[00:00:00] Speaker 00:
Yes.

[00:00:07] Speaker 00:
Mr. Bergstrom?

[00:00:11] Speaker 00:
Yes.

[00:00:11] Speaker 00:
May it please the court.

[00:00:12] Speaker 00:
My name is Bob Bergstrom.

[00:00:13] Speaker 00:
I'm a patent attorney in Seattle, Washington.

[00:00:16] Speaker 00:
And I actually wrote this application.

[00:00:18] Speaker 00:
I'm not a litigator, as you may have asked or gained from the briefing.

[00:00:22] Speaker 02:
Well, you hold good standing before the Washington State Bar.

[00:00:25] Speaker 02:
Yes, of course.

[00:00:28] Speaker 02:
At page 47 of the blueberry, you say some of the board statements are clearly and obviously false.

[00:00:38] Speaker 02:
And as a member of the Washington State Bar, you know under Washington State Rules of Professional Conduct 8.3 that a lawyer who knows that another lawyer has committed a violation

[00:00:54] Speaker 02:
the raises of substantial question as to that lawyer's honesty should inform the appropriate professional authority.

[00:01:03] Speaker 02:
Have you filed a complaint for the examiner's misconduct in making a statement that was clearly and obviously false?

[00:01:12] Speaker 00:
No, I haven't.

[00:01:13] Speaker 02:
Isn't it your obligation to do so?

[00:01:17] Speaker 00:
Well, I think it's often the case that an attorney may misread

[00:01:24] Speaker 00:
document or not understand the facts.

[00:01:27] Speaker 00:
But you accuse your attorney of dishonesty.

[00:01:31] Speaker 00:
I don't think so.

[00:01:33] Speaker 00:
I think a statement can be false and can be made honestly in error.

[00:01:39] Speaker 00:
You may proceed.

[00:01:40] Speaker 00:
You meant to say it was incorrect.

[00:01:43] Speaker 00:
Yes.

[00:01:44] Speaker 00:
To me, false and incorrect are synonymous.

[00:01:50] Speaker 00:
Be careful.

[00:01:54] Speaker 00:
So the main thrust of my arguments throughout the appeals in this case is that the USPTO did not provide any evidence for the 35 USC section 101 rejections.

[00:02:12] Speaker 02:
I pointed that out in the... You didn't invent orthogonal arrays, correct?

[00:02:20] Speaker 00:
No.

[00:02:20] Speaker 02:
Or their use on computers?

[00:02:23] Speaker 00:
No, there were other, in fact, the examiner cited some prior art where they used orthogonal arrays to generate experiment designs on a computer.

[00:02:38] Speaker 00:
However, there's a lot of differences between Webtrend's product and the pre-existing ones.

[00:02:44] Speaker 03:
Do the claims here claim the

[00:02:49] Speaker 03:
creation of new orthogonal arrays to add to the library from which a user might select?

[00:02:57] Speaker 00:
The independent claim does not.

[00:03:00] Speaker 00:
I can't recall now if one of the dependent claims does.

[00:03:05] Speaker 03:
We do describe that in the specification.

[00:03:09] Speaker 03:
But those have not been argued separately, right?

[00:03:12] Speaker 03:
No.

[00:03:12] Speaker 03:
So what we have here is a

[00:03:19] Speaker 03:
library of pre-existing orthogonal arrays, which are commonly used for test design.

[00:03:30] Speaker 03:
And the claim here focuses on a user deciding what criteria to use to select from this library.

[00:03:42] Speaker 00:
Not exactly.

[00:03:44] Speaker 00:
It turns out, and I describe this in the application,

[00:03:47] Speaker 00:
If you look at the entire experiment design space, it's enormous.

[00:03:52] Speaker 00:
There's only 117,000 known orthogonal arrays.

[00:03:56] Speaker 00:
There's billions of possible experiments.

[00:03:59] Speaker 00:
So it's almost impossible to find an orthogonal array that matches your design parameters.

[00:04:06] Speaker 00:
So what you do is you try to find an orthogonal array that you can modify in order to produce the experiment design that you specify with the design parameters.

[00:04:17] Speaker 00:
and it's a computationally difficult problem and by that I mean there's no known algorithm for doing this efficiently.

[00:04:26] Speaker 00:
It's a combinatorially intractable problem.

[00:04:33] Speaker 00:
So what WebTrans did is they've used a lot of techniques and most of these involve the templates that are associated with the orthogonal arrays in order to

[00:04:47] Speaker 00:
to record a lot of data prior to searching for matching orthogonal rays to facilitate the search.

[00:04:57] Speaker 00:
The template also includes a signature that allows you to rapidly compare the experiment design parameters to the experiment designs represented by the templates.

