Post-Bilski at the USPTO: Is the BPAI Looking to Create a "Software Per Se" Exception?
"The United States is a nation of laws: badly written and randomly enforced."Ex Parte Godwin, Appeal 2008-0130, November 13, 2008
- Frank Zappa (1940 - 1993)
This opinion issued from a Request for Rehearing that was filed before the Bilski decision, and the opinion followed after (side note: the original Notice of Appeal was filed June 6, 2006). It is not clear if this is one of those "bad" cases, but there are a number of troubling positions taken by the BPAI.
One of the claims rejected under 35 U.S.C. 101 follows:
The BPAI, acknowledging Bilski, stated that the claim did not recite patentable subject matter, as it was "directed to purely software components."7. A portal server system comprising:
a portal coupled to a plurality of portlets, each of said portlets having associated portlet rendering logic;
a portlet aggregator communicatively linked to said portlet rendering logic; and,
a visual service extension to said portlet aggregator programmed to process said portlet rendering logic to transform visual style attributes in said portlet rendering logic into markup language tags which can be rendered for display in a specified type of pervasive agent.
What is, well, weird about the decision is the BPAI's reasoning for upholding the 101 rejection. Apparently, the BPAI believes that, given the broadest reasonable interpretation, a claim may be rejected if non-statutory matter can be injected into the claim. Additionally, one of the issues considered by the Board is "[h]ave Appellants shown that the claimed invention recited in claims 7 and 12 is not directed to software per se? " (Freeman-Walter-Abele anyone ?)
From the opinion:
While Appellants’ position is that the "broadest reasonable interpretation of the terms ‘portal server’ and ‘portal server system’ yields a claim construction that includes a hardware device which is statutory subject matter," (Request 8), we find paragraph [0036] of the Specification contravenes appellants’ contention ("The present invention can be realized in hardware, software, or a combination of hardware and software."). A hardware device (i.e., computer) is not positively recited . . . Claims 7 and 12 do not sufficiently invoke a machine or a transformation of subject matter, and are therefore unpatentable under 35 U.S.C. § 101.And get ready for more weirdness. The Appellants submitted that the proper question to be asked is whether the claimed subject matter encompasses statutory subject matter, and not whether the claimed subject matter includes non-statutory subject matter. Here is how the BPAI responded, in its entirety:
[A]s per our discussion above, in the context of a software embodiment, we maintain that a server is broadly but reasonably interpreted as a provider of services. Therefore, we are not persuaded by Appellants’ argument that the scope of independent claims 7 and 12 requires hardware (i.e., a machine) in addition to the recited software components. We also note that software components can be communicatively linked in the abstract sense.
[W]e note that our reviewing court has clearly stated that "[t]he four categories [of § 101] together describe the exclusive reach of patentable subject matter. If a claim covers material not found in any of the four statutory categories, that claim falls outside the plainly expressed scope of § 101 even if the subject matter is otherwise new and useful." In re Nuijten, 500 F.3d 1346, 1354 (Fed. Cir. 2007).
Read the decision on Request for Rehearing here.
Read the original BPAI decision here.
Posts
The BPAI is a bunch of Nazis!
(sorry, I just had to invoke Godwin's Law on Ex Parte Godwin).
You do realize that the claim you quoted would cover:
Slashdot, Google, nearly every publication website (NYTimes, Washington Post, etc.), Yahoo!, Facebook, and nearly every website that collects data to display to their users, no?
So, the decision to reject is good. It would do far more harm to the public than good.
Setting aside the merits of this particular application and its claims, I think Peter's point was that BPAI's reasoning seems odd. They seem to be saying that a claim (any claim) becomes non-statutory and thus unpatentable if its disclosed subject matter COULD be implemented in software.
The BPAI should read about Turing machines, a core topic in any computer science curriculum. Turing machines are archetypal instruction-executing machines that can in theory be made to do pretty much anything.
So... if computer science says that software can implement the functions of pretty much any physical mechanism, and if BPAI is saying that any mechanism that might be implementable as software is unpatentable, wouldn't most utility patents become non-statutory?
A turing machine can make everything that software alone can make: However it can't make a better mousetrap. Or any other physical invention!
I note the reliance the last two commentators put on the adjective "physical". I think I know roughly what it means but I'm not sure when it comes to a standard computer running a new data processing program. Is it anything like the EPO's "technical"? Are physical and abstract mutually exclusive?
It is close to what the European Patant Office mean by technical.
An example a computerized breathelizer can be patented, as can an industrial robot.
A new data processing program running on a general purpose computer can't be. It can only affect the world outside the computer by the agent of a human.
Whether physical and abstract is mutually exclusive doesn't really come in to it.
But doesn't the decision say that any claim that can be even partially implemented in software strays into non-statutory subject matter?
These days, the breathalizer and the industrial robot will both be implemented using substantial amounts of software. Does this decision require that any patent claims on these devices clearly segregate the hardware from the software, and address only hardware aspects of the invention?
I doubt that such a clear segregation is possible in today's world.
