Federal Circuit says computer memory system claims patent eligible, not abstract

Judge Kara Stoll of the Federal Circuit wrote the opinion for the majority.

Yesterday, in Visual Memory LLC v. NVIDIA Corp., the United States Court of Appeals for the Federal Circuit once again issued the rare decision finding a computer-implemented invention (i.e., software) to claim patent eligible subject matter. The decision authored by Judge Stoll (joined by Judge O’Malley) was not a unanimous decision. The majority determined that the patent claims drawn to a computer memory system did not cover an abstract idea and, therefore, the second step of the Alice test was an unnecessary inquiry. Judge Hughes dissented and would have found the claims not only to relate to an abstract idea under the first step of the Alice test, but would have found the claims to be patent ineligible.

The patent at issue was Visual Memory’s U.S. Patent No. 5,953,740, titled Computer memory system having programmable operational characteristics based on characteristics of a central processor.

The ’740 patent teaches that computer systems at the time of filing frequently used a three tiers of memory to enhance performance – low speed, medium speed and high-speed memory. The high-speed memory acts as the processor cache and is expensive in comparison to the low and medium speed memory. A three-tier memory hierarchy alleviates the limitations imposed by the cache’s size limits because it allows code and non-code data to be transferred from the main memory to the cache during operation. Replacement algorithms determine which data should be transferred from the main memory to the cache and which data in the cache should be replaced. This types of prior art memory systems lacked versatility because they were optimized for a specific processor.

The invention described in the ‘740 patent overcame the limitations of the prior art by creating a memory system having programmable characteristics, which would allow it to be used with multiple different processors without any reduction in performance.


Visual Memory sued NVIDIA for patent infringement of the ‘740 patent. NVIDIA filed a motion to dismiss arguing that the claims asserted were not directed to patent eligible subject matter. The district court granted NVIDIA’s motion and Visual Memory appealed.

“Our review of the ’740 patent claims demonstrates that they are directed to an improved computer memory system, not to the abstract idea of categorical data storage,” Judge Stoll wrote. “The specification explains that multiple benefits flow from the ’740 patent’s improved memory system.”

Dissenting, Judge Hughes disagreed, saying that the claims covered “nothing more than a black box,” and that “the patent lacks any details about how [the invention’s purpose] is achieved.” The majority found three flaws with this conclusion.

First, the patent included an appendix having 263 frames of computer code. Thus, the majority found the assumption that those frames of computer code taught nothing was improper when reviewing a dismissal under Federal Rule of Civil Procedure 12(b)(6), where all factual inferences are supposed to be given in favor of the non-moving party – the patent owner.

“Second, whether a patent specification teaches an ordinarily skilled artisan how to implement the claimed invention presents an enablement issue under 35 U.S.C. § 112, not an eligibility issue under § 101,” Judge Stoll wrote.

Third, the majority explained that the dissent assumes the innovative effort was found in the programming, but that assumption is contradicted by the patent specification itself. After pointing to various places within the specification that explained that the invention is an improved memory system, Judge Stoll concluded: “Configuring the memory system based on the type of processor connected to the memory system is the improvement in computer technology to which the claims are directed.”

The majority did not express any opinion on the ultimate question of validity, since the case was disposed of on a motion to dismiss raising only the question of patent eligibility. Nevertheless, Judges Stoll and O’Malley concluded that the claims did not related to an abstract idea under the first step of the Alice inquiry, making it unnecessary to proceed to step two of the Alice test.

Obviously, this decision makes perfect sense. A computer memory system is not something that can be or should ever be capable of being considered abstract. The fact that the district court and a member of the Federal Circuit were able to convince themselves that something so clearly not abstract was, in fact, abstract and patent ineligible speaks volumes of the utter state of chaos that is patent eligibility law in the United States.

The fact that anyone would view 263 frames of computer code as not at a minimum being “significantly more” than an abstract idea under the second step of the Alice inquiry is inexplicable. It is as if Judge Hughes was reading the claims wholly devoid of any specification support, which is of course what decision-makers in this space frequently do. Of course, that flies in the face of volumes of black-letter patent law, including Supreme Court precedent, that commands that the claims be read in light of the specification, not in spite of the specification or dissected from the specification.

That this was not a unanimous decision on such a simple issue that is so fundamental to computer operation is all too predictable, yet still unbelievable. I’m afraid we really have fallen through the looking glass.

UPDATED at 12:22pm ET to add the last 2 paragraphs.


Warning & Disclaimer: The pages, articles and comments on IPWatchdog.com do not constitute legal advice, nor do they create any attorney-client relationship. The articles published express the personal opinion and views of the author as of the time of publication and should not be attributed to the author’s employer, clients or the sponsors of IPWatchdog.com.

Join the Discussion

45 comments so far.

  • [Avatar for step back]
    step back
    August 24, 2017 11:52 am

    B @44

    Regrettably we live in a time of blind obedience to incompetent authority and of industry capture of the courts (especially SCOTUS) and the legislature.

