In Defense of Software (Patentability)

Note from Matt LevyThe Alice v. CLS Bank case, which is being argued at the U.S. Supreme Court later this month, has invigorated the debate over the patentability of software. 

The following guest post is by Dana Rao, Vice President for Intellectual Property and Litigation at Adobe, and I’m thrilled to be able to publish it. Dana gets at the fundamental conflict in the debate in a way that’s very easy to understand. 

It’s a little longer than our typical pieces, but I hope you’ll read the whole post. It’s worth your time.

A few weeks ago I was asked to speak at my daughter’s 8th grade honors science class, regarding the weighty topics of “patents and innovation.” They are in the middle of a unit on inventions and they have to invent something as part of that class.  As I prepared my slides on interesting topics (to me) like “how Morse used the existing ideas of an electromagnet to solve the problems of instantaneous communication as illustrated in the beacon scene of the Lord of the Rings,” the thought occurred to me that the principles from these legendary inventions illustrate why the software patentability debate of today is disconnected with our country’s history of innovation.

Samuel Morse did not discover the abstract idea that running an electric current through a wire wrapped around iron transforms the previously nonmagnetic iron into a very powerful electromagnet.  Others had discovered this concept (though it was still a recent discovery in 1852).  Morse combined the principle of electromagnetism with his own, intangible, idea, that you can use a code to deliver messages that would reduce the number of wires required to transmit the message, and print the result on paper.  Other solutions required many wires, or used moving needles in scopes at the end to reveal the message.  Morse’s invention increased the practicality of using electromagnetism to solve the problems of instantaneous communication.

He first had an idea, the use of electromagnetism in a particular way, and then had an implementation, one that fit the medium of his idea, in this case, wires and metal and ink. Samuel Morse’s code was an intangible concept indeed, but was deemed patentable even in 1852, as implemented in his telegraph machine, for telegraphic purposes.

The main thrust of the argument against the patentability of software is that a software program, like an algorithm, describes nothing more than an “abstract idea.”  This argument has as its corollary the notion that patentable subject matter only can be found in an invention in something “tangible.”   Patents for intangible things (like an algorithm implemented in software) are conflated with patents on purely abstract ideas (like “time travel” or “electromagnetism”), and therefore, neither category should be patentable.  In reality, though, all ideas and inventions are at their very core intangible concepts.  It is when the inventor actually implements an idea in the medium best suited for its execution that the invention enters the realm of patentable subject matter.  Software is no different than any other medium, and inventions made in software are as deserving of patent protection as inventions anywhere else.

Every patent has a similar story as Morse’s.  In today’s world, the medium in which many of our inventions manifest themselves is software.  These software-implemented inventions typically are executed by a general purpose processor, coupled to a general purpose memory, often coupled to a network.  Instead of being implemented in software, these inventions could be constructed using specialized circuitry, or directly in silicon, or by some specific network topology to accomplish their means.  However, there would be nothing patentable about those implementation mediums (circuitry, silicon, or network topology) in the fulfillment of the inventions.  Nor was there anything patentable about the mediums of Morse’s iron, wire, and ink, or Thomas Edison’s carbon, electricity, and glass.  Instead, it was the ideas animating the use of those mediums to solve their problems (the “algorithms” as it were), that made those discoveries patentable.

If all inventions are intangible at their core, when should an idea be considered to be unpatentably abstract?  In the seminal case of O’Reilly v. Morse, the Supreme Court found Morse’s eighth patent claim, for the use of electro-magnetism however developed for printing letters at any distances, to be drawn to an abstract idea, and thus unpatentable.  Morse, for that one claim, explicitly stated his invention was not bound by the details of his specification.  In finding this claim unpatentable, the Court articulated the basis on which to determine whether an abstract idea is patentable subject matter:

Whoever discovers that a certain useful result will be produced, in any art, machine, manufacture, or composition of matter, by the use of certain means, is entitled to a patent for it; provided he specifies the means he uses in a manner so full and exact, that any one skilled in the science to which it appertains, can, by using the means he specifies, without any addition to, or subtraction from them, produce precisely the result he describes. And if this cannot be done by the means he describes, the patent is void. And if it can be done, then the patent confers on him the exclusive right to use the means he specifies to produce the result or effect he describes, and nothing more.