[00:05:13] Speaker 00:
In the process of actually generating these templates, which is all described in the application, they used a lot of heuristics so that you don't do a combinatorial cross-product.

[00:05:22] Speaker 00:
You don't generate all possible templates, but only a promising subset so that, again, the search is efficient.

[00:05:31] Speaker 00:
Prior to web trends, it could take hours, days, enormous amount of time to try to find an optimal experiment design.

[00:05:43] Speaker 00:
based on orthogonal arrays.

[00:05:44] Speaker 00:
WebTrends is able to do it in about a second.

[00:05:47] Speaker 00:
So it actually facilitated use of orthogonal arrays for experiment design in practical situations, including web testing experiments where things are done photographically.

[00:06:00] Speaker 03:
Can you help clarify for me which language in the claims addresses the question of finding

[00:06:13] Speaker 03:
the best arrays that can be modified for your purposes.

[00:06:17] Speaker 00:
So the independent claim sort of follows the initial flow diagram in the application.

[00:06:27] Speaker 00:
So identifying matching orthogonal arrays in a database of orthogonal arrays.

[00:06:32] Speaker 00:
And matching really means matching?

[00:06:35] Speaker 00:
No.

[00:06:37] Speaker 00:
This is described in a dependent claim.

[00:06:39] Speaker 00:
I believe it's claim two or three.

[00:06:41] Speaker 00:
Matching means finding orthogonal arrays that would be suitable to modify in order to generate the experiment design you're looking for.

[00:06:50] Speaker 00:
So they don't match in the sense that the orthogonal array itself is not the experiment design you're seeking, but it can be modified in order to do that.

[00:07:06] Speaker 02:
If we go to step two, what

[00:07:10] Speaker 02:
your proposed inventive concept.

[00:07:14] Speaker 02:
Is it the templates?

[00:07:16] Speaker 00:
Yes.

[00:07:17] Speaker 00:
I mean there's quite a few inventive concepts here but in drafting the claims I started out with a reasonably broad claim and most of the innovation does reside in the templates.

[00:07:29] Speaker 02:
Were they used in the prior art?

[00:07:31] Speaker 02:
No.

[00:07:33] Speaker 00:
No, in fact the PTAB found that the 103 rejections were invalid

[00:07:39] Speaker 00:
They identified no use of templates anywhere in the prior art.

[00:07:45] Speaker 00:
So they are unique from everything I know.

[00:07:51] Speaker 02:
Is there anything else?

[00:07:53] Speaker 02:
I'm trying to drill in.

[00:07:55] Speaker 02:
Other than the templates, is there anything else that you say constitutes the inventive concept?

[00:08:09] Speaker 00:
Well, so templates are a complex data structure.

[00:08:13] Speaker 00:
They include a bunch of different tables and data.

[00:08:18] Speaker 00:
And the signature within the template is highly critical for the efficiency of this method.

[00:08:25] Speaker 00:
And the signature allows one to rapidly compare your design parameters to an experiment design encoded within a template so that you can quickly search.

[00:08:37] Speaker 00:
Otherwise, the searching is very time-consuming.

[00:08:43] Speaker 00:
There can be literally thousands of different possible modifications of any given orthogonal array.

[00:08:50] Speaker 00:
So this technique increases the efficiency to allow you to find near-optimal experiment designs in under a second versus days or weeks

[00:09:07] Speaker 00:
you know, even longer periods of time.

[00:09:10] Speaker 00:
So it very much increases the efficiency of the computational approach to generating experiment designs.

[00:09:21] Speaker 02:
Do you want to reserve the balance of your time?

[00:09:23] Speaker 00:
Yes, I would.

[00:09:34] Speaker 02:
Is it Mr. Racila?

[00:09:36] Speaker 01:
Racila, that's correct.

[00:09:48] Speaker 01:
May it please the court, Brian Rossello for the appellee.

[00:09:52] Speaker 01:
As the examiner in the board recognized and concluded, Webtrend's claims are broadly directed to abstract ideas and the application of abstract ideas on generic computers.

[00:10:02] Speaker 01:
This renders the claims patent-ineligible under Section 101.

[00:10:05] Speaker 02:
What would an appellant have to add to claim one to bring it within the realm of patent-ineligible subject matter?

[00:10:13] Speaker 01:
He would have to.

[00:10:16] Speaker 01:
There's probably not much that he could do to save it under step one, because the claim is directed to the abstract idea of identifying and collecting information.

[00:10:29] Speaker 01:
We can do that with pencil and paper.

[00:10:31] Speaker 01:
Exactly.

[00:10:33] Speaker 01:
So then he would have to do something under step two.

[00:10:34] Speaker 01:
So he would have to identify something that was not conventional, well understood, or routine.

[00:10:43] Speaker 02:
Well, he says the examiner erred in his understanding of template.

