I’m not a jurisprudence guy, but as I noted in my post about In re Bilski here, I think there are some serious flaws in the way the Federal Circuit settled on the “machine or transformation” test as the only way a process patent claim can be patentable subject matter. As a recap, here is what I said:
Here’s how the court gets there:
1. We know that natural principles aren’t patentable.
2. We know it is difficult to determine what is and what is not a natural principle.
3. We know for sure that there are two ways to tell if something is not a natural principle:
A. If it is tied to a machine or
B. If it transforms the subject matter
4. Therefore, the only way for a process to be patent eligible is to be tied to a machine or tranform subject matter.
5. In order to make sure all principles are excluded, insignificant post-solution machines or transformations don’t count.
Anyone else see the logical jump from 3 to 4? I put this leap on the level of (i) if you weigh the same as a duck, you much be a witch (for you Monty Python fans) and (ii) Descartes’ proof of the existence of god (for you philosophy fans).
I want to take a look at the court’s logic (literally) and how it used precedent to support that logic. I think the result will be that the court’s test simply does not follow the precedent it cites. Of course, that doesn’t mean the court can’t extend or limit a test and see what the Supreme Court does with it, but to say that the test follows as a matter of fidelity to precedent seems an unlikely claim due to three errors: semantic, precedential, and jurisprudential.
Analysis follows after the jump.
The court does a good job of making its argument, and to avoid any claim that I am distorting the argument, I will quote the opinion directly where possible, though I do reorder quotes into their logical sequence. Thus, this post may run a bit long, but it’s easy reading.
The court starts with the basic proposition:
A claimed process is surely patent-eligible under § 101 if: (1) it is tied to a particular machine or apparatus, or (2) it transforms a particular article into a different state or thing.
Fair enough. I won’t quibble with this basic reading for the purposes of this analysis. Of course, even if a process is surely patentable in one of these ways does not mean there is no other such way.
The court then attempts to limit the rule to only this proposition. It starts with Benson:
Applicants and several amici have argued that the Supreme Court did not intend the machine-or-transformation test to be the sole test governing § 101 analyses. As already noted, however, the Court explicitly stated in Benson that “[t]ransformation and reduction of an article ‘to a different state or thing’ is the clue to the patentability of a process claim that does not include particular machines.
The court later notes:
We recognize, however, that the Court was initially equivocal in first putting forward this test in Benson. As the Applicants and several amici point out, the Court there stated:
It is argued that a process patent must either be tied to a particular machine or apparatus or must operate to change articles or materials to a ‘different state or thing.’ We do not hold that no process patent could ever qualify if it did not meet the requirements of our prior precedents.
So far, so good. M or T is the clue (heavy emphasis by court) but explicitly not the only test. The court fails to note that the Supreme Court in Benson was urged to limit the test and refused to do so. The court takes up the next case, Flook:
Flook, 437 U.S. at 589 n.9 (“An argument can be made [that the Supreme] Court has only recognized a process as within the statutory definition when it either was tied to a particular apparatus or operated to change materials to a ‘different state or thing'”
Hardly a ringing endorsement for the test – a footnote that says an argument can be made. The court goes on to say:
In Flook, the Court took note that this statement had been made in Benson but merely stated: “As in Benson, we assume that a valid process patent may issue even if it does not meet [the machine-or-transformation test].” [emphasis added in Bilski]
After Flook, it appears that the test is in the same state – one method but explicitly not the only way to get to eligibility. From a persuasive point of view, I think Flook moves me away from the test, but that’s not enough to say the analysis is wrong. Next comes Diehr:
And this caveat [the assumption from Flook] was not repeated in Diehr when the Court reaffirmed the machine-or-transformation test. See Diehr, 450 U.S. at 184 (quoting Benson, 409 U.S. at 70) (“Transformation and reduction of an article ‘to a different state or thing’ is the clue to the patentability of a process claim that does not include particular machines.”).
And also:
Diehr, 450 U.S. at 192 (holding that use of mathematical formula in process “transforming or reducing an article to a different state or thing” constitutes patent-eligible subject matter)
That’s right, a string cite. What of the “clue” language? Here is the argument, in a footnote:
We believe that the Supreme Court spoke of the machine-or-transformation test as the “clue” to patent-eligibility because the test is the tool used to determine whether a claim is drawn to a statutory “process”—the statute does not itself explicitly mention machine implementation or transformation. We do not consider the word “clue” to indicate that the machine-or-implementation test is optional or merely advisory. Rather, the Court described it as the clue, not merely “a” clue. See Benson, 409 U.S. at 70. [emphasis in Bilski]
From this, the court comes to the conclusion that M or T is the only test allowed by precedent:
Therefore, we believe our reliance on the Supreme Court’s machine-or-transformation test as the applicable test for § 101 analyses of process claims is sound.
