concurring in part and dissenting in part:
I agree with much of the Court’s opinion, including its determination that the EPA was not required to issue a SIP call as the exclusive means of revising the SIP approved in 2008. To that extent, I concur in the majority opinion. But I respectfully dissent from the Court’s determination that “[njeither the text of the final rule disapproving revision published in the Federal Register, nor the record before this Court, reveal that EPA affirmatively made the requisite error determination.” Majority Op. at 1287.
My disagreement is based on two specific differences with the Court’s opinion. First, the EPA’s notices and decisions in the Federal Register quite clearly describe how the agency believes it was in error in approving the 2008 proposed revision. Second, the process that the EPA followed was more accommodating to the State and to the affected industries than what will inevitably follow from Court’s decision, as applied in future cases. I believe we should dismiss the 2008 action for lack of jurisdiction and affirm the EPA’s 2011 disapproval.
I.
When the EPA published notice that it was proposing either to affirm its approval or disapprove Alabama’s proposed revision, it said:
EPA’s prior approval notice provides extensive discussion of the reasons why EPA concluded in that notice that section 110(0 had been satisfied. In particular, EPA stated as grounds for this conclusion that: “(1) The revision would not increase the allowable average opacity levels; and (2) the relationship between changes in opacity and increases or decreases in ambient PM2iS levels cannot be quantified readily for the sources subject to this SIP revision, and is particularly uncertain for short-term analy-ses.”
74 Fed. Reg. 50930, 50932-33 (Oct. 2, 2009) (quoting 73 Fed. Reg. at 60959 (Oct. 15, 2008)) (internal citations and footnote omitted).
Given those two reasons for approving the 2008 proposed revision, the error that the EPA believes it committed is identified in the thorough description set forth in the Federal Register:
Based on the information received to date, EPA believes that increases or decreases in PM2.5 emissions based on short-term increases in opacity cannot be quantified readily for the sources subject to this SIP revision. There are several contributors to the uncertainties associated with relating mass emissions to increases in opacity, including: (1) Differences between combustion technology characteristics and fuel components, (2) differences in control technology types, temperatures at which they operate, and load characteristics, (3) the recognition that both opacity and mass emissions are subject to significant variability over short periods of time and fluctuations such that one may act independently of the other, and (4) differences between the mass of particles that exists at the point of opacity measurement by the COMS (e.g., in the stack) and the direct PM2.5 that forms immediately upon exiting the stack (that are related to fuel components more than to control technology).
Id. at 50933. In this passage, the agency makes clear that because “the relationship between changes in opacity and increases *1296or decreases in ambient PM2.5 levels cannot be quantified readily for the sources subject to this SIP revision” and because there are at least four distinct reasons that little or no change in opacity might yet be accompanied by a significant increase in actual emission of particulate matter, the EPA cannot say whether the proposed revision “would interfere with any applicable requirement concerning attainment.” Because § 110(0 of the Clean Air Act requires the EPA to ensure that the proposed revision will not interfere with any applicable requirement concerning attainment, the EPA’s approval did not comply with the Act. That is the identified error or mistake the agency made.
The EPA also described a second error, an extension of the first, when it announced its disapproval of the 2008 proposed revision. It said that in its prior approval of the revision:
[W]e established a new metric of “average daily opacity” (and “average quarterly opacity”) and concluded that sec-' tion 110© did not prohibit approval of a SIP revision that allowed significantly increased opacity levels for longer consecutive periods of time because the revision would not increase the allowable average opacity levels (on either a quarterly or daily basis). This analysis was focused on opacity and operational conditions regarding opacity as opposed to a focus on the relationship between opacity and PM mass emissions ....
76 Fed. Reg. 18870, 18874 (Apr. 6, 2011) (emphases added). In short, the prior analysis looked at numbers, not at what the numbers were supposed to index or signify. The mistaken prior analysis was inconsistent with the requirements of the Act.
When it announced its final decision to disapprove the proposed revision, the EPA referred back to the two identified reasons in support of approval that were stated in its Notice of October 2, 2009. It said:
[B]oth of the findings that provided the foundation for [EPA’s] initial approval of the SIP revision were not strong enough to support approval under the [Clean Air Act]. EPA concludes that, as it was described in the Submittals, the concept of “average daily opacity” is not a useful tool for evaluating whether the Submit-tals are likely to maintain current air quality, particularly given the lack of other limitations on opacity exceedances in the Submittals ....