[00:10:48] Speaker 02:
And counsel says that's the inventive concept.

[00:10:53] Speaker 01:
Well, I think what the board did here was the correct analysis.

[00:10:58] Speaker 01:
The board first found under step one that it was directed to abstract ideas.

[00:11:02] Speaker 01:
Then under step two, it looked to see if any of the additional elements, whether together as an ordered combination or individually,

[00:11:09] Speaker 01:
provided anything beyond, any inventive concept beyond those abstract ideas.

[00:11:15] Speaker 01:
And the board looked to the claim language itself, and the additional elements were computer system and the stored program.

[00:11:23] Speaker 01:
That computer system and the stored program, there's no information in the claims about any specificity of those computers.

[00:11:31] Speaker 01:
And then it looked to the specification.

[00:11:33] Speaker 01:
In the specification, there's an acknowledgment that the identification steps

[00:11:38] Speaker 01:
And the assignment steps that are recited in claim one could be done on various types of computer systems.

[00:11:45] Speaker 01:
So in acknowledging that it could be done on generic computer.

[00:11:50] Speaker 01:
So looking to the previous case law of this court, because the claims are directed to abstract ideas and basically just tells the practitioner to apply them on a generic computer,

[00:12:04] Speaker 01:
That cannot save it from the ambit of being an abstract idea and not patent eligible under 101.

[00:12:12] Speaker 01:
With respect to the argument about the templates, again, the language of claim one is very broad.

[00:12:19] Speaker 01:
And it just says identifying templates and assigning factors and interactions to columns in the templates.

[00:12:29] Speaker 01:
Again, this court has found that collection and analysis of information

[00:12:33] Speaker 01:
is directed to abstract ideas.

[00:12:35] Speaker 01:
And there's no detail on.

[00:12:38] Speaker 01:
So we'll go back to my question.

[00:12:41] Speaker 01:
What could he have done?

[00:12:45] Speaker 01:
He would have to, perhaps, looking at some of the other case law before this court, identify some kind of interactive display or something that was

[00:13:01] Speaker 01:
not conventional, he could have claimed perhaps if there was a non-conventional arrangement of the memory and computer system and program, he could have claimed that.

[00:13:19] Speaker 01:
But I would submit that the specification just doesn't provide support for that.

[00:13:23] Speaker 03:
What if he identified a new way

[00:13:31] Speaker 03:
of making the selection of a orthogonal array that, using the introduction of some sort of measure, a distance measure, enabled one to find something that one would never have been able to find before.

[00:13:52] Speaker 01:
I think we would have to go through the analysis to see whether or not that process was something that could be done

[00:13:59] Speaker 01:
by human on pen and paper or whether it was a mental process and how that was applied on the computer.

[00:14:07] Speaker 01:
If it's just generically applied on a computer.

[00:14:10] Speaker 03:
Well, a computer can do anything.

[00:14:11] Speaker 03:
So the entire innovation here is actually doing various kinds of data analysis and probability analyses and introducing a measure of distance so that you want to get

[00:14:27] Speaker 03:
in some mathematical sense, the closest thing.

[00:14:31] Speaker 03:
And nobody had ever been able to make the selection process that efficient before.

[00:14:46] Speaker 03:
Would that count?

[00:14:48] Speaker 03:
It's like creating a new search tool.

[00:14:54] Speaker 01:
think so under that hypothesis, because just merely claiming improved speed or efficiency is not enough to take it out of the ambit of being an abstract idea.

[00:15:05] Speaker 03:
Why not?

[00:15:06] Speaker 03:
Because, again, in this case, running an old algorithm really, really fast might well be different from running a new algorithm that enables you to find what would otherwise take six billion years to find.

[00:15:24] Speaker 01:
But the Supreme Court in Alice has told us that just taking processes that could be done.

[00:15:30] Speaker 03:
Nobody has ever run, in my hypothesis, this new algorithm.

[00:15:33] Speaker 03:
That's what makes it new.

[00:15:36] Speaker 01:
The Supreme Court has also found that a new algorithm run on a generic computer would still be patent-ineligible under the abstract idea.

[00:15:46] Speaker 01:
Really?

[00:15:47] Speaker 01:
Where?

[00:15:48] Speaker 01:
In Alice and in Mayo.

[00:15:49] Speaker 03:
The court said that the application... Alice didn't involve algorithms.

[00:15:53] Speaker 01:
I'm sorry, in Mayo, the court said that... In Mayo didn't involve algorithms.

[00:15:57] Speaker 03:
It involved a new discovery of what the body does with particular substances.

[00:16:02] Speaker 01:
Of a mathematical... Mayo?

