Software Patent Law | Patents4Software.com

Last fall, I reported that the District Court for the Central District of California had used the Bilski abstract idea subject matter exclusion to invalidate the claims of Ultramercial’s U.S. Patent No. 7,346,545.  At that time I wasn’t 100% confident that the Federal Circuit wouldn’t agree.  Fortunately for holders of Internet-based inventions, Chief Judge Rader, joined by Lourie and O’Malley, in a well reasoned and common sense opinion, reversed and remanded in Ultramercial v. Hulu – Case No. 2010-1544.

As described in the decision, the claimed invention of the ‘545 patent is a “method for monetizing and distributing copyrighted products over the Internet.”  The court further observed that the ’545 patent seeks to remedy problems with prior art banner advertising, such as declining click-through rates, by introducing a method of product distribution that forces consumers to view and possibly even interact with advertisements before permitting access to the desired media product.  ’545 patent col. 2 ll. 14-18.

The court found that the claimed invention was “a practical application of the general concept of advertising as currency and an improvement to prior art technology.”  The court concluded that the claimed invention is not “so manifestly abstract as to override the statutory language of section 101.”

In its discussion, the Federal Circuit reinforced the notion that 35 U.S.C. Section 101 at least includes technical inventions:  “Moreover, title 35 does not list a single ineligible category, suggesting that any new, non-obvious, and fully disclosed technical advance is eligible for protection, subject to the following limited judicially created exceptions.”  The court also pointed out the claimed invention involved technical advances:  “[v]iewing the subject matter as a whole, the invention involves an extensive computer interface,” and that, “[b]y its terms, the claimed invention invokes computers and applications of computer technology.”  The notion that technical inventions in the software area fall within Section 101 was also put forth last year in Research Corp. Techs., Inc. v. Microsoft Corp., 627 F.3d 859 (Fed. Cir. 2010).  In that case, the court noted that “inventions with specific applications or improvements to technologies in the marketplace are not likely to be so abstract that they override the statutory language and framework of the Patent Act.”

The court also takes time to dismiss the contention that the software programming necessary to facilitate the invention deserves no patent protection or amounts to abstract subject matter or, in the confusing terminology of machines and physical transformations, fails to satisfy the “particular machine” requirement.  In doing so, the court reaches back nearly 20 years to its discussion in In re Allapat, Fed. Cir. 1994, where it observed that “programming creates a new machine, because a general purpose computer in effect becomes a special purpose computer once it is programmed to perform particular functions pursuant to instructions from program software.”

The court notes that, in other words, a programmed computer contains circuitry unique to that computer.  That “new machine” could be claimed in terms of a complex array of hardware circuits, or more efficiently, in terms of the programming that facilitates a unique function.  The digital computer may be considered by some the greatest invention of the twentieth century, and both this court and the Patent Office have long acknowledged that “improvements thereof” through inter-changeable software or hardware enhancements deserve patent protection.  Far from abstract, advances in computer technology—both hardware and software—drive innovation in every area of scientific and technical endeavor.

This notion, that a programmed computer is in itself a “particular machine”, refutes and rejects the notion that a claim to software in combination with a computer requires the computer itself to be “partricular” and not just a general purpose machine, in order to satisfy the “particular machine” requirement of the machine or transformation test.  Clearly this is good news for software patent claims.

On the other hand, the decision makes clear that claims reciting only a mathematical algorithm or steps that can be performed by the human mind are more likely to be abstract.  The court distinguished the claims of Cybersource from the Ultramercial claims: “Unlike the claims in CyberSource, the claims here require, among other things, controlled interaction with a consumer via an Internet website, something far removed from purely mental steps.”

While the Ultramercial case advances the principle that complex software programs with technical features (as carried out on a computing infrastructure) are in the statutory bucket, it  does not address what is to done with non-technical software inventions, indeed if there is anything such as that, that apply or implement abstract ideas.  To be on the safe side in drafting claims, it would be advisable to focus on describing and claiming the technical aspects of any invention involving a financial or otherwise suspect abstract concept.