The Court stated that, to determine whether an idea is abstract, the focus should be on whether the invention is disclosed and claimed fully and exactly.  An inventor can secure a patent in any art on any useful result, the Court says, as long as the exact means for accomplishing that result is specified.  The nature of the invention is irrelevant.

And it makes no difference, in this respect, whether the effect is produced by chemical agency or combination; or by the application of discoveries or principles in natural philosophy known or unknown before his invention; or by machinery acting altogether upon mechanical principles. In either case he must describe the manner and process as above mentioned, and the end it accomplishes. And any one may lawfully accomplish the same end without infringing the patent, if he uses means substantially different from those described.

The key question to understanding whether an idea is abstract, the Court says, is whether or not the inventor described the process in sufficient detail to allow others to achieve the same ends without infringing the patent if they use substantially different means.

We see so often under current law that the “claims are not bound to the specification.”  But why shouldn’t they be?  Samuel Morse’s code was found to be patentable because, in his other claims which lacked the explicit disclaimer of linkage to his specification, the Court limited their breadth to his disclosure. This makes sense. There is no sound policy reason to grant monopolies to people for implementations beyond the scope of their invention.  Unfortunately, patents today are allowed to have claim scopes defined by their relation to the prior art (or, more accurately, the prior art found at the time of examination), and are not grounded by any of the details in their specification.  Breaking the connection between claims and the specification has resulted in the rise of vaguely drafted patents that are the source of the patent troll suits. With no meaningful constraints imposed by the description of the invention itself, the patent trolls are free to make their abstract claims apply to all.

In response, many blame software itself as the cause of these meritless suits, and invoke Section 101 to simplistically assert that software related inventions are just too abstract to be patented.  But from both a technical and a legal perspective, this argument of convenience is plainly not correct.  Software is merely the medium by which the inventions of today are implemented.  If an invention, in software or in any other field, is for a “useful result” and specifies the means described in sufficient detail so full and exact that anyone skilled in the science can produce the same result, the invention should be considered to be more than an abstract idea.  If such disclosure is lacking, then the claim is invalid.  It is either for an idea that is too abstract, or described too abstractly.

If we bring our Section 112 jurisprudence in line with the doctrine articulated by the Morse Court, I believe we can rid our system of abstractly drafted patents. And stricter application of Section 103 will further curb the grant of software patents whose sole claim to novelty is that they implement ancient concepts in the new medium of software.  Both of these measures will appropriately shift the patentability focus from the medium to the method disclosed.

The teenagers in that 8th grade science class told me all sorts of cool ideas (which I can’t disclose), and most typically the medium of their inventions was software.  Let’s ensure the patent system continues to protect the inventors of the next generation, in whatever medium they choose to invent.

  • Owen

    The author is gyrating between two arguments and trying to answer two different questions. The first question is about patent law and precedent and the second is about what patents are for and when they should exist.

    The arguments about abstraction and Morse are arguments about what the statute and common law precedent and the Constitution say. The statute says new machines are patentable and centuries of common law say abstractions are not patentable.

    The author makes the usual patent maximalist contention that every computer program is a new machine and not abstract. The maximalists of the patent bar contend that there is no such thing as a general purpose computer because every computer runs some kind of software. Computer programmers point out that general purpose computers are a real and easily observable fact of reality, that you can buy them and probably have some already. Writing and running programs is what users do with general purpose computers. The maximalists on the CAFC are abusing language and reality when they contend that running a program on a general purpose computer literally makes it a new machine, but that is the argument they seriously use to justify software patents.

    There’s a paragraph where the author trots out the usual argument that software and hardware are the same thing because you could build a specialized circuit. The patent bar seldom tries actually claiming those in patents, though, because those patentees wouldn’t be able to monopolize original software written by unsuspecting programmers.

    But that’s not the only argument being made here. The author weaves in another argument about the practicality and morality of software patents. The first argument says, sure software patents are a baleful scam to legally rob working programmers of their rights and profits, but the law says we can ream you so we are sure going to continue doing so. The second argument says that it would be a fine thing to subject programmers to the patent system because it continues a tradition of innovation.