There you have it – the Federal Circuit’s transformation of a “clue” that is not explicitly limited to a mandatory and exclusive rule.
I believe the reasoning makes three errors, which I call semantic, precedential, and jurisprudential for no reason other than I couldn’t think of better names.
Semantic – Interpreting “the clue” to mean that one and only test may be applied seems to put a lot of faith in a single article. I’ll leave it to others to debate whether “the” means “a” in textual interpretation, but not before I note that the Federal Circuit addressed a similar issue in patent claim construction earlier this year in Baldwin Graphics v. Siebert. The court there ruled that the article “a” means “one or more” and that a later singular qualifier article (“the”) does not change that. For example, if a claim says
A computer monitor comprising
a liquid crystal display panel
wherein the panel is in color
then the claim covers monitors having multiple LCD panels, notwithstanding the reference to “the” panel. Analogized to the Benson case, the Supreme Court defined “a” test, and later referred to “the” clue. The later “the” does not mean that there cannot be one or more rules.
Even disregarding linguistic definitions of “the” – something the court would call extrinsic evidence – there are at least two instances within the opinion – let’s call it intrinsic evidence – that make clear that “the” clue cannot mean the M or T test is the one and only test. First, the clue provided is not about machines or transformations – it is only about machines: “Transformation and reduction of an article ‘to a different state or thing’ is the clue to the patentability of a process claim that does not include particular machines.” This means that there must be at least two clues – transformation or particular machines. Second, M or T is not the only part of the test – there is also the part that “insignificant post-solution activity” does not count as a machine or transformation. If “the” clue really meant that this was the one and only test, then the additional requirement would not be necessary – that is an additional clue. If there are two clues, then “the” cannot mean one.
Finally, extrapolating a single rule from “the” clue makes little sense. Consider all of the following statements, all of which can be true at the same time in the decision:
- the clue is whether fundamental principle is implicated
- the clue is preemption of a natural principle
- the clue is statutory categories – if not in a category no eligibility
- the clue is application of the principle
- the clue is data relating to physical objects
These too are clues, and calling them “the clue” does not detract from the import of the others.
Precedential – I call this error “precedential” because, quite frankly, the court mis-states the precedent. A lynchpin in the court’s analysis is that Diehr failed to include the caveat present in Benson and Flook that there may be other ways to show patentable subject matter than the M or T test. Whatever one thinks of this leap (discussed in the next section), the reality is that Diehr does include that caveat.
As Judge Newman points out in dissent,
The Court explained that the presence of a mathematical formula does not preclude patentability when the structure or process is performing a function within the scope of the patent system, stating:
[W]hen a claim containing a mathematical formula implements or applies that formula in a structure or process which, when considered as a whole, is performing a function which the patent laws were designed to protect (e.g., transforming or reducing an article to a different state or thing), then the claim satisfies the requirements of §101.
This statement’s parenthetical “e.g.” is relied on by the majority for its statement that Diehr requires today’s “machine-or-transformation” test. However, this “e.g.” does not purport to state the only “function which the patent laws were designed to protect.” This “e.g.” indeed describes the process in Diehr, but it does not exclude all other processes from access to patenting.
The majority addresses this in a footnote: “…language such as the use of “e.g.” may indicate the Supreme Court’s recognition that the machine-or-transformation test might require modification in the future.” Noting that additional tests might be considered in the future seems like a caveat to me. In the context of the argument it weakens – if not guts – the claim that Diehr simply “reaffirmed” a narrow holding of Benson and deliberately ruled that M or T is the only test by leaving out a caveat.
Jurisprudential – The final error is jurisprudential, by which I mean the court’s logic legal reasoning leaves something to be desired. Even assuming that Diehr did not include an “e.g.” and could be read as eliminating the caveat that M or T is not the only test, reliance on a missing caveat seems misplaced from a legal analysis standpoint. The logical concern is obvious – just because a disclaimer is missing does not mean that all other avenues are foreclosed, but the issue goes beyond simple logic in my mind. I think it goes to the heart of how we think about holdings and dicta.