    Most examiners are not law school graduates. So you’re not going to get any meaningful discussion with them about the constitutionality of Alice/Mayo or any other aspects of the multitude of laws that USPTO break every day.

    Examiners do what their bosses (SPEs) tell them to do and those middle managers tote the line for upper management. The game is to maximize revenues while fleecing the American inventor.

    That’s how we get to the goal line of making America great again. Right?

  • [Avatar for B]
    August 24, 2017 11:19 am

    “Try telling it to the judge. I mean a judge like Hughes of the Federal Circuit. All your talk about Turing and algorithms will go right over his head into the trash heaps of abstractionism.”

    Step Back – while I think the various contributors to this debate have so many interesting and valid points – you owe me a new keyboard given the coffee I just spewed everywhere. I so agree. Judges like Hughes think there’s no difference between library science and computer science.

    With this said, I am thinking of conducting an official survey of patent examiners on their opinion of Federal Circuit cases. My informal survey results are overwhelming in that examiners are totally confused by the inconsistency of Federal Circuit cases. The problem is that abstract no longer means abstract (thanks to Alice), and the preemption doctrine (thanks to Ariosa) no longer requires actual preemption, in whole or in part, of an “abstract” idea. Also leading the charge from the plane of reasoned and arguable debate over the cliff of vast stupidity is about half the Federal Circuit, who are convinced that a clever metaphor (clever in their minds) and/or specious phrase hold the key to determining abstractness. Determining abstractness based on all limitations as an ordered combination? Soooo passé. Determining abstractness based on evidence? Not necessary.

    The Supreme Court warned the judiciary not to go overboard lest the except swallow the rule. I’m convinced most the CAFC never read Alice in its entirety.

  • [Avatar for step back]
    step back
    August 24, 2017 12:14 am

    David Stein @0101011… oh what the heck

    Try telling it to the judge. I mean a judge like Hughes of the Federal Circuit. All your talk about Turing and algorithms will go right over his head into the trash heaps of abstractionism.

    All that is fine if you’re a theoretician.
    But if you’re an inventor who has just sunk 100’s of thousands into your patent and it goes poof into the night when the good judge gets his claws on it, it won’t be such theoretical fun anymore.

    (BTW we fully agree that digital hardware and software are functionally interchangeable. Doubt it as you will, but I’ve written my share of code and soldered chips into boards in my days of yore.)

  • [Avatar for David Stein]
    David Stein
    August 23, 2017 10:20 pm

    Mark @39:

    At the risk of repeating other points that I’ve raised for other posters, I’d like to address your characterization.

    Consider this landmark paper from the venerable Xeroc PARC Research Institute:


    This research covers the development of sorting algorithms with performance optimizations based on their usage of the memory cache. The paper recognizes that if you must perform comparisons among a whole lot of words – more than you can cache – then it’s helpful to choose methods that will perform several comparisons with the same word in chronological proximity, so that it remains in the cache.

    Now, hopefully you’ll agree with me that:

    * The design focus and innovation of this paper are 100% software. Given *a* cache, *any* cache, the paper considers whether certain algorithms will perform better than others.

    * The actual cache in use is immaterial. Section 2 of the paper presents an introduction of caching technology. Some parts discuss tuning the algorithms to the properties of a particular cache – its size, access time, or caching strategy – but not of modifying any of those properties.

    * The caches discussed in this paper, as a physical memory device, have no novelty. The research and functional advantages are purely in *software design choices* that make better use of the ordinary functionality of the cache.

    If you agree with me on those points, then you must also accept that the innovation in Visual Memory’s patent is purely software.

    The cache in this patent is just a cache. It stores data on request; it provides data on request. It uses a caching strategy to choose which data to keep. Whether the computer featuring this cache uses *this technique* or *any other technique*, it performs, as a memory device, in exactly the same manner.

    Rather, all of the novelty in Visual Memory’s patent is in a logical process that controls what types of data get cached – to make better caching decisions. And it does so by keeping some metadata around to indicate something about what is in the cache.

    The innovation is not hardware. It is a caching algorithm – it is described as logical functionality that operates on the data *in* the cache, for the purpose of optimizing applications that cache data of different types. All of the innovation is on the logical organization of data stored in the cache – irrespective of how the cache, as a memory structure, *physically* operates.

    As I’ve noted repeatedly, I do not believe that this characterization demeans the patent, diminishes the value of the invention, nor should alter the application of 35 USC on this patent. This is a purely technical analysis of where the novelty in the claim resides.

  • [Avatar for David Stein]
    David Stein
    August 23, 2017 09:49 pm

    step back @40:

    “Computer haters?” Patenting technology in computing generally – and patenting software technology specifically – is my bread-and-butter. I’ve earned a master’s in computer science, and I’ve written quite a lot of useful software. You’d probably have a lot of difficulty finding someone less in the “computer hater” camp than me.

    Many computing inventions involve novel computing machinery. A processor that handles new instruction sets; a new bus configuration that provides faster and more reliable throughput; a memory device that is physically distinct, such as with a new physical data layout – these are all novel computing machinery.