[00:16:06] Speaker 01:
The use of a mathematical... Well, let me circle back to the second point of your question, which was the

[00:16:19] Speaker 01:
So the application of a new algorithm, if that algorithm is based on mathematical principles that could be done as a mental process, but the computer allows it to be done more efficiently or quicker than

[00:16:50] Speaker 01:
then if that's an abstract idea, increasing the speed of it doesn't take something that's already an abstract idea and take it out of that ambit.

[00:17:02] Speaker 01:
Now, part of your hypothesis was this idea about a distance vector, which may be some kind of new information or something that might be

[00:17:19] Speaker 01:
patentable as depending on how it's claimed by the applicant.

[00:17:26] Speaker 01:
Because that might be something that has never been discovered before or something that a human being would be unable to do using algorithms.

[00:17:56] Speaker 01:
If there's no further questions.

[00:18:17] Speaker 00:
So my main complaint again is no evidence has been provided by the USPTO to support the

[00:18:25] Speaker 00:
Alice test or the 35 USC section 101 rejections as a whole.

[00:18:30] Speaker 00:
I can't find any evidence in the examiners for conclusory statements and I find no evidence in the board decision.

[00:18:37] Speaker 00:
They're simply conclusory statements.

[00:18:40] Speaker 00:
It's easy to say that something is an abstract idea or that the fundamental building block in this case is the use of... Put aside fundamental building block.

[00:18:52] Speaker 03:
I mean, I took it that the fairly simple point here is that if you look at claim one, the language of template is so broad that it doesn't require any evidence, because it's not a point about facts in the world, to say that is at a level of generality that is definitionally abstract.

[00:19:17] Speaker 03:
And that's the only locus of innovation in this claim.

[00:19:21] Speaker 00:
I still maintain under the APA that some kind of evidence needs to be cited for the proposition.

[00:19:29] Speaker 03:
It's not a fact question.

[00:19:31] Speaker 03:
Why would you need evidence?

[00:19:33] Speaker 03:
What you need evidence about is states of fact in the world, like whether something that is not abstract, a particular computer, was routine conventional.

[00:19:48] Speaker 03:
But to characterize a template,

[00:19:50] Speaker 03:
where it's not defined except at this extremely high level of generality.

[00:19:57] Speaker 03:
That's not a fact about the world.

[00:19:59] Speaker 03:
That's about the meaning of the term template.

[00:20:03] Speaker 00:
But bear in mind that all of the aspects of the template are claimed in the dependent claims.

[00:20:09] Speaker 03:
But you didn't argue the dependent claim separately.

[00:20:11] Speaker 00:
I have argued throughout the briefing that all of the claims were summarily rejected without evidence.

[00:20:18] Speaker 00:
And there's no indication

[00:20:20] Speaker 00:
that any of the details in the dependent claims or even the template itself was considered in the decision.

[00:20:26] Speaker 00:
They talk about the fact that we... They were rejected without evidence.

[00:20:31] Speaker 02:
What evidence exactly?

[00:20:32] Speaker 00:
Well, for example, the identified abstract concept here is the use of something, mathematical use of orthogonal arrays for experiment design.

[00:20:45] Speaker 00:
Where is the evidence that this is in any way mathematical?

[00:20:48] Speaker 00:
We don't agree to that.

[00:20:49] Speaker 03:
Forget about the term mathematical.

[00:20:51] Speaker 03:
It's just selecting information that's sitting in a library.

[00:20:55] Speaker 03:
What did you say, 117,000 members in this library?

[00:20:58] Speaker 03:
Currently.

[00:20:59] Speaker 03:
Selecting the ones you want.

[00:21:02] Speaker 03:
What more is in this claim?

[00:21:03] Speaker 03:
Except, you know, doing it extremely fast because the library is fairly big and you may have a complicated set of selection criteria.

[00:21:10] Speaker 00:
The thing that allows me to increase the efficiency so much is the fact

[00:21:16] Speaker 00:
that I associate these templates with the orthogonal array in the database.

[00:21:20] Speaker 00:
This isn't so unlike the ENFISH case where there was a self-referential table that was added to the database and that increased the efficiency of the system.

[00:21:31] Speaker 00:
It's the same thing.

[00:21:33] Speaker 00:
That template includes enough information to make my search incredibly fast versus the pre-existing attempts to use orthogonal arrays computationally.

[00:21:50] Speaker 00:
Do you have anything further, counsel?

[00:21:54] Speaker 00:
It seems to me, again from reading the case law, that the Alice two-part test is the prima facie case for asserting a judicial exception.

[00:22:11] Speaker 00:
And prima facie case to me means there's an evidentiary burden that the examiner must meet.

[00:22:20] Speaker 00:
I don't see, again, any evidence cited anywhere in this rejection.

[00:22:27] Speaker 00:
So even if you say that the temple is a data structure, but if you say it's just information or whatever, I still think you need to cite some kind of evidence for these propositions.