Here’s a breakdown of the three key precedents:
- Benson: The patent does not meet M or T, and is not patentable, though there may be other ways to show patentability
- Flook: The patent does not meet M or T, and is not patentable, though there may be other ways to show patentability
- Diehr: The patent does meet M or T, and is therefore patentable
The first problem is in Benson and Flook – if a patent is to be rejected, then shouldn’t every test be explored to see if one of them is met? There are three explanations for the court’s failure to do so: (1) M or T really is the only test (belied by the explicit language of the case); (2) The court knows a bad patent when it sees it – also known as “insignificant post-solution activity” (not terribly palatable, because it still ignores the possibility of other ways to acheive patentability); or (3) some other test is actually being applied (this is my theory). Because of these issues, it is difficult to ascertain the actual rule that leads to the holding of these two cases. However, the court is explicit that whatever that rule is, there may or may not be others.
The second problem goes deeper – and it is associated with Diehr. Taking Benson at its face, one may show patentable subject matter by (i) machine; (ii) transformation; (iii) maybe some other way. The Diehr Court rules that the process at issue is at least a transformation, and is thus patentable. From a holding/dicta point of view, that’s it – end of story. The applicant need not prove that the process is tied to a machine, and the court need not address it. The applicant need not prove that there is some other way to show patent eligibility, and the court need not address it. If the court did address either of these subjects, it would have been an alternative rule at best or dicta at worst. Either way, the posture of Diehr – granting a patent by meeting one of many tests – makes reliance on its silence about other rules quite strange.
Jurisprudentially, the failure to provide an alternative rule or further dicta mandates a later reading that no such alternative rules or dicta exist. And that is the real failing here – it took a lot of words to say something pretty simple – the failure to perpetuate a dicta-based caveat does not warrant an elimination of the caveat. I’ll push it further – even if Benson and Flook did not include explicit caveats but instead simple logic allowed for the possibility of other rules, then the failure of later cases to recognize that logic would still not eliminate the possibility of such rules until the court actually considered the issue and rejected a patent after applying such a rule.
So what?
That’s a good question. Reasonable minds can differ on whether the Federal Circuit got the test right or not (I don’t think so), whether machine or transformation is the only test possible (I don’t think so), and whether or not this limitation is good for the patent system (I don’t think so). I think that it would be difficult to justify the ruling as the only one justified by logic and precedent.
Posted by Michael Risch on November 3, 2008 at 08:02 AM
Comments
I didn’t mean to imply that the court should have considered our test (though that would have been nice). I agree that amicus tests are rarely considered, but the situation here was a bit different. First, for whatever reason amicus suggested tests are often considered in subject matter cases (at least recently). Second, the court actually heard argument from (I think two) amicus brief authors who proposed specific tests.
Whether or not possible test ideas came from amicus briefs, before rejecting all other tests I would want some consideration of at least a few of those test. Indeed, the “useful, concrete and tangible” test is rejected because it doesn’t meet the Supreme Court precedent. But if the precedent is not construed to require only machine or transformation, the court would have had to take a hard look at whether the useful, concrete and tangible test was an appropriate alternative, and I don’t think it wanted to do that analysis.
Posted by: Michael Risch | Nov 3, 2008 1:02:34 PM
Thanks for the response. I hadn’t realized that you wrote an amicus advocating a different rule: Given your participation, I can understand your frustration with why the court didn’t explicitly consider the pros and cons of other possible rules. (On the other hand, in my experience it is rare for courts to have such explicit discussions of alternative rules, especially ones advocated in amicus briefs.)
Posted by: Orin Kerr | Nov 3, 2008 12:49:17 PM
Thanks for the comment and the good question. I would say in this case that my objections would remain the same – primarily because the Supreme Court was asked in Benson (and I’m sure in Flook and Diehr) to limit the test and it explicitly refused to do so. I don’t think any plausible reading can then say the test is limited.
That said, what I would not have objected to (on jurisprudential grounds at least) would have been something like “The court explicitly left the question of other tests open. In the 30+ years since Benson, no one has come up with a test that we think should also apply, and so we apply only this test today.” Or something like that. Maybe even further, “And we won’t consider another one until someone convinces the Supreme Court first.”
Of course, the problem with that is that there are all sorts of other tests that might apply (we suggested one in our amicus brief) and to admit that the door is open would have required the court to actually either analyze or actively ignore those possibilities without the pretext of Supreme Court authority.
Posted by: Michael Risch | Nov 3, 2008 11:49:35 AM
Michael,
Would your objection be answered if the court had said that the precedents didn’t clearly answer what test was correct, but that the test they adopted seemed to be the most plausible reading of those precedents and therefore it would be considered correct?
Posted by: Orin Kerr | Nov 3, 2008 10:48:34 AM