    And many computing inventions don’t. A compression algorithm can be run on any computing device that supports a basic set of logical operations. A data structure, such as a hashtable, can be stored on any memory device that is capable of storing a collection of 0s and 1s.

    Both classes of inventions are, generally, technology. The term “generic computer” does not demean the value of the invention – software inventions can present just as much technology, and just as much value, as hardware. Neither should be ruled abstract if there is a functional, significant difference.

    Indeed, the suitability of *nearly any* modern computing device for such logical power is a fundamental principle of computer science: the Church-Turing thesis. (Yes, that Turing.)

    Indeed, if you’re aware of the Church-Turing thesis and modern computing techniques, then you will appreciate that the fundamental distinction between hardware and software. Is an FPGA hardware or software? How about a very simple microcontroller that repeatedly executes four arithmetic operations in series? Does it actually matter if control signals for a device are stored as an array, with some form of addressing, or as voltage values in a set of latches or flip-flops? … etc.

    With respect, your frustration could not be more poorly misdirected. I am discussing, in a purely technical sense, whether *this claimed invention* is software or hardware. And as I’ve stated repeatedly, this technical characterization should not matter for the patent-eligibility of the invention: 35 USC draws no distinction.

  • [Avatar for step back]
    step back
    August 23, 2017 09:17 pm

    Mark @39,

    No matter how many times you try to explain computer hardware to the computer haters, they will still insists that there is something called a “generic” computer (whatever that means) and that all computer architectures are equivalent (because Turing said so).

    It’s hopeless. The only poetic justice would be if their 14nm integrated semiconductor circuits could be replaced with vacuum tube circuits because “they” insist it’s all generic and equivalent. Sigh.

  • [Avatar for Mark Summerfield]
    Mark Summerfield
    August 23, 2017 08:14 pm

    David Stein @34: “Here’s the software part…”

    With respect, the limitation you quote does not recite software. It defines behaviour of the cache controller (hardware) based upon how it is configured via software, but the software itself is not claimed. The specification describes hardware, in the form of a block diagram and accompanying microfiche appendix that includes a complete implementation expressed in a hardware description language.

    I agree, of course, that the statute makes no relevant distinction between hardware and software, although I think it pretty clear that the Supreme Court has done so (as have courts in other jurisdictions, including my own home country of Australia). In this case, however, the claim cannot be directed to software, and therefore should not be treated as such, to the extent that the courts have drawn a distinction. By definition, a cache controller is responsible for determining where to store/retrieve data and instructions required by the processor on a cycle-by-cycle basis. It therefore necessarily operates at the hardware level.

    I also agree that defining operation of the hardware in such broad functional terms may be problematic. But, if so, then that is a 112 issue, not a 101 issue. And, as others have pointed out, it is an issue that may require evidence of the skill and knowledge of the PHOSITA, and the state-of-the-art, in 1993. I actually taught computer systems engineering and digital design to third year EE students from 1997 to 1999, so I may have a better handle than most on what was considered ordinary skill and knowledge at around that time. It does not strike me as self-evident that the claim is invalid, on any ground.

    The District Court decided the case on the pleadings, and that is the decision that has been rightly overturned by the Federal Circuit. There is no way that the validity of these claims can be assessed without being properly construed with the benefit of suitable expert evidence. That the District Court felt that it was able and entitled to do so emphasises the growing problem with Alice and its progeny.

  • [Avatar for David Stein]
    David Stein
    August 23, 2017 07:51 pm

    B @36:

    > I assert that the data-type cache storage was not generic.

    The novelty is not in the hardware. The hardware itself is generic. 100% of the novelty is in the *use* of this generic hardware by software and the data that it consumes.

    Let’s say I developed a particular kind of data structure – let’s say, a binary tree with upward links as well as downward links. If I stored a data structure, organized in that manner, in a completely generic memory – would that comprise a “hardware” case?

    By way of analogy: if you take a generic car, and put a uniquely skilled human in the driver’s seat, one who can make the car do amazing things – have you invented made a new car?

    This rationale creates a lot of problems. Every single software invention is a “hardware” case, because all of them necessarily require a processor and a memory. And by that same rationale, every hardware case is more properly classified as a chemistry case (since computing hardware is fabricated through processes such as chemical deposition, etc.) And by that same rationale, every single form of technology is, at its core, a “physics” case.

    If we’re actually going to pigeonhole technologies into technical fields – which, I maintain, is immaterial and even dangerous – then this exercise must be based on the level of abstraction at the point of novelty exists. Here, the point of novelty is in the data, and the processes that consume it. QED: software.

  • [Avatar for step back]
    step back
    August 23, 2017 07:22 pm

    In reality, there is no such thing as mere “data” and generic “information”.
    The electrons don’t care what we self important primates call them.
    They move nonetheless.
    In their respective localities.

    (To paraphrase Galileo in his rumored whisper after recanting the heliocentric model before the grand inquisitors.)