76 Fed. Reg. 18870, 18874-75 (Apr. 6, 2011) (emphases added).
All of these statements describe errors of judgment. The EPA does not use the specific words “error” or “correction.” Nor does it identify a mathematical mistake or linguistic slip. And it does not provide an easy descriptor of the error or expressly invoke § 110(k)(6). But, just as the majority does not require the EPA to intone “magic words,” Majority Op. at 1290, there is no reason why such omissions should doom the agency’s action.
The majority states, “Congress has required the EPA to articulate an ‘error’ and provide ‘the basis’ of its determination that an error occurred.” Id. at 1287 (emphasis added). But that is not exactly what the Act says. The Act says the EPA may revise its prior approval, disapproval, or promulgation whenever it determines that its action “was in error.” Section 110(k)(6) (emphasis added). When error is spoken of as a place, the intent is often to include misjudgment rather than mere mistake or omission.
Because the EPA described the errors of judgment that it sought to correct, § 110(k)(6) authorized its reconsideration of its approval of Alabama’s proposed SIP revision. That being so, the agency had not “completed its decisionmaking pro*1297cess” as to the 2008 approval, Franklin v. Massachusetts, 505 U.S. 788, 797,112 S.Ct. 2767, 120 L.Ed.2d 636 (1992), and the erstwhile approval became “interlocutory in nature,” National Parks Conservation Ass’n v. Norton, 324 F.3d 1229, 1236-37 (11th Cir.2003) (quoting Bennett v. Spear, 520 U.S. 154, 177-78, 117 S.Ct. 1154, 137 L.Ed.2d 281 (1997)). See also Whitman v. Am. Trucking Ass’ns, Inc., 531 U.S. 457, 479, 121 S.Ct. 903, 149 L.Ed.2d 1 (2001). Therefore, I would dismiss the 2008 action for lack of jurisdiction.
II.
I would also uphold the EPA’s action in 2011 disapproving the proposed revision. The proposal, after all, would apply not only to the coal-fired utilities represented in this action but also to cement manufacturing facilities, as well as pulp and paper facilities, entities which might generate different kinds of particulate matter in different ways with different consequences. See 76 Fed.Reg. 18873 & n. 8; 74 Fed.Reg. at 50933. The EPA found that, while “[o]ne of the primary purposes of opacity limits is to ensure that PM control devices are operating within normal parameters,” the proposed revision might result in failure to alert an operator or the State or EPA to an equipment malfunction that is not only causing increased opacity but failing to control types of PM emissions that the operator mistakenly believes to be under control. 76 Fed.Reg. at 18874. “[A] rule that allows no more than one 6-minute exceedance per hour and opacity readings no greater than 40 percent clearly requires more effective control equipment and/or operating procedures than ... a rule that allows longer consecutive periods of exempt opacity excursions and at higher opacity levels.” Id. at 18876.
Further, the EPA found:
[Elements that are missing from the submitted modeling include: data from all the sources and source categories affected by the Aabama Submittals!1]; a demonstration of the relationship between PM emissions and opacity at a particular facility and source-category; consideration of emissions from other sources in the modeled area; condensa-ble PM data; explanation for background PM levels used in the evaluation; and an explanation of the use of PM10 as a surrogate for PM2.5; among other concerns.
Id. at 18875. The EPA concluded that “it is reasonably foreseeable that approving the Aabama Submittals would allow increased mass emissions, for at least some sources and under at least some conditions, over the PM emission levels that would have been allowed” under the previous rule. Consequently, “[SJection 110(0 requires disapproval ... absent additional limitations which would significantly diminish the likelihood that mass emissions increases will occur.” Id. at 18876.
In its 2011 disapproval, in my view, the EPA permissibly construed § 110(0 and acted neither arbitrarily nor capriciously when it determined that an increase in opacity limits will “interfere” with the PM2.e standard for attainment by making it more difficult for operators to spot malfunctions and by exempting more emissions of particulate matter. Athough opacity is not attainment and is not an air-quality standard, it is aptly described as an “applicable requirement concerning attainment.” These determinations are squarely within the heartland of EPA expertise and *1298the sphere of judicial deference to it under Chevron. That being so, the law requires us to defer to the agency determination in this case where the meaning of words or phrases in a statute or regulation is evident from context. See Nat’l Ass’n of Home Builders v. Defenders of Wildlife, 551 U.S. 644, 666, 127 S.Ct. 2518, 168 L.Ed.2d 467 (2007).
