Egenera prevails at CAFC on "judicial estoppel" question.

The CAFC finds no judicial estoppel: Prior to claim construction, and alongside an ongoing inter partes review (“IPR”) proceeding, Egenera separately petitioned the United States Patent and Trademark Office (“PTO”) to remove one of the eleven listed inventors from the ’430 patent. Following the district court’s claim construction and a trial on inventorship, Egenera asked the district court to add the removed inventor back to the patent. The district court determined that judicial estoppel prevented Egenera from relisting the inventor and held the ’430 patent invalid for failing to name all inventors. See Egenera, Inc. v. Cisco Sys., Inc., 379 F. Supp. 3d 110 (D. Mass. 2019) (“Invalidity Decision”); Egenera, Inc. v. Cisco Sys., Inc., 348 F. Supp. 3d 99 (D. Mass. 2019) (“Judicial Estoppel Decision”); Egenera, Inc. v. Cisco Sys., Inc., No. 16-11613, 2018 WL 717342 (D. Mass. Feb. 5, 2018) (“Claim Construction Decision”). Egenera appeals, challenging both the district court’s claim construction and the application of judicial estoppel. For the reasons described below, we affirm the district court’s claim construction but vacate the invalidity judgment based on judicial estoppel and remand for further proceedings. Of the issue: First, we address whether Egenera could correct inventorship, even absent judicial estoppel. The Constitution authorizes awarding patent exclusivity only to an inventor. U.S. CONST. art. I, § 8, cl. 8. And so courts have historically held that if a patent does not reflect its true inventorship, it is invalid. See Pannu v. Iolab Corp., 155 F.3d 1344, 1349–50 (Fed. Cir. 1998) (collecting cases). Inventorship is sometimes easy to determine. But sometimes it is complicated, as with complex projects involving many contributors at various times. Ultimately, inventorship is a legal conclusion premised on underlying factual findings, and one that depends on claim construction. In re VerHoef, 888 F.3d 1362, 1365 (Fed. Cir. 2018); Trovan, Ltd. v. Sokymat SA, 299 F.3d 1292, 1302 (Fed. Cir. 2002). And like validity, inventorship is a claim-by-claim question. Trovan, 299 F.3d at 1302. Accordingly, who should be listed on the face of a patent may vary depending on what, exactly, is claimed and what, exactly, a court determines the claim scope to be. The Patent Act allows a listing of inventors to be corrected either upon petition to the Director, see 35 U.S.C. § 256(a), or upon court order, see § 256(b). Our precedent recognizes that a patent cannot be invalidated if inventorship can be corrected instead. Pannu, 155 F.3d at 1350. The statutory text recognizes this too: The error of omitting inventors or naming persons who are not inventors shall not invalidate the patent in which such error occurred if it can be corrected as provided in this section. 35 U.S.C. § 256(b) (emphases added). Section 256 applies if “through error a person is named . . . as the inventor, or through error an inventor is not named.” § 256(a). As previously noted, the inventorship question involved Egenera’s attempt to add back an inventor, Mr. Schulter, who had previously been removed alongside the IPR. The district court declared in a footnote that Mr. Schulter’s removal by petition was therefore “a considered act that is unlikely to qualify as an omission ‘through error.’” Judicial Estoppel Decision, 348 F. Supp. 3d at 102 n.1. Cisco agrees, arguing that Egenera’s petition was a “tactical ploy” rather than a “reasonable, but mistaken, effort to get inventorship right.” Appellee’s Br. 27–30 (arguing that “error” requires a “good-faith mistake”). We note that although the district court declined to credit Egenera’s witnesses’ accounts of conception, it also declined to find inequitable conduct on these facts, despite being urged by Cisco to do so. See Judicial Estoppel Decision, 348 F. Supp. 3d at 101; Invalidity Decision, 379 F. Supp. 3d at 129. Our precedent provides that “error” in § 256 includes “all varieties of mistakes—honest and dishonest”—rather than only unintentional inaccuracy. Stark v. Advanced Magnetics, Inc., 119 F.3d 1551, 1554–56 (Fed. Cir. 1997).3 That is, Stark expressly construed “error” to “embrace more than simply honest mistakes.” Id. at 1554. Stark’s broad interpretation was intended to “harmonize[] well with the title 35 policy of seeking to reward the actual inventors of technological advances.” Id. Indeed, § 256 is a savings provision, functioning to prevent invalidation when correction is available. It is the inequitable-conduct rules that provide a safety valve in the event of deceit. Id. at 1555–56. Cisco invites us to sidestep Stark because the AIA has since amended § 256. Previously, § 256 contained an additional requirement that the “error” of omitting an inventor occurred without “deceptive intention” on the inventor’s part. The scope of “error,” standing alone, was therefore broad enough to otherwise include acts amounting to “deceptive intention”—that is, intentional inaccuracy. See Stark, 119 F.3d at 1554. With the AIA, “without . . . deceptive intention” was struck from that section. See AIA, Pub. L. No. 112-29, § 20(f)(1)(B), 125 Stat. 284, 334 (2011). The essence of Cisco’s position is that when Congress removed this restrictive language that excluded intentional inaccuracy in certain cases, it somehow narrowed the meaning of “error” to exclude intentional inaccuracy entirely. Appellee’s Br. 30. Cisco argues that Congress’s removal of this language was meant to harmonize “error” with what Cisco views as its plain meaning—one that excludes intentional inaccuracy. We had not yet addressed the impact of the AIA on the holding of Stark. See, e.g., CODA Dev. S.R.O. v. Goodyear Tire & Rubber Co., 916 F.3d 1350, 1358 n.6 (Fed. Cir. 2019). But we now