  • [Avatar for B]
    August 23, 2017 06:56 pm

    “That’s it. Even by 1993 standards, it was not only generic but a component of every computer in production.”

    Respectfully, I assert that the data-type cache storage was not generic.

    “so much navel-gazing, so little clarity.”

    I don’t think Judge Hughes has seen his navel in 30 years. (cheap-shot!)

    ‘But if you take the leap from the claim language to “categorical data storage” it’s easy to make an analogy to a librarian.’

    Yes, but have you ever heard of a librarian storing instructions/code on one shelf and variables/data on another shelf? Just goes to prove these black-robed monkeys read out what they want on gut feelings.

  • [Avatar for JS in DFW]
    JS in DFW
    August 23, 2017 06:36 pm

    B @ 31,

    FYI, here is the quote from the lower court (Andrews):

    Here, the claims are directed to the abstract idea of categorical data storage.
    Humans have categorized data for many years. For instance, a library may have an easily accessible section for popular novels, while maintaining less accessible storage for less-requested materials. This is an “undisputably well-known” practice that “humans have always performed.”

    I think this is a great example of how slippery the Alice step 1 slope is. These claims are far removed from a librarian doing her job. But if you take the leap from the claim language to “categorical data storage” it’s easy to make an analogy to a librarian.

  • [Avatar for David Stein]
    David Stein
    August 23, 2017 06:24 pm

    B @32:

    But here’s the hardware in claim 1:

    “A computer memory system connectable to a processor…, wherein said system is connectable to said processor by a bus, said system comprising: main memory connected to said bus; and a cache connected to said bus.”

    That’s it. Even by 1993 standards, it was not only generic but a component of every computer in production.

    Here’s the software part:

    “one or more programmable operational characteristics, said characteristics being defined through configuration by said computer based on the type of said processor, … wherein a programmable operational characteristic of said system determines a type of data stored by said cache.”

    This is merely data. It is data with certain (poorly defined) properties, but it is still data. It may be consumed by functional processes (not present in claim 1) that affect how the cache is used, but *it is still data*.

    Again, and more importantly – this determination is immaterial. Pigeonholing this claim as “hardware” vs. “software,” with the implication that patent law applies differently based on this determination, is contrary to 35 USC, which neither states nor implies any such principle.

  • [Avatar for David Stein]
    David Stein
    August 23, 2017 06:19 pm

    B @29:

    Fair enough.

    I do find it interesting that both Stoll and Hughes were willing to operate on the tacit assumption that the term has some specific meaning. I’d have presumed that Hughes, at least, would have raised the prospect that § 101 issues cannot be decided over a term with a completely open-ended meaning. In some instances, § 112 should be even more of a “threshold test” than § 101.

    Of course, this is a consequence of the conflation of § 101 with §§ 102, 103, and 112(a)/(b)/(f). Even the judges can’t keep the principles, rationale, and cases straight any more. Consider the ongoing confusion about whether “preemption” vs. “overbreadth” vs. “abstractness” vs. “definiteness,” etc. … so much navel-gazing, so little clarity. Everyone loses when basic rules are this muddy.

  • [Avatar for B]
    August 23, 2017 06:14 pm

    To Mr. Stein,

    ” Look at the claim: the hardware is 100% generic.”

    I respectfully disagree. It is generic today, but not in 1993. Using a cache to store only specific forms of information was (IMHO) an advancement in the art

  • [Avatar for B]
    August 23, 2017 06:11 pm

    To JS – regarding your statement: ‘Hughes would have affirmed held that all of the claims, including claim 8 above, were “directed to” activities like a librarian putting a book on a shelf.’

    Hughes did state: “I would find the ’740 claims are directed to the abstract idea of categorical data storage,” so your point is valid IMHO.

    Hughes would assert that Diamond v. Diehr was directed to monitoring a rubber curing process, and invalidated the claims. The man has an amazing ignorance of technology, and believes that all technology can be understood in simplistic metaphors.

  • [Avatar for David Stein]
    David Stein
    August 23, 2017 06:03 pm

    Step Back @22:

    > The fact is that this is a dammn hardware case. It deals with how the high speed caches are used for different kinds of CPU’s.

    First – nothing in 35 USC or 37 CFR provides that patent law operates differently for “software” inventions, “hardware” inventions, or anything else. The same laws and standards, including § 101, should apply to every patent irrespective of its technical field.

    Second – claim 1, on which this case turned, is not a “hardware” patent. Look at the claim: the hardware is 100% generic. The entire point of novelty is in a particular type of data stored in a completely ordinary memory.

    Sure, it’s a fairly low-level case – in the stack of abstractions that comprise a typical computer, this one is pretty close to the hardware. But that is not the proper delineation of “software” cases vs. “hardware” cases. Hardware drivers and BIOS / UEFI are as low-level as you can get – yet they are both composed of software. So is firmware, which is software that you can transfer to a dedicated EEPROM for the device.

    > One needs to dive into the details.

    One should not need to dive into the specification to salvage subject matter that should have been in the claims.

