September 19, 2017

The Supreme Court and Software Patents

I’m very excited that Doug Lichtman, a sharp law professor at UCLA, has decided to take up podcasting. His podcast, Intellectual Property Colloquium, features monthly, in-depth discussions of copyright and patent law. The first installment (mp3) featured a lively discussion between Lichtman and EFF’s inimitable Fred von Lohmann about the Cablevision decision and its implications for copyright law. November’s episode focused on In Re Bilski, the widely-discussed decision by the United States Court of Appeals for the Federal Circuit limiting patents on abstract concepts like software and business methods. The podcast featured two law professors, John Duffy (who argued the Bilski case before the Federal Circuit) and Rob Merges.

As I noted at the time it was decided, people care about Bilski largely because of what it says about legality of software patents. Software patents are intensely controversial, with many geeks arguing that the software industry would be better off without them. What I found striking about the conversation was that both guests (and perhaps the host, although he didn’t tip his hand as much) took it as self-evident that there needed to be patents on software and business methods. As one of the guests (I couldn’t tell if it was Merges and Duffy, but they seemed to largely agree) said around minute 47:

The easiest criticism of the [Bilski] opinion is that it invites this kind of somewhat pointless metaphysical investigation. What you say is “look, I’ve got an invention, I wrote some code, I’d like to a patent for that.” Why do we have to play this kind of sophomoric philosophical game of “well, what changes in the real world when my code runs?” The [Supreme Court] case law arose fairly early in the information technology revolution. We’re kind of stuck with this artifactual, residual overhang of physicality. It’s just the price we have to pay to get a software patent these days. Someday maybe it will drop away or wither away, but that’s where we find ourselves now.

On this view, the Supreme Court’s historical hostility toward patents on software is merely an historical accident—a “residual overhang” that we’d do well to get beyond. Guided by a strong policy preference for the patentability of software and business methods, Duffy and Merges seem to feel that the Federal Circuit should give little weight to Supreme Court decisions that they regard as out of touch with the modern realities of the software industry. After all, this is “just the price we have to pay to get a software patent these days.”

I don’t agree with this perspective. I’ve long sympathized with software patent critics such as Ben Klemens who argue that the Supreme Court’s precedents place clear limits on the patenting of software. But I thought it would be interesting to take a closer look at the Supreme Court’s classic decisions and talk to some patent scholars to see if I can understand why there are such divergent opinions about the Supreme Court’s jurisprudence. The result is a new feature article for Ars Technica, where I review the Supreme Court’s classic trilogy of software patent cases and ponder how those cases should be applied to the modern world.

Like most Supreme Court decisions, these three opinions are not the clearest in the world. The justices, like most of the legal profession, seem slightly confused about the relationships among mathematical algorithms, software, and computer programs. It’s certainly possible to find phrases in these cases that support either side of the software patent debate. However, a clear theme emerges from all three cases: mathematics is ineligible for patent protection, and software algorithms are mathematics. The high court struggled with what to do in cases where software is one part of an otherwise-patentable machine. But it’s hard to avoid the conclusion that many of the “pure” software patents that have generated so much controversy in recent years cannot be reconciled with the Supreme Court’s precedents. For example, it’s hard to read those precedents in a way that would allow Amazon’s famous “one-click” patent.

I also argue that this result is a good one from a public policy perspective. Software has several important properties that make it fundamentally different than the other categories of now-patentable subject matter. As Klemens points out, almost every significant firm has an IT department that creates software. That means that every significant firm is a potential target for software patent lawsuits. This is a very different situation than, say, pharmaceutical patents which only affect a tiny fraction of the American economy. Second, software is already eligible for copyright protection, rendering software patents largely redundant. Most important, we now have 15 years of practical experience with software patents, and the empirical results have not been encouraging. I don’t think it’s a coincidence that the explosion of patent litigation over the last fifteen years has been concentrated in the software industry.

As the Federal Circuit struggles to craft new rules for patent-eligible software patents, it should take a close look at the far more restrictive rules for patent eligibility that were applied in the 1970s and early 1980s.

Comments

  1. If software patents didn’t require physical implementation, automated-discovery and hypercompiler programs would be a great way to build up a portfolio of patents very rapidly. Especially with the (still) ridiculous paucity of searchable prior art for software.

  2. Michael Johnson says:

    The biggest fault I find in software patents isn’t the mathematical nature of them (although that is a big one). It’s obviousness.

    If I understand correctly, something isn’t supposed to be patentable if it’s obvious to an expert in the field. Well, I’m an expert in the field (writing software professionally since 1990), and I haven’t seen a software patent yet which I couldn’t have come up with or implement given the need. You used the “one-click” patent. It extends directly from

    Using cookies to store state on the client.
    Finding user information on the server given that state.
    Wanting to make it as easy as possible for a user to purchase an item.

    If a user’s browser identifies the user to the server via a cookie and the server recognizes the user, it is a very small leap to making a purchase into a single button click. The number of basic implementations is really very small. Endless variations on the specifics (hence copyrightability) but only one or two basic paths of logic.

    • I haven’t seen a software patent yet which I couldn’t have come up with or implement given the need.

      Consider Dynamic Circular Work-Stealing Deque. It’s software, I’m an author, it’s patented. I think this is an existence proof for “not obvious”. It helps that it’s a lockless concurrent algorithm — even the starting point for this work (the original Arora-Blumofe-Plaxton work-stealing deque) is not “obvious”, and we didn’t make it simpler.

      I also recall first seeing a presentation of Joel Bartlett’s conservative-compacting garbage collection algorithm, and about falling out of my chair from the cleverness of it.

      • Yes it is obvious…

        • The point of the above comment is not really to claim that the Dynamic work stealing deque is obvious – just to point out that “obvious” (and be extension “not obvious”) is a very difficult to define criterion.