Those holding or seeking software patents should be encouraged by this sensible decision, and I am encouraged the opinion continues to bring a solid common sense approach back to analyzing the patentability of software inventions.  While the “abstract idea” exception is far from dead, it at least appears that software inventions that are arguably “technical” in nature will survive a challenge from Bilski, at least as far as the Federal Circuit is concerned.

For your reference, the ’545 patent claims:

A method for distribution of products over the Internet via a facilitator, said method comprising the steps of:

a first step of receiving, from a content provider, media products that are covered by intellectual property rights protection and are available for purchase, wherein each said media product being comprised of at least one of text data, music data, and video data;

a second step of selecting a sponsor message to be associated with the media product, said sponsor message being selected from a plurality of sponsor messages, said second step including accessing an activity log to verify that the total number of times which the sponsor message has been previously pre-sented is less than the number of transac-tion cycles contracted by the sponsor of the sponsor message;

a third step of providing the media product for sale at an Internet website;

a fourth step of restricting general public access to said media product;

a fifth step of offering to a consumer access to the media product without charge to the consumer on the precondition that the con-sumer views the sponsor message;

a sixth step of receiving from the consumer a request to view the sponsor message, wherein the consumer submits said request in response to being offered access to the media product;

a seventh step of, in response to receiving the request from the consumer, facilitating the display of a sponsor message to the consumer;

an eighth step of, if the sponsor message is not an interactive message, allowing said con-sumer access to said media product after said step of facilitating the display of said sponsor message;

a ninth step of, if the sponsor message is an interactive message, presenting at least one query to the consumer and allowing said consumer access to said media product after receiving a response to said at least one query;

a tenth step of recording the transaction event to the activity log, said tenth step including updating the total number of times the sponsor message has been presented; and

an eleventh step of receiving payment from the sponsor of the sponsor message displayed.

With all the confusion about Section 101 and how it includes or excludes computer software inventions, or what level of disclosure constitutes “structure” for the purposes of 35 U.S.C. Section 112, it may help to take a quick step back and look at exactly what “software” is.  While Wikipedia is far from a perfect source to quote, I like its definition of software:  “Computer software, or just software, is a collection of computer programs and related data that provide the instructions for telling a computer what to do and how to do it. In other words, software is a conceptual entity which is a set of computer programs, procedures, and associated documentation concerned with the operation of a data processing system.” (http://en.wikipedia.org/wiki/Software)

In other words, “software” constitutes both a definition of specific electronic operations to be performed by an electronic computing system, and also, at its base level, actual instructions, in the form of electrical logic signals, to cause the computer to perform the electronic operations.  Generally, software can be divided into two categories:  low-level software that constitutes instructions in the form of digital logic elements that can be read by the computer, transformed into digital signals usable by the computer’s electronic circuits, and executed from a collection of possible machine operations.  Different combinations of these electronic operations can be used to accomplish a multitude of modern data processing results.

High-level software is software that can be used, by compilation or by interpretation using a computer itself, to produce the low-level software needed to program/cause the computer to perform desired electronic operations.  A computer, suitably programmed, can read high-level software to produce the low-level software and in turn electronic operations required to cause a computer to perform electronic computer operations.

High-level software itself if often represented by a flowchart, which itself typically describes a sequence of desired computer operations, at the electronic level, to perform a given computerized process.  In some cases, a flowchart can represent exactly the precise, lowest level electronic operations performed by a processor, such as peforming add functions, shift functions, load functions, bus functions, etc…   Such a flowchart would also serve as an exact specification for the construction of a fully hardware electronic device, i.e., one in which no software was involved, but rather all functionality was built into the electrical architecture/design of the electronic device.

Higher level flowcharts by nature and perhaps by definition, define less precisely exactly how a particular processor will perform the desired sequence of electronic operations, and as such omits the details of such operations.  However, the ultimate result or functionality to be achieved by whatever exact sequence is chosen is defined.  Where such “high level functionality” is novel, it could meet the test of non-obviousness.

