concurring.
I fully agree with the majority opinion and join it. I write separately to note what I believe are additional reasons why retaining Cybor is wise.
First and foremost is that the Supreme Court has held that claim construction is a question for the court rather than the jury. Thus, for us to appear to be cutting back from that holding by giving formal deference on so-called fact-like questions, which normally would go to the jury, to the district court judge, would seem to me to be an attempt to partially retreat from the Court’s holding, which is unwise.
We have held that claim construction is a question of law, going only minimally beyond the Court’s explicit holding that it is only a question for the court. Cybor Corp. v. FAS Techs., Inc., 138 F.3d 1448, 1454-55 (Fed.Cir.1998) (en banc). The Court’s holding, including its statement that it is a “mongrel” question, does not encourage fractionation of the process, making part of it subject to de novo review and part clearly erroneous review. The Court in Markman stated that construction of terms of art should be ceded to the judge “notwithstanding its evidentiary underpinnings.” The “notwithstanding” applies to any factual as well as legal aspects of claim construction. Evidentiary underpinnings do not lean toward a clearly erroneous standard any more than they do to the jury.
Equally important, one of the purposes of Congress in creating our court was to achieve uniformity in the patent law. Consistent with that goal should be uniformity of interpretation in construction of patent claims. It is not rare that a patent is asserted against more than one defendant in different forums, with conflicting holdings on infringement. It would hardly promote uniformity of patent law for us to bless a claim construction in one district court, based on that court’s judging the credibility and demeanor of the expert witnesses in one case, when a different case might lead to a different result based on a different district judge’s appraisal of dif*1293ferent witnesses. We in fact might be confirming conflicting claim constructions, the antithesis of uniformity. This problem might increasingly exist in light of the AIA’s limits on the number of accused infringers that can be joined as defendants in one lawsuit, thereby creating the possibility of more lawsuits on the same patent, and more inconsistency, than existed in the past. See 35 U.S.C. § 299(a); In re EMC Corp., 677 F.3d 1351 (Fed.Cir.2012).
By our deferring to those determinations, conceding our full review of the meaning of a claim term, which should be based on the patent’s written description and prosecution history, not the witnesses, hampers our ability to interpret claims with full authority and hence to ensure uniformity.
Furthermore, claim construction is not a process that normally involves historical facts. It primarily involves reading the patent’s written description as well as the prosecution history of the patent, and this court is quite as able to do that as any district court, sometimes better. It is true that there may be questions concerning what a particular claim term meant to one skilled in the art at a particular time, but, in my view, when the trial judge is subjected to dueling experts selected for their views, choosing which of them to credit hardly amounts to historical fact-finding. To the extent that it does, the relevant inquiry should be, not what the dueling experts say, but what the inventor understood the term to mean when he or she filed the patent application containing the claim term in question. Courts should be reluctant to go beyond the written record to help answer that question. It is too subject to ex post facto thinking based on self-interest; the inventor had his chance to define his invention and should not be heard in later testimony to get another bite at the apple by redefining that language.
A realistic assessment of the problem in claim construction in litigation recognizes that the patenting process begins with an inventor and his or her attorney drafting a written description and claims to describe and specifically claim his or her invention. The claims will usually then get negotiated in the Patent Office. The issued claims are then used, pursuant to some perceived business need or desire, by a different lawyer who had no part in the drafting of the written description and claims, who then tries' in front of a lay judge to shoehorn an accused infringer into claims that usually do not fit (or else claim construction and infringement would not be at issue).
Hired “experts,” supporting the parties’ theories of infringement or non-infringement take positions that are also distinct and isolated from (and often different from) those originally taken by the inventor and attorney, who knew what the invention was and what positions were taken in the Patent Office during prosecution. Thus, the problem lies, not with lack of deference to district court interpretation of claims by the Federal Circuit, but to the multiplicity of actors contending in a competitive economy. The actors striving to deal with a patent in district court are often not those who made the invention, created the patent, and hence knew exactly what it meant. The solution does not lie in depriving the one institution charged with ensuring uniformity of part of its authority.
Much criticism has been directed at this court for allegedly ignoring all the fine work of district court judges in construing patent claims. That criticism is premised on the misperception that we do not give a district court’s claim construction any def*1294erence. That is incorrect; perhaps one even might say “clearly erroneous.”
The rubric governing issues of law unfortunately does read “no deference.” But even though we all know that the rubric governing procedural errors is the harsh “abuse of discretion” language, all judges know that a finding of a procedural error does not normally justify the term “abuse.” Similarly, the “no deference” language is simply established legal jargon for a holding that, having reviewed the record, we disagree. It has been stated in some ami-cus briefs before the court that there are truly factual issues involved in claim construction, particularly what a claim term meant to one skilled in the art at a particular time, and that such a determination should be given deference. But we should not complicate the law and change our precedent for such a situation. This court should rarely overturn a district court’s claim construction on a finding of that nature.
