By Michael Borella --
In August, 2009, Allvoice sued Microsoft in the Western District of Washington, alleging infringement of its U.S. Patent No. 5,799,273. In December, 2013, the District Court granted Microsoft's motion for summary judgment finding that Microsoft did not infringe the asserted claims, and that claims 60-68 of the patent were invalid under 35 U.S.C. § 101. Allvoice appealed. The Federal Circuit panel of judges Prost, Dyk, and O'Malley heard arguments, and affirmed the District Court.
Of particular interest is the reasoning behind the finding of invalidity in light of the claim language at issue. As a representative claim of those held invalid, claim 60 recites:
A universal speech-recognition interface that enables operative coupling of a speech-recognition engine to at least any one of a plurality of different computer-related applications, the universal speech-recognition interface comprising:
input means for receiving speech-recognition data including recognised words;
output means for outputting the recognised words into at least any one of the plurality of different computer-related applications to allow processing of the recognised words as input text; and
audio playback means for playing audio data associated with the recognised words.
In other words, Allvoice's invention, as recited by claims 60-68, is directed to a form of application programming interface (API) that allows applications running on a computer to receive the output of a speech recognition program and "to link the relationship between the output data and the audio data to allow the audio data to be played back for any output data which has been dictated."
After discussing the parameters of § 101 in a broad sense, the Court focused in on the four statutory categories listed therein (machines, processes, articles of manufacture, and compositions of matter). Rather than analyzing whether the claims fell into one of the judicial exceptions to these categories (e.g., abstract ideas), the Court considered whether the claimed invention even fit into any of the categories at all.
Noting that "subject matter that falls outside the four statutory categories of § 101 . . . is not patent eligible," the Court explained that "except for process claims, the eligible subject matter must exist in some physical or tangible form." For instance, a machine must be "a concrete thing, consisting of parts, or of certain devices and combination of devices," an article of manufacture must be "a tangible article that is given a new form, quality, property, or combination through man-made or artificial means," and a composition of matter "requires the combination of two or more substances and includes all composite articles."
When before the Court, Allvoice explained that the interface recited by claim 60 consisted of software instructions. The Court did not find this point sufficient to result in the claims meeting the requirements of § 101, because while "[s]oftware may be patent eligible . . . when a claim is not directed towards a process, the subject matter must exist in tangible form." Claims 60-68, in the Court's view, were "software instructions without any hardware limitations" and therefore intangible. Allvoice's proposed construction of the claims' means-plus-function limitations did not involve a tangible form either.
Allvoice contended that "software must necessarily be in a machine readable, physical state in order to exist," and therefore is patent-eligible by its very nature. However, citing Digitech Image Techs., LLC v. Elecs. for Imaging, Inc., the Court observed that "instructions, data, or information alone, absent a tangible medium, is not a manufacture," and thus refused to import a tangible medium into Allvoice's claim language.
The four categories of statutory subject matter cover most anything that one can imagine. For computer-implemented inventions, however, claims directed to pure signals or disembodied data are not within the ambit of § 101. This case serves are a reminder to claim drafters that claiming software as anything but a process can be a tricky endeavor. After all, software instructions are indeed just another form of data, and without a medium to be stored on, lack physical structure.
But, this case also begs the question of what Allvoice could have done in order to mitigate this problem. For instance, if the preamble to claim 60 had read "an article of manufacture including a universal speech-recognition interface . . . ," would the outcome be different? Or, would Allvoice's claims be valid if they had been written in Beauregard style, such as "a non-transitory computer readable medium including computer instructions that, when executed by a processor, cause the processor to perform operations of a universal speech-recognition interface . . . ." Deviation from well-known and well-accepted preambles remains risky.
Allvoice Developments US, LLC v. Microsoft Corp. (Fed. Cir. 2015)
Nonprecedential disposition
Panel: Chief Judge Prost and Circuit Judges Dyk nd O'Malley
Opinion by Circuit Judge O'Malley
The sentence "After all, software instructions are indeed just another form of data, and without a medium to be stored on, lack physical structure." goes too far, and like the court itself, misses what software is.
But setting that aside for the moment, let's look at another impact here: copyright.
If this particular definition were to hold, then a gaping hole (the size of the Grand Canyon) has just been blown through copyright protection. All one has to do is copy the "signal" and one is free and clear. Since copyright also extends only to that which is captured in a tangible media, anything intangible is not - and cannot - be protected.
Does this extend to copying the signal of movies and music?
Pandora called - she wants her box back.
Posted by: Skeptical | May 26, 2015 at 06:24 AM
US readers should bear in mind that In re Nuijten and Digitech do not apply in Europe, and that the EPO will in appropriate cases grant claims to signals. However, because of the severe tests for added subject matter and intermediate generalization under a.123(2) EPC it is essential to include appropriate statements of invention in the specification if such claims are to be pursued here.
That being said if we analyze claim 60, apart from a broad statement of the goal of providing a universal speech-recognition interface without an indication of any new and surprising or ingenious technical features that make this possible, there are specified input means, output means and audio playback means. These features are present in the now fairly venerable iMAC on which this comment is being typed, were present in the Epson HX20 portable computer which was the first that I owned in the early 1980's and were also present in the Sinclair Spectrum that I bought for my children at that time. It is a little surprising that this claim was considered litigating at all, still less as far as the Federal Circuit.
Posted by: Paul Cole | May 26, 2015 at 10:53 PM
If software is not data, then what, pray tell, is it?
Posted by: Mike Borella | May 27, 2015 at 06:02 AM
@ Mike
From my recollection software is an ordered assembly of instructions for be supplied to a computer processor. The processor operates on data in accordance with the instructions. As concepts these are quite distinct, as any software engineer will tell you.