Interestingly, cases such as In re Aoyama, discussed in a previous posting, suggest that some such software inventions can be described too generally to constitute “structure” under 35 U.S.C.  § 112 ¶ 6, and as such claims directed to such functionality in a means plus function format are unpatentable as indefinite under 35 U.S.C. § 112 ¶ 2.  This begs the question, what constitutes “structure” in a software environment.  Certainly, if the software is low-level, in that it exactly describes an exact electronic process, i.e., sequence of computer operations to be peformed, it must constitute an acceptable level of structure or else it is not possible to achieve a means plus function claim for a software invention.  When the description of functionality is at such a high level that a multitude of different low-level software-specified processes are possible for achieving the defined functionality, it can be said that constraints on “equivalents” under 35 U.S.C.  § 112 ¶ 6 can only be applied at the high level of the software description, since there is no low-level definition.  If the high-level description is a series of high level functions or steps, then one could argue that the specific means to achieve each function or step is unconstrained, and that the only constraint on equivalents is whether the high level functionality, i.e, the high level functions or steps, are equivalent.

The majority in In re Aoyama seems to be suggesting that where a software process is claimed under 35 U.S.C. § 112 ¶ 6, a machine-specific implementation must be provided, i.e., an implementation that describes with particularlity how the low level electronic operations available in a particular type of computer are used to perform the claimed operation.  This description would provide the most definitive description of software structure possible to describe, and therefore would have to pass muster or software description ever could.

In the end, software practicioners need to be aware that the safest grounding for a software case is in the actual electronic operations performed by a computer in response to the software.  The more examples of how high-level software can be implemented by low-level code, the more bullet proof a software application will be from the perspective of 35 U.S.C. § 112.  While one would think that actual source code (and the compiler or interpreter to be used) or object code may be overkill in this day and age, when it is available, submission of such code in an appendix may be the easiest way to always provide at least one very specific implementation in low-level software that is unassailably as structural as one can achieve.

My thanks to Justin McCarthy for this guest post:

On July 29, 2011, The Federal Circuit decided The Association for Molecular Pathology v. Myriad Genetics, 2010-1406 (Fed. Cir. July 29, 2011).  While this case deals with the patentability of gene fragments and methods of use, the case does present interesting questions surrounding the patentability of software.   Myriad was brought by a number of medical organizations, researchers, genetic counselors, and patients as a declatory judgment proceeding against a number of patents that Myriad had secured which protect both the isolated gene fragment BRCA1 and BRCA2 as well as methods for using those gene fragments to detect certain breast and ovarian cancers.  Myriad, at 8-9.   The court’s decision has two parts, both having implications for the software patent landscape.

Part I – The method claims:

The patent included claims drawn both to the gene fragments themselves as well as methods for using the gene fragments.  The court’s decision on the method claims was the only part of the decision that was unanimous.  The decision on the patentability of the gene fragments was split 2-1.

The challenged method claims recite:

1. (‘999 patent) A method for detecting a germline alteration in a BRCA1 gene, said alteration selected from the group consisting of the alterations set forth in Tables 12A, 14, 18 or 19 in a human which comprises analyzing a sequence of a BRCA1 gene or BRCA1 RNA from a human sample or analyzing a sequence of BRCA1 cDNA made from mRNA from said human sample with the proviso that said germline alteration is not a deletion of 4 nucleotides corresponding to base numbers 4184-4187 of SEQ ID NO:1.

1. (‘001 patent) A method for screening a tumor sample from a human subject for a somatic alteration in a BRCA1 gene in said tumor which comprises [] comparing a first sequence selected from the group consisting of a BRCA1 gene from said tumor sample, BRCA1 RNA from said tumor sample and BRCA1 cDNA made from mRNA from said tumor sample with a second sequence selected from the group consisting of BRCA1 gene from a nontumor sample of said subject, BRCA1 RNA from said nontumor sample and BRCA1 cDNA made from mRNA from said nontumor sample, wherein a difference in the sequence of the BRCA1 gene, BRCA1 RNA or BRCA1 cDNA from said tumor sample from the sequence of the BRCA1 gene, BRCA1 RNA or BRCA1 cDNA from said nontumor sample indicates a somatic alteration in the BRCA1 gene in said tumor sample.

