Srinivasan, Circuit Judge:

Unwanted robocalls are an all-too-familiar phenomenon. For years, consumers have complained to the Federal Communications Commission about automated telemarketing calls and text messages that they did not seek and cannot seem to stop.

Congress sought to address consumers' concerns with undesired robocalls in the Telephone Consumer Protection Act of 1991. The TCPA generally prohibits the use of certain kinds of automated dialing equipment to call wireless telephone numbers absent advance consent. The Act vests the Commission with authority to implement those restrictions.

In this case, a number of regulated entities seek review of a 2015 order in which the Commission sought to clarify various aspects of the TCPA's general bar against using automated dialing devices to make uninvited calls. The challenges encompass four issues addressed by the agency's order: (i) which sorts of automated dialing equipment are subject to the TCPA's restrictions on unconsented calls; (ii) when a caller obtains a party's consent, does a call nonetheless violate the Act if, unbeknownst to the caller, the consenting party's wireless number has been reassigned to a different person who has not given consent; (iii) how may a consenting party revoke her consent; and (iv) did the Commission too narrowly fashion an exemption from the TCPA's consent requirement for certain healthcare-related calls.

We uphold the Commission's approach to revocation of consent, under which a party may revoke her consent through any reasonable means clearly expressing a desire to receive no further messages from the caller. We also sustain the scope of the agency's exemption for time-sensitive healthcare calls.

We set aside, however, the Commission's effort to clarify the types of calling equipment that fall within the TCPA's restrictions. The Commission's understanding would appear to subject ordinary calls from any conventional smartphone to the Act's coverage, an unreasonably expansive interpretation of the statute. We also vacate the agency's approach to calls made to a phone number previously assigned to a person who had given consent but since reassigned to another (nonconsenting) person. The Commission concluded that calls in that situation violate the TCPA, apart from a one-call safe harbor, regardless of whether the caller has any awareness of the reassignment. We determine that the agency's one-call safe harbor, at least as defended in the order, is arbitrary and capricious.

We therefore grant the petitions for review in part and deny them in part.

I.

The federal government's efforts to combat unwanted robocalls have spanned nearly three decades, involving two federal agencies and a number of congressional enactments. In the Telemarketing and Consumer Fraud and Abuse Prevention Act of 1994, 15 U.S.C. § 6101 et seq., Congress empowered the Federal Trade Commission to regulate the telemarketing industry. The FTC's measures include a general bar against calling any telephone number on the "do-not-call registry" without consent or an established business relationship. 16 C.F.R. § 310.4(b)(1)(iii)(B); see 15 U.S.C. § 6151(a). This case does not concern the FTC's initiatives.

This case instead concerns the Federal Communications Commission's efforts to combat unwanted robocalls pursuant to its authority under the TCPA. Some of the Commission's restrictions on telemarketing calls mirror measures established by the FTC. Compare 16 C.F.R. §§ 310.4(b)(1)(iii)(B), 310.4(c), with 47 C.F.R. § 64.1200(c). But the agencies' initiatives also differ in various respects. Of relevance here, only the TCPA specifically restricts the use of an "automatic telephone dialing system" to make calls. 47 U.S.C. § 227(b)(1)(A). Petitioners challenge the Commission's interpretation and implementation of various TCPA provisions pertaining to automated dialing equipment.

A.

Congress enacted the TCPA in 1991 based on findings that the "use of the telephone to market goods and services to the home and other businesses" had become "pervasive due to the increased use of cost-effective telemarketing techniques." 47 U.S.C. § 227 note, Pub. L. No. 102-243, § 2(1), 105 Stat. 2394, 2394. "Many consumers," Congress determined, "are outraged over the proliferation of intrusive, nuisance calls to their homes from telemarketers." Id. § 2(6)-(7).

The TCPA restricts calls both "to any residential telephone line" and to "any telephone number assigned to a ... cellular telephone service." 47 U.S.C. § 227(b)(1)(A)(iii), (B). This case solely concerns the latter restrictions on telephone calls to wireless numbers.

Congress, in that regard, made it "unlawful... to make any call (other than a call made for emergency purposes or made with the prior express consent of the called party) using any automatic telephone dialing system ... to any telephone number assigned to a ... cellular telephone service," "unless such call is made solely to collect a debt owed to or guaranteed by the United States." Id. § 227(b)(1)(A)(iii). The statute defines an "automatic telephone dialing system" (ATDS, or autodialer) as "equipment which has the capacity — (A) to store or produce telephone numbers to be called, using a random or sequential number generator; and (B) to dial such numbers." Id. § 227(a)(1).

In short, the TCPA generally makes it unlawful to call a cell phone using an ATDS. And an ATDS is equipment with the "capacity" to perform each of two enumerated functions: (i) storing or producing telephone numbers "using a random or sequential number generator" and (ii) dialing those numbers. The general prohibition on autodialer calls to wireless numbers is subject to three exceptions. The central exception for purposes of this case is for calls made with "prior express consent." There are also exceptions for emergency calls and calls made to collect government debts.