For example, consider FPGAs (field programmable gate arrays), a common way of creating short production runs of special purpose integrated circuits from general-purpose blanks. An FPGA might well be considered to comprise software.
On the other hand, once traditional software is burned into read-only memory (such as is done with BIOS (basic I/O subsystem) software in a personal computer, is it no longer software because it has been converted to a permanent physical representation, and thus patentable?
"A turing machine can make everything that software alone can make: However it can't make a better mousetrap. Or any other physical invention!"
Yeah, and a turing machine can make a digitally controlled mousetrap, so nothing is patentable!
Carl,
Exactly.
Anon E. Mouse
The scope of the claim exceeds statutory subject matter because software per se is non-statutory, and the specification states that a possible embodiment of the claim can be software by itself, without any hardware component. The claim could be amended to fall within statutory subject matter by any number of ways that explicitly recites physical structure, such as for example, a physical storage medium containing the software. By the way, software per se cannot be executed by any known processor so that this claim (considered as pure software as the spec explicitly says this claim covers)lacks enablement and utility under 112, 1st. The BAPI decision isn't by any standard weird, it's routine interpretation according to 101 training and MPEP 2100.
Yes, Beauregard claims. But if my claim is limited to a "physical storage medium containing software that...", doesn't my claim cover only the presence of the software on the storage medium? Or at most one physical representation of my software?
How can my Beauregard claim protect the algorithm implemented by the software (which is of course the goal of a software patent)? There are many ways of recording an algorithm on a physical storage medium. Mulitple of these ways can result in mathematically equivalent but physically very different representations on the physical media. If abstract descriptions of algorithms are not patentable, how do I write a Beauregard claim that covers all possible physical representations of my algorithm without inserting a mathematical definition of the algorithm into the claim (which would make it non-statutory)?
(You can probably tell I'm a computer guy, not patent counsel.)
On reflection, I suppose the issue is this "statutory" thing. A 19th century statute is trying to deal with 21st century technology. The courts are just trying to cope.
I'm a computer practitioner, so I see software on a general purpose computer as the ultimate erector set. Just one more "material" with which I can build things.
Defining any embodiment based on software to be non-statutory is (to me at least) rather like making machines assembled with bolts statutory but machines assembled with welds non-statutory. To one skilled in both arts, these two embodiments are equally reasonable and obvious once the disclosure of basic machine principles is made in a patent.
I believe the same is true with respect to machine-tool embodiments versus software-on-a-computer embodiments. Problem is, the statutes don't see it that way.
Guess I'll have to learn welding if I want to be an inventor.
One thing to keep in mind is that "patentable subject matter" means different things for different parts of the patent statute.
Under sections 102/103, the claims can be rejected if someone else has done the same thing, or an obvious variant. As a side note, the same claims were also rejected in the opinion due to prior art.
Under section 101, the claims can be rejected if they are not tied to a particular machine or transform an article to a different state or thing. Here, the "weirdness" is that the BPAI's rationale for rejection does not appear to be rooted in any US caselaw dealing with section 101. "Software per se" has no application whatsoever in US patent caselaw.
Turing machines were invented as an abstract model to determine what is computable by rule or function (not by machine - the electronic computing machine had not been built yet). Anything that can be computed by a Turing machine can be expressed in Lambda calculus.
So yes, if a Turing machine can do it then it's mathematics. If it's mathematics it's not patenable.
Note that a Turing machine cannot produce a better digitally controlled mousetrap, but it can execute mousetrap-control software. You can patent your trap and your controller, but not your control software algorithm :).
Peter,
I just finished going round and round with some "software" gurus on the PLI site as to the meaning of "software".
My position (and I'll stick to it) is that there is no clear cut definition for "software". It is a fuzzy, touchy feely word with no precise definition.
Moreover, I have no idea what the BPAI is saying when they say that a method claim (whose body BTW makes antecedent reference to the preamble) is "directed to" software. What the heck does the term "directed to" mean? Do we draw some sort of pointing vector, and if so, from which point to what other point?
______
p.s. Perhaps you should post hyperlinks to both Ex parte Godwin decisions, the original and the rehearing. Also, I did not check to see if the patent app itself is published and if so whether the spec defines claim terms such as "rendering", "portal", "portlet", etc. Thanks for the interesting discussion.
"My position (and I'll stick to it) is that there is no clear cut definition for "software". It is a fuzzy, touchy feely word with no precise definition."
Then I suppose it makes sense not to allow people to patent it.
[Warning - lengthy post! Someone asked a computer practitioner what 'software' was.]
Maybe the problem is that we tend to anthropomorphize machines -- that is, talk about them like they were human. This confuses the courts because -- while we are actually talking about machines and very physical attributes of those machines -- it sounds like we are talking about abstract human concepts, which of course are non-statutory.
In human terms, software is simply a set of instructions that a machine executes. That's how we humans think of it. But of course a machine does not "obey" humans and has no concept of "instructions".