Posted by: Paul Cole | May 27, 2015 at 06:50 AM
Software instructions are data - they can be read, written, modified and deleted like any other form of data. They may have a functional difference from other forms of data, in that they can be used to tell a processor what to do, but they do not have a structural difference.
One of the first programming assignments I had at university was to parse through an executable file and look for certain types of information. The software code was the data to another program.
Posted by: Mike Borella | May 27, 2015 at 10:21 AM
Mike,
Careful there with your "definition" of data, as along that logical abstraction, everything is "data" (you fall into the Tegmark camp).
Posted by: Skeptical | May 28, 2015 at 06:13 AM
"misses what software is"
if the claim recited "what software is", there would not have been a problem. the claim recites "what software does" (function laden language) and does not fit into the four statutory categories. my grandmother can come up with dozens of items to magically do stuff she wants done, but that's not patentable. squeeze the magic out of the claim and you'll make progress.
Posted by: Gene | May 28, 2015 at 08:28 AM
"If software is not data, then what, pray tell, is it?"
at what point in time of the existence of "software" do you want a definition? it's not patentable when it exists initially in the mind of the programmer. it's not patentable when written down on paper. it's not patentable when stored in memory. it's not patentable as an ordered assembly of instructions. it's patentable when compiled because it is then a MACHINE--an electric circuit where a higher voltage at one node causes electrons to travel toward a node at a lower voltage. does one claim it that way? not if the prior art does not require such narrow language. perhaps my generation of electrical engineers is obsolete--where we had to write programs and compile logic circuits. a processor is not a black box to my classmates. i understand that college coursework is so specialized these days that graduates majoring in software need not understand hardware details, and vice versa. claiming hardware with high level functional language will not get you a patent, a disclosed and claimed novel and non-obvious algorithm will.
Posted by: Gene | May 28, 2015 at 08:47 AM
The point (well, my point anyway) is that the CAFC made a reasonable determination that the claim covered software as data, bereft of a tangible medium, and is therefore unpatentable. This is in line with precedent and makes technical sense, as opposed to what the Supremes did with Alice.
Posted by: Mike Borella | May 28, 2015 at 09:36 AM
With respect to how Claims 60-68 might have been written, note the other claims in the `273 patent that are directed to various methods and "An editor work station" and were held indefinite and not infringed in this suit. Claims 73 and 74, directed to a "computer usable medium," were not asserted against Microsoft, though they and Claims 60 - 68 were successfully asserted against Dragon in AllVoice Computing PLC v. Nuance Communications, Inc., No. 06-1440 (Fed. Cir. Oct. 12, 2007). In that earlier case, the Federal Circuit reversed district court rulings that these claims were indefinite or were invalid for failure to disclose the best mode of practicing them.
Also there are two related AllVoice patents, 5,857,099 and 6,961,700, with further sets of claims including "computer usable medium" and method claims. So there are a variety of claims of various types, some of which have been held invalid and some not.
I confess that I do not understand this technology and am not inclined to study the claims in these patents to assess their differences and similarities to Claims 60 - 68 of the '273 patent. It would be interesting to know just how the infringement cases were laid out in both AllVoice lawsuits. It would be interesting to know just how the district court held most of the asserted apparatus and method claims to be indefinite and why Allvoice did not appeal that decision.
Returning to the AllVoice Computing PLC v. Nuance Communications, Inc. decision, I see that the panel found corresponding "structure" in the specification to support all three "means" expressions of Claim 60. But, in pursuing Microsoft, Allvoice conceded that these "means" limitations were limited to software instructions without any hardware limitations. Why was AllVoice forced to make this concession?
Would someone address these two cases?
Posted by: Joe Breimayer | May 28, 2015 at 12:50 PM
What I find interesting is the lack of focus on the use of means-plus-function claim language. As you know, MPF claims inherently limit the claim element to structural embodiments (disclosed in the specification, and equivalents, etc.).
Thus, if the holding was that the MPF claim element didn't have corresponding structure in the specification, it would seem to be more appropriate to invalidate as indefinite under 35USC112.
What an interesting time for patent law!
Posted by: Henry | May 28, 2015 at 01:56 PM
That Claim 60 was a disaster from the git-go.
1. The preamble should have explicitly referenced statutory matter -- i.e. "an interface device," which is what is described in the spec.
2. The novel/non-obvious characteristic should not be in the preamble, it should be cited as either an element or as a limitation of an element. I would have used MPF here:
means for enabling operative coupling of a speech-recognition engine to at least any one of a plurality of different computer-related applications;
This MPF is adequately supported by flo charts in the spec so as to convert the algorithm into "structure" ala the recent Eon case.
It was a disaster in the making. Didn't affect the outcome in this case, however. The CAFC's bizzare upholding the DC's dismissal of the case for lack of timely filing of a motion to amend the claims deep six'd all the litigated claims.
Posted by: Denis | May 29, 2015 at 10:27 AM
Michael:
Please compare and contrast the findings re Claims 60 - 68 of the '273 patent in
Allvoice Developments US, LLC v. Microsoft Corp. (Fed. Cir. 2015) with
AllVoice Computing PLC v. Nuance Communications, Inc., No. 06-1440 (Fed. Cir. Oct. 12, 2007)
Help us understand.
Posted by: Joe Breimayer | May 29, 2015 at 07:29 PM
In one case, the issue was under Section 101, in the other, the issue was under Section 112. Thus, the analyses and outcomes can easily be different.
Posted by: Mike Borella | June 01, 2015 at 07:17 AM