    U.S. patent law does not allow omnibus claims, of the form: “We claim a device substantially as sown and described in the specification.” Claims that provide empty, generic terminology – those which have no identifiable meaning apart from the specification – do not satisfy the fundamental principle of peripheral claiming.

  • [Avatar for B]
    August 23, 2017 06:03 pm

    Dear Mr. Stein,

    I admittedly did not bother to address the 112 issues. With this in mind, Judge Hughes apparently understood what was meant by “programmable operational characteristic” in light of evidence (e.g., spec and other claims) and argument set before him. There were no apparent issues of claim construction based on the published decision.

    That said, I do not believe NVIDIA had an opportunity to discuss 112 issues as this case was on appeal after a dismissal based on a 12(B)(6) motion.

  • [Avatar for JS in DFW]
    JS in DFW
    August 23, 2017 05:56 pm

    @ step back, nice article. In this case, the lower court’s decision that Hughes would have affirmed held that all of the claims, including claim 8 above, were “directed to” activities like a librarian putting a book on a shelf. (I am not joking) Just to say that again, in Hughes and Andrews opinion, the claim above is just like a librarian putting books on a shelf. Clearly they have never solved a technical problem.

  • [Avatar for David Stein]
    David Stein
    August 23, 2017 05:43 pm

    B @21:

    I generally agree with your remarks. Claim 1 isn’t “abstract” as per Alice – it does not merely recite data.

    I will pose this to you, B: Claim 1 has a potentially more significant, non-Alice § 101 problem involves the printed-matter doctrine. Claim 1 literally recites a physical memory structure, and disembodied metadata stored *somewhere* (not even specified where!) The data itself does nothing; rather, actually-functional processes – which do not appear in claim 1 – consume it. The Digitech Image case is relevant here. In this specific case, I’m willing to give it pass because the data relates to the operation of the physical memory. But it is a more interesting question that the court did not (openly) consider.

    However, regarding this:

    > Hughes mistakes breadth for abstract. Hughes states: “The ‘programmable operational characteristic” is nothing more than a black box for performing the abstract idea of storing data based on its characteristic.”

    I agree that this claim element is not abstract – but it is most certainly indefinite.

    “Programmable operational characteristic.” What does that term mean? This is a made-up term – it has no identifiable meaning in the technical field. It uses intentionally generic language (note the M-W definition of “characteristic”: “a distinguishing trait, quality, or property”). Claim 1 provides not even a hint of a definition. If you replace it with the equally generic term “data,” what you get is: a computer that stores data indicating *something* about the data in the cache, and it has *something* to do with the type of processor.

    Yes, claims 6 and 8 are both better; yes, the specification provides more information. But the case turned on claim 1. And claims cannot feature significant elements that are given a generic name, with the tacit suggestion to refer to the specification to figure out what the term may or may not mean. Such tactics are anathema to a patent system where scope is defined by the periphery of the claims.

    Interestingly, both Judges Stoll and Hughes briefly addressed § 112 (a)/¶1 enablement, which seems fine: given any particular definition of the term “programmable operational characteristic,” the specification and level of skill seems sufficient to implement it. But neither judge addressed § 112 (b)/¶2 ambiguity: the term itself has an unclear (and indeed, unlimited!) meaning. Judge Stoll notes that “(other) challenges under § 112 could prove successful, but these questions are not before us.” It appears that NVIDIA’s patent counsel missed the opportunity to challenge this claim based its most obvious deficiency. Perhaps the siren song of “any computer-related patent can be invalidated under § 101” was too strong to resist.

  • [Avatar for step back]
    step back
    August 23, 2017 05:36 pm

    JS @25

    When one is an important and all knowing federal judge, one does not belabor oneself with trivial matters such as facts.

    A smell test is good enough. 😉


  • [Avatar for JS in DFW]
    JS in DFW
    August 23, 2017 04:25 pm

    Everyone is talking about claim 1 – Claim 1 is the most broad and I was very pleased that the CAFC decided it was eligible. But note that the district court (Judge Andrews) threw out all of the patent’s claims, and the dissent did too. Check out claim 8 below, it is not nearly as broad as claim 1. It includes the “programmable operational characteristic” which seems to be be problematic for a lot of people, but I am certain that a person having ordinary skill in the art understands it, and it lays out the architecture and requirements for different caches. How on earth is this abstract? Even to a patent-hating liberal arts major, how on earth is this abstract?

    8. A computer memory system, connectable to a processor and bus master by a bus and having programmable operational characteristics based on characteristics of said processor, comprising:

    a plurality of caches connected to said bus;

    a main memory connected to said bus and including a plurality of pages of a first and second type; and

    a register connected to said memory for holding the page address of the most recently accessed of said first type pages;

    wherein a first programmable characteristic of said system determines whether a first of said caches stores only code data or whether said first of said caches stores both non-code data and code data;

    wherein a second of said caches performs buffering of data writes to said main memory, and a second programmable characteristic of said system determines whether said second of said caches performs buffering of data solely from said bus master and whether said second of said caches performs buffering of data both from said bus master and said processor;

    wherein said page address is used to reopen a most recently accessed of said first type pages after one of said second type pages has been accessed; and
    wherein a third programmable operational characteristic of said system determines a type of data stored by said main memory.