When I write software in some programming language, a compiler eventually reduces those human-readable instructions to a sequence of hardware-level codes called "machine language". (This, of course, is another example of anthropomorphizing.) When the machine language codes are loaded into a suitable machine, the machine (due to its design) acts in certain hopefully-predictable ways. Computers are complex machines and have a wide repertoire of possible actions, but in concept loading machine code into a computer is no different than than twiddling the controls on your iPod. You are just arranging the machine to make it perform as you want it to. "Software" is simply that set of arrangements, resulting in a machine that takes certain actions.
The machine has no "language" in the human sense. What is happening is that the compiler generates a set of data that, when loaded into a specific type of computer, causes that computer to execute computations and generate results in a manner that corresponds to the original human-written specifications.
But what is really happening inside the machine is that certain registers are set in certain ways that cause the machine to take certain actions. These registers might as well be physical cams and levers, but in modern computers they tend to be bits of (equally physical) semiconductor materials. Instead of a physical position state, we have (an equally physical) electrical conductivity state.
So in my view, typical software (at least when reduced to concrete machine language code) is no different from an arrangement of cams and levers. In fact, computer science arguably began with Charles Babbage (q.v. on Wikipedia), who built the equivalent of a modern electrically-based computer using purely mechanical components.
From the Wikipedia article: "The data and program memory were separated, operation was instruction based, the control unit could make conditional jumps and the machine had a separate I/O unit."
This tells me that Babbage's machine used "software" that directed its operation. But this "software" was purely physical in a manner that I assume would qualify as statutory under 101.
So why can't I patent an arrangement of register values that causes a general purpose computer to accomplish certain tasks? How is this different from an arrangement of cams and levers that accomplishes those same tasks?
Or maybe I can, and it's called a Beauregard claim...
Bravo Carl,
You wrote the long thesis that I did not want to bother with. Yes, exactly. There are no "instructions". There are only energetic input signals (be they electrical, optical, mechanical, etc.) and/or physical inputs applied to the physical machine that we humans name as a "computer" and there are the energetic output signals or output products produced by the machine that we humans interpret as being intelligible "results".
Accordingly, I can supply analog "instruction" signals to an analog electrical computer machine and interpret its analog output signals as being intelligible "results". I can supply chemical "instruction" signals (reactants) into an analog chemicals processing machine and interpret its outputs (chemical products) as being intelligible "results".
All these theories about mythical Turing machines and mythical man months are distractions.
There are some things for example which an analog computer can do (well come pretty close to doing) that no Turing machine can do. For instance, subtract one irrational number from another and produce a rational result. i.e., (Pi +1) minus Pi equals 1.0. This is a computation and yet no Turing machine can perform it because its one-times-infinity long tape cannot store two irrational numbers.
This is just a legal pish posh, an exersize in nonsence writing
"Software per se", software on a physical storage medium, software written on a piece of toilet paper... blah... blah...blah...
What is the f****** difference ?
Just give me a break already, pleeeeeeeeeeeeeeeeeeeeeeeze !!!
"Note that a Turing machine cannot produce a better digitally controlled mousetrap, but it can execute mousetrap-control software. You can patent your trap and your controller, but not your control software algorithm :)."
Nope, for the "better mousetrap" produced in mass quantities all controlling software would be implemented as ASIC, that is in pure hardware.
Making software control algorithm unpatentable just creates an unhealthy legal dichotomy and a HUGE loophole for infringers.
Nothing positive is achieved here.
I agree. Requiring software to be "physically stored" in order to be patentable (a la Beauregard claims) provides no meaningful protection to a "software" invention.
If my claim discloses my software invention in physical terms (for example, as "a pattern of fused circuit paths within an ASIC, said pattern acting to..."), then anyone can steal my invention and reduce it to physical embodiment on any other form of computing device without infringing. For example, a Beauregard claim that specifies a "disk storage medium" would be evaded by storing the software instead on a portable USB memory stick.
On the other hand, if my claim discloses the sequence of steps that defines my invention and explains how these steps can be reduced to physical embodiment on a range of commonly known computing devices, then my invention appears to be non-statutory under 101.
Either way, my invention is unprotected.
Unless... can we invoke the doctrine of equivalence and say that a Beauregard claim disclosing software stored on a disk also covers software stored on any other functionally equivalent storage medium generally known to those skilled in the art? Or would non-disk embodiments fall into the "obvious" category: non-infringing as to the original claim, but also non-patentable in their own right over the original claim?
Carl wrote:
"This tells me that Babbage's machine used "software" that directed its operation. But this "software" was purely physical in a manner that I assume would qualify as statutory under 101.
So why can't I patent an arrangement of register values that causes a general purpose computer to accomplish certain tasks? How is this different from an arrangement of cams and levers that accomplishes those same tasks?"
Because the really smart software gurus on PLI say so.
I agree with all you've said, Carl.
Anon E. Mouse
"Because the really smart software gurus on PLI say so"
Yeah, right...
those "really smart software gurus" also happen to be "software punks" - they like you coding for free (as in beer)
It is a well-recognized fact (among real tech gurus) that software operation is more appropriately described as a functional patentable machine, rather than a copyrighted arrangement of printed weird-looking characters on paper or display called "software"
anonymous - what is the last number in PI? then go f**k yourself for being rude
Post a Comment