20. (‘282 patent) A method for screening potential cancer therapeutics which comprises: growing a trans-formed eukaryotic host cell containing an altered BRCA1 gene causing cancer in the presence of a compound suspected of being a cancer therapeutic, growing said transformed eukaryotic host cell in the absence of said compound, determining the rate of growth of said host cell in the presence of said compound and the rate of growth of said host cell in the absence of said compound and comparing the growth rate of said host cells, wherein a slower rate of growth of said host cell in the presence of said compound is indicative of a cancer therapeutic.

The court held all but claim 20 invalid under 35 U.S.C. 101 as directed to abstract ideas and therefore not patent eligible.  The court found that claims 1 and 2 recited nothing more than abstract mental steps necessary to compare two different nucleotide sequences.  The court summarized the claims as “look at the first position in a first sequence; determine the nucleotide sequence at that first position; look at the first position in a second sequence; determine the nucleotide sequence at that first position; determine if the nucleotide at the first position in the first sequence and the first position in the second sequence are the same or different, wherein the latter indicates an alteration; and repeat for the next position.” Myriad, at 50.

The court stated that while the application of a formula or abstract idea in a process may describe patentable subject matter, the claims at issue did not describe applying the step of comparing two nucleotide sequences in a process, rather the comparison was the process.

The court also briefly distinguished Prometheus Labs., Inc. v. Mayo Collaborative Servs., 628 F.3d 1347, 1350 (Fed. Cir. 2010), cert. granted 2011 WL 973139 (June 20, 2011), which upheld a claim including the steps of  “(a) “administering” a thiopurine drug to a subject, and/or (b) “determining” the drug’s metabolites levels in the subject, wherein the measured metabolite levels are compared with predetermined levels to optimize drug dosage.” Myriad, at 52.  The court stated that in addition to the “administering” step being transformative, that the “determining step” was both transformative and central to the purpose of the claims.  Because the determining step could not be determined merely by inspection, the determining step required a transformation, i.e. the metabolites must be extracted and analyzed from a bodily sample.  The court held that the method claims in the instant case were capable of being performed by mere inspection and thus were not transformative. Id.

The court did hold that claim 20 was valid stating: “the claim recites a method that comprises the steps of (1) ‘growing’ host cells transformed with an altered BRCA1 gene in the presence or absence of a potential cancer therapeutic, (2) ‘determining’ the growth rate of the host cells with or without the potential therapeutic, and (3) ‘comparing’ the growth rate of the host cells. The claim thus includes more than the abstract mental step of looking at two numbers and ‘comparing’ two host cells’ growth rates. The claim includes the steps of ‘growing’ transformed cells in the presence or absence of a potential cancer therapeutic, an inherently transformative step involving the manipulation of the cells and their growth medium. The claim also includes the step of ‘determining’ the cells’ growth rates, a step that also necessarily involves physical manipulation of the cells.” Id. at 53.

The impact of this case is likely to be important to software patents as many such patents commonly contain method claims which rely on analyzing and comparing data.  To avoid invalidation after Myriad, practitioners’ should consider adding transformative limitations which attempt to remove the method from physically being able to be performed by a human mind.  For example, practitioners’ may consider adding limitations relating to the computer system as a whole, such as using calculated information to control a peripheral device or another computing device; transforming data from one form to another using a processor; receiving input (preferably reciting the source of the input – e.g. a network, a keyboard, or the like); or any other tangible connection that ties into the software method.  This will leave an argument that the method cannot be performed simply by a human mind and that some tangible step (such as physically sending electrons to a device) is necessary.  Such a connection should be as substantial as practically possible as a court may find such connections to be merely post-solution activity or data gathering steps which will not save the claim.  However, at least it leaves the litigators an argument for why the claim is not so abstract as to be unpatentable.