  • [Avatar for step back]
    step back
    August 23, 2017 10:48 am

    ALL claims, meaning absolutely all claims “are” abstract.

    They are merely words scrivenered on parchment (e.g., in the English language) to represent the concept of a corresponding invention.

    This is not a pipe. Get it? If not, link to the following or similar explanations of the existentialist concept:

    What is important to understand is that Judge Hughes possesses a rare and magical field direction indicating compass.

    Instead of the usual N, E, S, W markers found on a normal compass, his magical compass denotes the major circle points as A, A, A and A; where “A” means Abstract. A select few angular micro-strokes on the circle get the NA notation, meaning Not Abstract. Only judge Hughes and those secretly sworn into his power group know where those are. They know them when they see them.

    Whenever confronted with the words of a claim, the good Judge pulls out his magical mystical compass and hovers it above the words. Its needle quickly and without belaboring itself too much aligns with the hidden field forces of the words and tells the Judge what those words are “directed to”.

    Yes, of course it’s almost always “A”. But that is not the fault of the good Judge. It is the reality of the universe as reflected by the angle markings on his pocket compass.

    No point swearing to the G-d you worship. It is as He hath willed it. The universe is just full of mysterious abstract misdirections.


  • [Avatar for B]
    August 23, 2017 12:11 am

    “Do the Fed. Cir judges understand that? At least one (Hughes dissenting) appears not to.”

    I swear to the G-d I believe in and every other one I don’t. Last year Hughes was practically screaming at me to confess that a particular claim was abstract. When I refused and asked whether there should be evidence that the claim was abstract, he crossed his arms, swiveled his chair to the side, and pouted.

  • [Avatar for step back]
    step back
    August 22, 2017 08:59 pm

    Stein @20

    Yes, this isn’t the best written claim in the world.
    But “feelings” shouldn’t be how we deal with it.

    The fact is that this is a dammn hardware case.
    It deals with how the high speed caches are used for different kinds of CPU’s. One needs to dive into the details. See for example, https://patentu.blogspot.com/2017/08/shellacking-chicken-shell.html and the point about the hardware descriptor language, CDL. Do the Fed. Cir judges understand that? At least one (Hughes dissenting) appears not to.

  • [Avatar for B]
    August 22, 2017 07:14 pm

    Dear Mr. Stein,

    The problem with Alice Corp. is that the SCOTUS broadened the idea of “abstract” to encompass that which is a legit machine or process under 101 by equating abstract to “well-known,” “routine,” and “conventional.” This was the slippery-slope absurd end result of Gottschalk v. Benson (claim 8 only, not claim 13 analysis) and Parker v. Fluke.

    Anyway, nothing in the present claims strikes me as “abstract” under real abstract, or neo-abstract based on 1990s technology.

    Of course Hughes dissent is telling of his capacity to adjudge 101 issues. He mistakes breadth for abstract. Hughes states: “The ‘programmable operational characteristic” is nothing more than a black box for performing the abstract idea of storing data based on its characteristic'”

    Well, hell, Hughes merely assumes that “storing data based on its characteristic” is abstract – or was at the time of the invention. The “wherein” clause of the present claims isn’t merely a “goal,” it is an operational characteristic of the machine.

    Perhaps Hughes, THE KING OF BLIND CONCLUSORY ASSERTIONS, should read claims 1-7 in O’Reilly v. Morse. He’d overturn their patent eligibility in a heartbeat.

  • [Avatar for David Stein]
    David Stein
    August 22, 2017 05:24 pm

    I have mixed feelings about this case.

    On the one hand, I’m always glad to see more cases to fill in the patent-eligibility side of the Alice equation. I still receive rejections on § 101 grounds that basically state: “Anything done on a computer is per-se abstract unless you garbage up the claim with enough implementation details to make the claim worthlessly narrow” – which, of course, is just bizarrely inconsistent with case law and the Interim Eligibility Guidance.

    On the other hand… look at the claim:

    “1. A computer memory system (blahblahblah), said characteristics being defined through configuration by said computer based on the type of said processor, (blahblahblah), said system comprising:

    (some routine hardware),

    wherein a programmable operational characteristic of said system determines a type of data stored by said cache.”

    Claims that recite the point of novelty in a wherein clause – as a *quality* of a piece of data – are problematic. There’s no recitation of:

    1) Where the “programmable operating characteristic” comes from, or

    2) That the “programmable operating characteristic” is used for anything in the described device, including how it is used to affect the use of the cache.

    All the claim provides is that the “characteristics” indicate something about the type of data in the cache, and is also *in some unspecified way* related to the processor type. The End.