    The author never outright contends that software patents promote innovation, that software companies invest in useful research and experimentation for the primary purpose of making money on patents, that the industry depends on patents to protect valuable innovations, that programmers will benefit from patents, or that programmers want software patents to be available at all. Nobody who knows anything about this debate would say those things because you can’t contradict the known facts so grossly and be taken seriously. He even briefly agrees that the so far utterly toothless §103 and §112 prohibitions against obvious and vague patents — which the courts and patent office ignore and dismiss — could be strengthened, without saying how.

    He does, though, insinuate that the kids’ inventions might benefit from software patents, that Sam Morse would be retroactively validated by software patents, and that we connect with a grand tradition by subjecting our profession to the whims of exacting legalistic micromanagement of permission to write programs under the patent laws. Those are weak suggestions of a weak argument. There is a reason that programmers oppose patents in software by as much as ten to one. Try to find even a handful of people in a field where patents work who want them abolished — maybe ask some biochemists.

    The reality is that if the patent regime suggested by the author is allowed to continue, those kids will be forever legally blocked from making and using their inventions by government privileges granted to others. The same will apply to all programs we write for our computers.

    We don’t want them and we should look forward to Congress and the courts telling them to leave us alone.

  • Musical Missionary

    As problematic as I feel many software patents are, especially in the hands of trolls, I have to agree with the author. 101 is not the right statutory provision to hold against software patents. While I see the abstract idea argument, it is weakened by the ultimate reality that software is useless without a physical device. So taken as a whole, properly disclosed and claimed, software inventions are not all that abstract. What most of them are, however, is obvious. The right way to knock out highly vague and seemingly abstract software patents is through Section 103 (or 112).

  • Brian Kahin

    While fixing functional claiming would improve matters by making claims narrow and less abstract, the result might simply be to encourage a lot more patents filled with more detail as attorneys try to make up in volume for what they can no longer achieve in scope. Software is designed with many levels of abstraction and at the lowest level, the implementation is code.

    Is code in the specification the ultimate answer? The number of software patents doubled from 38761 to 75368 in the last four years (following Bessen’s classification – the PTO/GAO classification, which covers more industrial classifications, would show roughly twice as many). Perhaps Adobe’s patent department can evaluate that many patents per year or perhaps Adobe can bear the risks of not looking too closely, but small developers cannot.

    Limiting patents to implementations at lower levels would increase the administrative burden and expand the firehose already spewing large numbers of patents of dubious validity. If code-level implementation is required, patent ends up looking a lot like copyright – except that independent creation is not allowed since infringement need not involve copying.

    This is much less of problem for industrial software where the application is limited to an environment where entry is naturally limited. In the case of abstract or “pure” software, there are millions of innovators manipulating multipurpose code. Benson warned against this problem – that the algorithm at issue could be used in many different contexts, but the author apparently rejects Benson.

    The author simply avoids addressing the uncertainty and risk
    software patents create because the sheer volume of software generated is
    matched by the sheer volume of patents and the inter. Twenty years ago, Adobe testified strongly against software patents – without distinguishing
    industrial from abstract – and the company spoke not through a lawyer but
    through its principal scientist, Doug Brotz. In fact, nearly all software companies in BSA, including IBM and Microsoft, originally expressed opposition to all software patents.

    What happened? As large software companies began to build their own portfolios (they had no choice), they saw that a large portfolio served as a thicket that discouraged market entry. Newcomers on the other hand – Google, Facebook, Twitter — have had to buy portfolios as the price of entry.
    We now see that even questionable patents have value, because they can
    be sold to trolls (or used to settle with trolls), who can use then them to
    attack others that are too small to fight back or others who will pay a small
    license fee simply because it makes business sense.

    BSA’s opposition to expansion of covered business method review (which
    Senator Schumer has roundly castigated — see Matt Levy’s post on Patent Progress for Dec 17) is just a matter of business. Why should its members put their accumulated patents at risk, so that those challenging bad patents can have a low-cost alternative to litigation? It would be publicly embarrassing if the challenges were successful.