Bottom line, practitioners’ may be better off with a slightly narrower claim with more specific limitations that concretely tie the method to a particular machine, rather than a broad method claim which could literally be performed by a human mind.

Part II the Gene Fragment Claims

The gene fragment claims are only applicable to computer software patents in the sense that the arguments used by the Federal Circuit to uphold the validity of Myriad’s gene fragment claims undercut the very reasoning that the Federal Circuit used weeks later to invalidate a patent drawn to a machine readable medium in Cybersource Corp. v. Retail Decisions, Inc., App. No. 2009-1358 (Fed. Cir. Aug 16, 2011).  In Cybersource, the court addressed both method and machine-readable medium claims.  The court first held that the method claims were invalid under 35 U.S.C. 101 as drawn to an abstract idea.  Continuing with the machine-readable medium claim, the court held that “[r]egardless of what statutory category . . . a claim’s language is crafted to literally invoke, we look to the underlying invention for patent-eligibility purposes.  Here, it is clear that the invention underlying both claims 2 and 3 is a method . . . “ Id. at 17.  So in essence, the court ignored the structural limitations that are inherent when discussing a machine readable medium (e.g. the physical re-arrangement of electrons on a medium) to look at what the machine-readable medium does as opposed to what it is.

The Myriad panel had three separate opinions.  The opinion from Judge Lourie took an approach that is inconsistent with the Cybersource court to save the gene fragment claims.  Namely, Judge Lourie ignored the functional similarities between the function of an isolated gene fragment and the function of that same gene fragment as a part of the full gene in the human body and focused instead on the structural differences between the gene in the human body and the patented fragment.  Judge Lourie found that in nature, isolated DNA fragments are covalently bonded to other such molecules.  When cleaved to form the claimed fragment, those bonds are broken forming an isolated DNA molecule that is a “distinct” entity. Myriad, 43-4.  Dismissing arguments from the Plaintiffs’ that despite any change in the chemical makeup of this molecule, the serve the same purpose – they both store the same set of information (e.g. to encode for protein production), the court stated: “[w]e disagree, as it is the distinctive nature of DNA molecules as isolated compositions of matter that determines their patent eligibility rather than their physiological use or benefit.”

Judge Lourie here is clearly going to great lengths to uphold the patentability of the gene fragment, going into extreme detail regarding slight changes in chemical composition (one wonders why these small changes aren’t an obvious variation of the gene found in nature – but I digress) and ignoring altogether the function and purpose of the gene fragments.  This is contrary to the court’s reasoning in the Cybersource case where they only looked at the function of the product as opposed to its structure.

Judge Moore’s opinion further highlights the structural differences between natural DNA and the patented fragments, but qualifies it saying that the structural differences are not enough without some new utility.  And what was Judge Moore’s unique utility for the gene fragments? Nothing more than the fragment’s diagnostic abilities, something Judge Moore points out, cannot be inherently done by the body.  However, as with the chemical structure, this seems to be a thin distinction, and certainly the machine readable medium in Cybersource achieves some “new” utility – namely it causes a computer to perform the method steps.

In some ways the structure of the machine readable medium claim may be to blame for the Cybersource decision as it is a hybrid claim type in that it recites both structure and method limitations.  However, this is by necessity, as the actual structure of the machine readable medium is a series of electrons arranged in a certain fashion and such an arrangement is indescribable. Rather than claim a machine-readable medium comprising: one, zero, one, one, zero, one (etc…), the claim is written in terms of its result – i.e. what the machine-readable medium does, rather than what it is.  This may give the false impression that a machine-readable medium claim is a method claim wrapped in a product shell.  Perhaps if practitioners’ began describing how the machine readable medium is created (e.g. including the instructions AND including compiling steps, etc…) rather than simply what it does, these claims would find a more favorable reception.

Regardless, it is clear that the court is viewing these two technologies differently and appears to be applying a more lenient standard to biotechnology cases than computer science cases.  It will be interesting to see how this plays out as Cybersource is likely headed to further review.