reject Cisco’s proposed interpretation because it is contrary to the text of § 256, the structure of the AIA,4 and the AIA’s legislative history.5 We hold that the AIA did not narrow the meaning of “error.” Accordingly, § 256 does not exclude “considered acts,” or even “deceptive intention,” from the meaning of “error.” Cf. Judicial Estoppel Decision, 348 F. Supp. 3d at 102 n.1. “Error” is simply the incorrect listing of inventors. Egenera asserted in its inventorship petition to the PTO, concurrent with the IPR, that Mr. Schulter was incorrectly listed as an inventor. At the time, no one had argued that “logic to modify” was a means-plus-function term. Indeed, it presumptively was not. And Egenera opposed such a construction when Cisco later advanced it; likewise, its position that Mr. Schulter was not an inventor was seemingly consistent with its preferred construction. But the court rejected Egenera’s construction in a way that also illuminated Mr. Schulter’s necessary presence as an inventor. After a three-day trial, and this appeal, the claim-construction and inventorship questions have at last been resolved. Because of these legal determinations, in retrospect, Egenera’s assertion in its inventorship petition was incorrect: Mr. Schulter was an inventor. According, we conclude that Mr. Schulter’s omission was “error.” Of judicial estoppel: Judicial estoppel is an equitable doctrine that prevents a litigant from taking a litigation position inconsistent with one successfully asserted in an earlier court proceeding. See id. “The purpose of the doctrine is to protect the integrity of the judicial process.” Id. Although the “contours of judicial estoppel are hazy,” and its application is case-dependent, the First Circuit applies the New Hampshire factors. RFF Family P’ship v. Ross, 814 F.3d 520, 527–28 (1st Cir. 2016) (Souter, J., sitting by designation). Accordingly, a court examines (1) whether a party’s earlier and later positions are “clearly inconsistent”—that is, “mutually exclusive”; (2) whether the party “succeeded in persuading a court to accept” the earlier position; and (3) whether the party would “derive an unfair advantage or impose an unfair detriment” on the other side if not estopped. New Hampshire v. Maine, 532 U.S. 742, 750–51 (2001). (...) To be “clearly inconsistent,” positions must be “mutually exclusive” and “directly inconsistent.” RFF, 814 F.3d at 528 (quoting Alt. Sys. Concepts, Inc. v. Synopsys, Inc., 374 F.3d 23, 33 (1st Cir. 2004)). In its § 256(a) petition, Egenera asked the PTO as a matter of formality “to delete Peter Schulter as an inventor of the invention being claimed.” J.A. 9367; see also 37 C.F.R. § 1.324; MPEP § 1481.02 (9th ed. Rev 10.2019, June 2020). Cisco argues that the “clearly inconsistent positions” were the two contrary uses of § 256.7 See Oral Arg. at 19:05–25, 26:29–38. Here, of course, in light of the district court’s claim construction and accompanying inventorship determination, Egenera is asking the district court to add Mr. Schulter back as an inventor. The district court concluded that “Egenera’s suggestion that Schulter may be relisted as an inventor as circumstances may dictate” was clearly inconsistent with its “September 2017 petition . . . that Schulter’s name was erroneously listed.” Judicial Estoppel Decision, 348 F. Supp. 3d at 102. This was incorrect. We do not think that multiple corrections under § 256 are per se “mutually exclusive.” In any event, the district court’s intervening claim-construction and inventorship determinations further justify any seeming inconsistency. Cf. Biomedical Pat. Mgmt. Corp. v. Cal. Dep’t of Health Servs., 505 F.3d 1328, 1341–42 (Fed. Cir. 2007) (noting that “inconsistency” in the judicial estoppel context “is excused by an intervening change in the law”); see generally 18B Wright & Miller, Federal Practice & Procedure § 4477.3 (2d ed., Apr. 2020 update). Inventorship, a complex legal conclusion, can depend on claim construction. Here, the underlying presumption was that Egenera’s claim terms, lacking “means,” were not means-plus-function. Egenera’s inventorship petition was consistent with that presumption. Indeed, it may well be that Mr. Schulter would not be an inventor under Egenera’s preferred construction; but inventorship under that claim construction was not decided. And Egenera consistently protested the means-plus-function construction both at the district court and on appeal—a construction that the inventorship question was directly predicated on. Therefore, once those issues were decided, it was entirely consistent for Egenera to request an accompanying formal correction of inventorship. Accordingly, at least due to the intervening claim construction, it was not “mutually exclusive,” as judicial estoppel requires, to again request formal correction of inventorship. The district court thus erred in discerning “clearly inconsistent” positions. AND This determination is narrow. We do not hold that judicial estoppel cannot apply to statements made during substantive prosecution, ex parte reexamination, or other quasi-adjudicatory proceedings—an issue not before us. And we do not go so far as to say that other theories of estoppel cannot apply to ministerial filings or representations before the PTO. But judicial estoppel cannot be stretched beyond persuading a tribunal, and it does not apply here. 3 Third, Egenera would gain no unfair advantage, and Cisco would suffer no unfair prejudice, if judicial estoppel were not applied. The focus of this inquiry is on whether not applying estoppel would result in unfair advantage or prejudice. See New Hampshire, 532 U.S. at 751 (“[Courts ask] whether the party seeking to assert an inconsistent position would derive an unfair advantage . . . if not estopped.”); RFF, 814 F.3d at 528.