This appellate court, when asked to interpret the claims of a patent carefully, notes and considers how the district court construed the claims. If we disagree, it is not without a degree of informal deference. Claims are to be interpreted in light of the written description in the patent specification, and in light of the prosecution. The choice of expert witnesses by parties’ counsel, and their demeanor, do not override those basic documents. Very few scientists called as expert witnesses will lie, hence the term “credibility,” useful in more conventional fact determinations, such as whether a traffic light was red or not, should not be controlling in construing claim limitations in a patent.
Claim construction is analogous to interpretation of other legal instruments, such as contracts and legislation. Each of these determinations is for the court, not a jury, although each can be found to contain factual components (it should be noted though that, at least in patent prosecution, the intent of neither the inventor-attorney nor the patent examiner is usually at issue in claim construction). Thus, beyond what the Court has held in Markman, there is good analogical basis for considering claim construction as similar to interpretation of these other legal instruments. Moreover, to the extent that underlying considerations create the “mongrel” nature of claim construction, considering factual components to be subject to deference under a clearly erroneous rule would implicate the Seventh Amendment right to a jury trial on factual questions. Such a procedure would further threaten the uniformity that Congress intended in setting up this court as well as the Supreme Court’s ruling that claim construction is not for the jury.
What the proponents of splitting construction into legal and factual issues are in essence contending for is that some issues in patent infringement are for the judge (some claim construction issues), some are also for the judge, but are of a factual nature, ordinarily as for a jury, and some clearly for the jury, ie., infringement. However, in claim construction, simpler is better-claim construction in all its aspects for the judge, subject to review by the appellate court, with sensible reliance on the prior work of the trial judge. Creating a formal distinction between fact-sounding issues subordinate to claim construction and the ultimate claim construction is a complication that we should not foist on this court.
A further point is that ultimately it should not matter whether claim construction has a factual component to which formal deference attaches or not. If, as I believe we should, and do, give proper informal deference to the work of judges of a subordinate tribunal, then we will or *1295should affirm when affirmance is appropriate. If, on the other hand, we were to apply a more formal clearly erroneous standard, judgments of subordinate courts are still not unreviewable. If we were to find that the so-called factual component, based on our review of the intrinsic record, has been determined incorrectly, clearly we could find it to be incorrect even with a clearly erroneous standard. Thus, this is an argument that should not much matter.
Moreover, to the extent we were to overrule Cybor, or modify it, and give formal deference to district courts, but reserve the right to decide the ultimate issues of obviousness and validity as questions of law, we would be engaging in a kind of sham, giving with one hand and taking back with the other. Doing so would bow to what amounts to a cosmetic public, or judicial, exercise in order to overcome the harsh rubric of “no deference.”
To the extent that critics assert that de novo review has not achieved the goal of uniformity, I believe that deferring to district court judges on subsidiary, extrinsic fact-related issues, and relying on experts hired for having positions favorable to particular parties would likely result in even less uniformity. At least under our current regime, claim construction in all its aspects is reviewed by one appellate court. And providing formal deference to district courts in evaluating fact-related issues would encourage migration away from reliance on the intrinsic written record of the patent specification and its prosecution history.
One should also make no mistake about it: if deference were to be given to rulings on complicated subject matter, intensive appellate review would fade away (how many appeals from the PTO are now reversed following Zurko’s increase in the degree of deference given to the relatively expert examining agency?), and so will uniformity. In addition, if determining whether an issue is fact or law would determine the degree of deference granted, parties would be arguing over that question, as in appeals in veterans cases, rather than the real merits of claim construction. As for the relatively high reversal rate of claim construction at this court, I very much doubt that it is primarily due to so-called issues of historical fact; they are primarily due to our court’s review of the claims in light of the specification, not to failure to judge the credibility of contending expert witnesses. Besides, the reversal rate on claim construction is apparently coming down. See J. Jonas Anderson & Peter S. Menell, Informal Deference: An Historical, Empirical, and Normative Analysis of Patent Claim Construction, 108 Nw. U.L.Rev.-, 1 (forthcoming 2014).
This case did not involve subsidiary findings resolving disputes of historical fact. What was involved was whether there was corresponding structure to support “voltage source means” for “providing a constant or variable magnitude DC voltage between the DC input terminals”. The panel found the means clause in the claim lacked sufficient structure, and the specification similarly was lacking, so it reversed the district court on the ground that the claims were indefinite. Historical fact-finding was not involved; reading the claims and written description was. The en banc court should arrive at the same conclusion, as the district court did not rely on any subsidiary findings of fact. How a means plus function term is construed under § 112, ¶ 6 is not fact, but claim construction, ie., law.
For the above reasons, in addition to the majority’s reliance on the doctrine of stare decisis, I support the court’s decision not to overrule Cybor.