          It derives from the following story:

          A famous Professor of pure mathematics was lecturing an advanced class. He was going through a proof. He said “it is obvious that ****”” ” a student hand was raised “excuse me sir – is it obvious?” The professot thought for a moment – then left the room. The students dutifully waited and ten minutes later he returned. “Yes it is obvious” he said….

  3. ‘Second, software is already eligible for copyright protection, rendering software patents largely redundant.’
    You state this as if it were self-evident, but copyright and software patents provide completely different types of protection. Copyright is used to keep someone from copying my program without permission, but software patents protect the innovative concept. Not all software development qualifies for a software patent, but if there is true innovation, why shouldn’t it qualify for patent protection?

    • “if there is true innovation, why shouldn’t it qualify for patent protection?”

      Well – patent offices have been setting that bar rather low in recent years – and why should having i=vented something five minutes earlier (or got your forms to the patent office five minutes earlier) give you total control over someone who independently invents the same thing.?

    • In fact, copyright and patent offer what were originally intended to be mutually exclusive forms of protection. Copyright protects expressions of ideas but not functional objects. Patent protects reductions to practice of innovative ideas, but not (until business method patents arose) the innovative ideas themselves. And any number of copyright cases have been won by showing that certain identical chunks of program text were the only sensible way to express a particular idea of what the code should do.

      So maybe redundancy isn’t quite the right word, or maybe it’s used in the wrong sense. When someone copyrights some code and then applies for a patent on the underlying ideas, they’re effectively taking out insurance: if the patent is granted, the copyright is quite possibly invalid because the code is a functional object; but if the patent isn’t granted, the copyright still interferes with other people implementing the same thing.

    • Copyright applies to speech. You cannot restrict speech with patents. This is a violation of fundamental rights.

    • Lawrence D'Oliveiro says:

      Abattoir: “…copyright and software patents provide completely different types of protection.”

      So what? Why should that entitle software to be the only field of human endeavour having both sorts of “protection”? Other fields can make do with one; what’s so special about software?

    • Lawrence D'Oliveiro says:

      Abattoir: “…copyright and software patents provide completely different types of protection.”

      So what? Why should that entitle software to be the only field of human endeavour having both sorts of “protection”? Other fields can make do with one; what’s so special about software?

  4. The elephant in the room here is the fact that patents give the holder rights over people who independently re-invent things.

    To my mind this is just plain immoral (a form of extortion) and shouldn’t be in the law at all for any type of invention.

    When the total patent portfolio is so large that no individual could ever be familiar will all of it and patents are granted easily for things which are obvious then the corollorary is that no individual or small company can play in this game at all without taking a huge legal risk.

    Limiting patents to “physical” things does help – but mainly because it greatly reduces the number of patents – it doesn’t really solve the problem and it creates a huge philosophical argument about what exactly is and isn’t a physical process.

    Having said that I think there are two cases where patent protection does(at least in theory) provide a public good.

    1. It gives a small company or individual some rights when they have to deal with a larger organisation to get their invention implemented.

    2. It protects inventions which have required a large ampunt of expensive experimental work (eg drug testing), which could never be economically justified without some protection.

    However in case 1 the current legal system often backfires anyway and favours the big company and in the second case the patent should clearly be linked to the development expenditure – rather than to the nature of the invention.

    The solution:

    Abolish patents com[pletely and replace them with:

    1. A standard “non-disclosure right” that would automatically apply like copyright does. This would protect inventors from being scooped by anyone who they could show that they had communicated their information to. but would not affect third parties. It would be broader than copyright – but should not last very long (Perhaps five years from the date of disclosure)

    2. A new type of patent based on demonstrated development and test expenditure rather than on the nature of the idea

    • The idea was that a LIMITED monopoly was granted in return for full disclosure. Thus, it would advance the state of the art for everybody because after the time expires it becomes a part of the public domain and the patent disclosure is published for all to learn from.

      As with many government plans, it was derailed by vested interests to such an extent that no one even recognizes what the original design really was.

      A new type of patent based on demonstrated development and test expenditure rather than on the nature of the idea

      Who cares how much money was spent? How strange that we have moved from a free market to an expectation that spending money creates a sense of deserving return. I guess this shows that the purpose of the entire economic system has been lost so what we do with patents doesn’t matter anymore.

      • “Who cares how much money was spent? How strange that we have moved from a free market to an expectation that spending money creates a sense of deserving return.”

        Obviously the person spending the money cares! You won’t spend money on a worthless idea – and of course you would have to demonstrate that spending the money was necessary to develop or prove the idea.

        Of course if you have to put a large investment into an idea then you should expect some protection on the returns – otherwise no-one would spend the money.

        Contrast this to the present situation where companies spend huge sums building enormous portfolios of trivial patents (that should never have been granted).

        The main problem with the present system is simple – too many patents. This is what undermines the value of disclosure. Making patents expensive to get would at least fix that one.

  5. Jess Austin says:

    ‘Guided by a strong policy preference for the patentability of software and business methods, Duffy and Merges seem to feel that the Federal Circuit should give little weight to Supreme Court decisions that they regard as out of touch with the modern realities of the software industry. After all, this is “just the price we have to pay to get a software patent these days.” ‘

    I’m shocked, SHOCKED that lawyers who argue patent suits think the creation of patent suits is a more important goal than the fostering of innovation.

  6. Andrew Chin says:

    I’ve responded to Klemens’s “software is mathematics” argument in On Abstraction and Equivalence in Software Patent Doctrine: A Reply to Bessen, Meurer and Klemens, 16 J. INTELL. PROP. L. ___ (2009). You can email me for the current draft.