    (Of course, we can’t exactly fault the practitioners for the quality of this patent: it issued in 1999 (!) after being filed in 1993 (!). It most likely passed muster in the permissive era of State Street, so judging its quality based on 18 years of case law after it issued would be unfair.)

    I suppose that we should applaud the Federal Circuit’s newfound recognition that computing technologies – *even if badly claimed*, in ways that may fail other sections of 35 USC – should not be casually invalidated under § 101, which was always intended (and, until recently, always utilized) as a “threshold test.” Still, for every case we get like this one, we get a half-dozen others that stand for the opposite principle – such as Intellectual Ventures v. Symantec, in which malware scanning techniques were invalidated as “abstract.” (And the majority analysis in that case is disingenuous to the point of being offensive: it begins by citing Alice for the principle that “fundamental… practices” are abstract, which of course omits an important word: ECONOMIC.)

    Meanwhile… examiners continue to ignore all of the recent case law, and Robert Bahr’s memorandum updates to the Interim Eligibility Guidance. § 101 rejections are 99% boilerplate with MadLibs-style “copy-and-paste claim 1 here” inserts, and absurd statements like “all of claims 2 through 20 fail to provide ‘significantly more.'” These terrible rejections will weigh down the PTAB until either the USPTO or the courts do something that *forces* examiners to follow a more sane process.

    Gene, I agree that this area is devolving to Freeman-Walter-Abele levels of hopelessness. I don’t know how much longer we will have to suffer under a regime of uncertainty in such a fundamental, crucial question of valuable property rights.

  • [Avatar for Night Writer]
    Night Writer
    August 20, 2017 10:21 pm

    @17 and 18: The goons were selected for their ignorance of patent law and technology. The ignorant or outsiders are capable of great violence. The USSR would use the ignorant outsiders to quell the revolts.

    The group of judges appointed by Obama are just goons.

  • [Avatar for step back]
    step back
    August 19, 2017 02:49 pm

    If they read, they might understand that the claims are an integral part of the spec and not words floating in a vacuum.

    For example, the Visual Machine spec provides specific definitions for its caches.

  • [Avatar for step back]
    step back
    August 19, 2017 02:44 pm

    Clearly, the new and “improved” judges on the Fed. Cir. never bother to read and understand the laws passed by Congress:

    35 U.S.C. 112 Specification.
    (a) IN GENERAL.–The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention .
    (b) CONCLUSION.–The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.

  • [Avatar for Night Writer]
    Night Writer
    August 19, 2017 12:10 pm

    Still, it is disturbing that this only made it by a technological improvement to the computer.

  • [Avatar for Night Writer]
    Night Writer
    August 19, 2017 11:19 am

    People really have to wrap their head around the fact that these judges were selected by Google. And that these judges very likely received a call prior to nomination where an assistant to Obama extracted a pledge to get patents under control in exchange for an appointment. We are dealing goons not judges. Hughes is a charter member of the goon squad.

    Stoll is an incredibly poor substitute for Rader, and Google likely has less control over her as Obama was replacing Rader and likely didn’t want the hue and cry to get too loud. But, none-the-less, Stoll is not a person that should have been selected based on her record and personality at Finnegan.

    O’Malley is pre-Google. So, great that Stoll didn’t trash this patent. But, we have seen that she is happy to be part of the goon squad on most issues.

  • [Avatar for step back]
    step back
    August 17, 2017 03:13 pm

    err … I meant to also thank Mark Summerfield @9 re HDL

  • [Avatar for step back]
    step back
    August 17, 2017 03:11 pm

    Thanks CP/DC @11 !!!

    I did a bit more digging into the CDL hardware descriptor language and posted a link to it inside the following posting:


    Yes, ignorance is bask-worthy bliss.
    And blindsight is 20/20 times zero (0).

  • [Avatar for B]
    August 17, 2017 12:58 pm

    “No electronics engineer would regard it as disclosing anything other than hardware, just as surely as if the equivalent circuit diagrams had been provided.”

    Being an EE and a patent attorney, I’ve been well aware of the crossover between hardware and software since the introduction of FPGAs. I once designed a digital-processing-based pulse-shaping circuit for a wireless transmitter in C. I wrote the code – a converter, compiler and fitter took care of the rest. It is this background of mine that gives me insight as to just how ignorant the CAFC is, but for God’s sake let us not disturb the technical (mis)understandings of these art history and English majors wearing black robes.


  • [Avatar for CP in DC]
    CP in DC
    August 17, 2017 11:00 am


    The only way to avoid patent ineligibility under the Mayo/Alice framework is to avoid meeting step 1. I have not seen one case from either the Federal Circuit or PTAB where a claim survives Step 2. All cases on reversal focus on claim construction, interpreting the claims as not directed to an exception.

    In bio/pharma, we often argue the Mayo/Alice test requires two inventions in one claim. One if you have a biomarker or genetic test, which often triggers the natural correlation, mental process, or natural product 101 rejection. And a second one directed to the methodology of applying or detecting the biomarker or genetic test (the dreaded something more). Step 2 is often applied at such a high level of generality (contrary to Mayo and Alice) that anything used is deemed routine and conventional.

