") I agree with the author's assertion that the use of "science" instead of "engineering" is unfortunate because it fails to capture the significant ways in which computer scientists/engineers use computers to develop new solutions to real-world problems.
There has been similar debate about the use of the term "machinery" in the name of the computer profession's preeminent association: The Association for Computing Machinery. R.W. Hamming, in his acceptance speech for the 1968 Turing Award (ACM Digital Library subscription required), said:
At the heart of computer science lies a technological device, the computing machine. Without the machine almost all of what we do would become idle speculation, hardly different from that of the notorious Scholastics of the Middle Ages. The founders of the ACM clearly recognized that most of what we did, or were going to do, rested on this technological device, and they deliberately included the word “machinery” in the title [of the ACM]. There are those who would like to eliminate the word, in a sense to symbolically free the field from reality, but so far these efforts have failed. I do not regret the initial choice. I still believe that it is important for us to recognize that the computer, the information processing machine, is the foundation of our field.
The focus on the "scientific" and theoretical aspects of what computer people do has affected how the law has viewed computer software. For example, an argument that continues to be raised against the patenting of software to this day (and particularly strongly in Europe at the moment) is that software is "abstract" or "intangible" and therefore lacking in the "practical" or "technical" nature required for patent protection. I think that this dispute about whether software is "abstract" (and hence not patentable) or "technical" (and hence susceptible of patent protection) has some of the same roots as the debate about whether computer people are doing science, mathematics, engineering, some combination of them, or something completely different.
Although what many computer people do what is reasonably classified as "science," many are using computers to design new solutions to practical problems -- in other words, to engage in engineering. The debate over legal protection for software needs to incorporate a more nuanced understanding of what computer people do if there is to be any hope of a rational resolution to that debate.
A software patent puzzle (part 1)
The controversy over software patents just won't die, with the row over the proposed European Directive on the Patentability of Computer-Implemented Inventions being the latest bit of evidence.
I think this debate is fundamentally misguided, and that the debate may be more fruitfully framed as one about the patentability of computer-generated inventions, not just software. To shed light on this alternative perspective, consider the following puzzle:
You are shown two black boxes that are identical in external appearance and behavior. Each box has an input slot into which an original X-ray print may be fed. After a short delay, each box produces a highly clarified X-ray print in which any tumors are highlighted. The clarified X-rays produced by both boxes are indistinguishable from each other. Assume that the quality of X-ray clarification produced by both boxes is better than that which may be obtained using any preexisting X-ray processing device.
Upon opening both boxes and peering inside, you find in the first box a complex jumble of circuitry and are informed that such circuitry was custom-designed by an expert electrical engineer. In the other box you find a small laptop computer running X-ray image processing software written by a computer programmer. The circuitry in the first box and the software in the second box implement precisely the same X-ray clarification algorithm.
Question: Is there any basis for deeming the circuit-implemented X-ray clarification device to be patentable subject matter, but not the software-implemented clarification device? Post your answers below.
(This hypothetical first appeared in an article that I wrote, but which is not yet available online. All the better -- now you can't cheat!)