Of greater concern is what I perceive will be the consequence of the majority’s reasoning. The EPA’s Notice of October 2, 2009, solicited input on whether its 2008 approval was, in fact, “in error.” I believe that the EPA complied with § 110(k)(6) in making this request for input because the statute requires it to undertake any revision of its prior action “in the same manner as the approval” but “without requiring any further submission from the State.” When it gave notice of its intent to reconsider, the EPA said:
This rulemaking is part of EPA’s reconsideration process on our October 15, 2008, final action approving Alabama’s visible emissions SIP revisions. EPA is seeking public comment on proposals to affirm our prior action, which approved the SIP revisions, or amend and disapprove the revisions to Alabama SIP rule 335-3-4-.01 (“Visible Emissions”), submitted initially in 2003 and significantly revised and re-submitted on August 22, 2008.
74 Fed. Reg. at 50934.
The EPA could have simply decreed in 2009 that it made an error of judgment in approving Alabama’s proposed revision because the evidence was not sufficient to ensure that allowing opacity greater than 20% for more extended periods of time would not result in emission of more particulate matter. Had it done so, under the majorities’ reasoning we could not now conclude that the EPA’s reconsideration was faulty. It is not particularly unusual for a change in philosophy within an agency to result in a policy about-face or to perceive error in earlier decisions. See, e.g., FCC v. Fox Television Stations, Inc., 556 U.S. 502, 514-15, 129 S.Ct. 1800, 173 L.Ed.2d 738 (2009); Nat’l Ass’n of Home Builders v. EPA, 682 F.3d 1032, 1034-39 (D.C.Cir.2012); Chamber of Commerce v. EPA, 642 F.3d 192, 196-97 (D.C.Cir.2011); see also Sierra Club v. Ga. Power Co., 443 F.3d 1346, 1353-54 & n. 11 (11th Cir.2006) (describing the EPA’s clarification in 2001 of guidance that it provided in 1999). What one person calls “a reevaluation of which policy would be better in light of the facts,” Home Builders, 682 F.3d at 1038, another person may reasonably call a determination that prior policy was “in error,” § 110(k)(6).
But, here, instead of taking wholly on itself the determination that an error occurred, EPA invited the interested parties to provide input on the question of whether it erred. It specifically asked for plant- or industry-specific data about the relationship between opacity and real amounts of particulate emissions into the ambient air. As a result of the EPA’s willingness to listen to these responses and consider additional data from interested parties before it made a final decision whether to change course, the majority concludes that the EPA exceeded its authority. If this reasoning is correct, and I do not think it is, the lesson for the EPA is — declare errors first, and investigate whether they exist later. It has a “verdict first, evidence later” feel to it.
No one affected by EPA’s decision in this case claims they were unable to understand why the agency was concerned about its prior action. On the contrary, both the State and the utility industries avoid making any such claim, asserting instead that EPA failed to identify a “technical or clerical error” — -words that cannot be found in § 110(k)(6)’s phrase “in error.” *1299No one has even attempted to refute the EPA’s concern about the relationship between opacity and particulate matter or that consideration of the relationship is an issue wholly within the mainstream of the agency’s responsibilities.
In my view, the process functioned just as it is supposed to. The EPA did not hide the ball, it did not do anything that was irrational, arbitrary, or capricious, it gave ample opportunity to interested groups to explain their own positions and allay the EPA’s concerns, and, at least in my opinion, it clearly explained its own concerns and final decision. I believe the record fails to support the appellant’s position and, under applicable law, supports the EPA’s actions and decision.
The majority’s conclusion that the EPA “failed to articulate an error” reflects a determination of what the majority expects the articulation of error to sound like and less to do with the EPA’s process or its findings about air quality and particulate matter.
When the question is whether the EPA had authority to reconsider under § 110(k)(6), we should read what the agency said in the Federal Register in the particular case at hand and ask whether the agency says that it was wrong on a prior occasion. Doing so remains the best way to avoid asking for “magic words” and is the best way to determine whether the agency’s actions — regardless of its brief or oral argument, see Majority Op. at 1288-89 & n. 11 — were consistent with the law enacted by Congress.
For these reasons, I respectfully dissent.
. For instance, while Alabama Power Company submitted models, cement manufacturers and pulp and paper facilities, whose operations would also be affected by the proposed revision, apparently did not. 74 Fed. Reg. at 50933; 76 Fed. Reg. at 18873.