    The courts must reign in this overly broad application of Mayo/Alice test. If step 1 can be met by anything remotely associated with an exception, then few claims will survive Alice. The Federal Circuit is zeroing in on claim construction, but the PTO (the PTAB or examiners) is not.

  • [Avatar for Mark Nowotarski]
    Mark Nowotarski
    August 17, 2017 08:55 am

    It’s interesting that the case turned on whether or not the claims were directed to an abstract idea. I’ve just finished reviewing recent decisions by the PTAB on post Alice 101 ex parte business method appeals. In most of the reversals, the panels based their decision on the claims meeting step 1 (not an abstract idea) instead of step 2 (something more).

  • [Avatar for Mark Summerfield]
    Mark Summerfield
    August 17, 2017 06:34 am

    As I have also written in the comments on that other, rather less civilised, blog, this is not, in fact, a software patent. The 200+ frames of microfiche ‘code’ are, in fact, in a hardware description language and define the netlist for an integrated circuit implementation of a cache manager embodying the invention. The specification is actually very clear about this. No electronics engineer would regard it as disclosing anything other than hardware, just as surely as if the equivalent circuit diagrams had been provided.

    The fact that it was ever treated as ‘abstract’, or even as software, is absurd.

    And, Gene, the fact that you could read the court’s opinion (though obviously not the patent specification) and still think it relates to a ‘software patent’ says a great deal about the lack of qualification of the CAFC judges to do their job. It may be the correct result, but it cannot possibly be for the right reasons!

  • [Avatar for JS in DFW]
    JS in DFW
    August 17, 2017 06:28 am

    NVIDIA’s briefing, and oral argument, attempted to cram 112 into the 101 question, so it was nice to see the court squarely address that issue. Today infringers are trying to litigate 102, 103 and 112 at the pleadings stage with Alice, and this case helps stop the 112 arguments.

    Also, it was very surprising that NVIDIA decided to create precedent under the Alice exception that can be used against its own portfolio. A large portion of NVIDIA’s patent portfolio covers improvements to computing systems that include a programming element. I am not sure if the company did this knowingly or not – but in my opinion it took an incredible risk in trying to destroy Visual Memory’s patents under the Alice exception.

  • [Avatar for B]
    August 16, 2017 11:59 pm

    “If the PTO can’t provide guidance, who can?”

    Answer: Anyone. The nards writing the guidelines are clueless

  • [Avatar for B]
    August 16, 2017 11:58 pm

    Of course Judge Hughes dissented. He knows what is abstract and what is not – law and evidence be damned

  • [Avatar for CP in DC]
    CP in DC
    August 16, 2017 05:13 pm


    The Mayo/Alice test is completely unworkable for both hi tech and bio/pharma. Judges know this, the PTAB knows this, the examiners know this, and practitioners know this. Even the PTO guidelines don’t understand eligibility.

    The PTAB decided In re Chamberlain, a case with claims directed a method of treating a bone disorder, osteoporosis. The claims follow the PTO guidelines for the Julitis example. The PTAB, after the Alice two part analysis, found the claims directed to patent ineligible subject matter. This case demonstrates the problem. Even when applicants follow the examples by the PTO, the claims don’t survive the Mayo/Alice test.

    If the PTO can’t provide guidance, who can?

  • [Avatar for Gene Quinn]
    Gene Quinn
    August 16, 2017 03:26 pm


    The Mayo/Alice framework is now officially just like the FWA test, which is to say completely meaningless and untethered from any objective reality. FWA collapsed because if you were a judge that didn’t like software patents the claims were ineligible and if you liked patents they were eligible. That is officially what we have now without any doubt with Mayo/Alice. We all knew that was the case, but Visual Memory v. NVIDIA conclusively proves that to be the case. 263 frames of computer code and the claims are still to “nothing”? It is time for a group of judges on the CAFC to stand up and state the obvious — the Mayo/Alice test is unworkable in its present configuration and needs to be adjusted so that it can be objectively and evenly applied.


  • [Avatar for EG]
    August 16, 2017 03:19 pm

    Hey Gene,

    Hughes dissent is frankly mind-boggling: Is 200+ pages of code really meaningless for the purposes of patent-eligibility? Just proves that I could determine patent-eligibility better using a Ouija board than with the broken and nonsensical Mayo/Alice framework.

  • [Avatar for step back]
    step back
    August 16, 2017 01:37 pm

    It just gets stranger and stranger.

  • [Avatar for Ternary]
    August 16, 2017 01:34 pm

    “The fact that the district court and a member of the Federal Circuit were able to convince themselves that something so clearly not abstract was, in fact, abstract and patent ineligible speaks volumes of the utter state of chaos that is patent eligibility law in the United States.”

    Great article Gene. It is even more upsetting that Examiners, who supposedly have an engineering background, go along with this nonsense. Some engineering common sense has to return to our patent system.

    The only way to address this is by Congress better defining patent eligibility or by rescinding Alice.