The TCPA vests the Commission with responsibility to promulgate regulations implementing the Act's requirements. Id. § 227(b)(2). The Act also grants the Commission specific authority to fashion exemptions from the general prohibition on autodialer calls to wireless numbers, where the calls are "not charged to the called party." Id. § 227(b)(2)(C). As Congress explained, the FCC "should have the flexibility to design different rules for those types of automated or prerecorded calls that it finds are not considered a nuisance or invasion of privacy." Id. § 227 note, Pub. L. No. 102-243, § 2(13), 105 Stat. 2394, 2395.

Since the TCPA's enactment, the FCC has issued a series of rulemakings and declaratory rulings addressing the Act's reach. In 2003, for instance, the agency concluded that the statute's restrictions on "mak[ing] any call" using an ATDS encompass the sending of text messages. See In re Rules and Regulations Implementing the Telephone Consumer Protection Act of 1991 (2003 Order), 18 FCC Rcd. 14,014, 14,115 ¶ 165 (2003).

The Act contains a private right of action permitting aggrieved parties to recover at least $500 in damages for each call made (or text message sent) in violation of the statute, and up to treble damages for each "willful[] or knowing[]" violation. 47 U.S.C. § 227(b)(3). There is no cap on the amount of recoverable damages. The Commission has noted a surge in TCPA lawsuits (including class actions) in recent years, likely attributable in part to the "skyrocketing growth of mobile phones." In re Rules and Regulations Implementing the Telephone Consumer Protection Act of 1991 (2015 Declaratory Ruling), 30 FCC Rcd. 7961, 7970 ¶¶ 6-7 (2015).

B.

In a Declaratory Ruling and Order issued in 2015, the Commission (with two Commissioners dissenting) addressed 21 separate petitions for rulemaking or requests for clarification. In this court, petitioners and intervenors seek review of four aspects of the Commission's order.

First, the Commission sought to clarify which devices for making calls qualify as an ATDS — i.e., equipment that "has the capacity" to "store or produce telephone numbers to be called, using a random or sequential number generator," and "to dial such numbers." 47 U.S.C. § 227(a)(1). With regard to whether equipment has the "capacity" to perform the enumerated functions, the Commission declined to define a device's "capacity" in a manner confined to its "present capacity." Instead, the agency construed a device's "capacity" to encompass its "potential functionalities" with modifications such as software changes. 2015 Declaratory Ruling, 30 FCC Rcd. at 7974 ¶ 16.

The Commission also addressed the precise functions that a device must have the capacity to perform for it to be considered an ATDS. The Commission reaffirmed prior orders deciding that "predictive dialers" — equipment that can dial automatically from a given list of telephone numbers using algorithms to predict "when a sales agent will be available" — qualify as autodialers. Id. at 7972 ¶ 10 & n.39. The Commission further explained that a "basic function[]" of an autodialer is to "dial numbers without human intervention." Id. at 7975 ¶ 17. At the same time, the Commission also declined to "clarify[] that a dialer is not an autodialer unless it has the capacity to dial numbers without human intervention." Id. at 7976 ¶ 20.

Second, the Commission spoke to whether, and when, a caller violates the TCPA by calling a wireless number that has been reassigned from a consenting party to another person without the caller's knowledge. The Act specifically permits autodialer calls "made with the prior express consent of the called party." 47 U.S.C. § 227(b)(1)(A). If the "called party" for those purposes refers to the intended recipient of a call or message, a caller would face no liability when using an ATDS to call a number believed to belong to a consenting party, even if the number in fact has been reassigned to another person who has not consented.

The Commission, though, determined that the term "called party" refers not to "the intended recipient of a call" but instead to "the current subscriber" (i.e., the current, nonconsenting holder of a reassigned number rather than a consenting party who previously held the number). 2015 Declaratory Ruling, 30 FCC Rcd. at 7999 ¶ 72. But the Commission did not hold a caller strictly liable when unaware that the consenting party's number has been reassigned to another person. Instead, the agency allowed one — and only one — liability-free, post-reassignment call for callers who lack "knowledge of [the] reassignment" and possess "a reasonable basis to believe that they have valid consent." Id. at 8000 ¶ 72.

Third, the Commission clarified the ways in which a consenting party can revoke her consent to receive autodialer calls. The Commission decided that callers may not unilaterally designate the acceptable means of revocation. It also declined to prescribe its own set of mandatory revocation procedures. Rather, it concluded that "a called party may revoke consent at any time and through any reasonable means" — whether orally or in writing — "that clearly expresses a desire not to receive further messages." Id. at 7989-90 ¶ 47; id. at 7996 ¶ 63.

Fourth, and finally, the Commission exempted from the autodialer provision's consent requirement certain calls to wireless numbers "for which there is exigency and that have a healthcare treatment purpose." Id. at 8031 ¶ 146. It declined, however, to give the exemption the reach desired by certain parties that are in the business of healthcare-related marketing calls.

We will take up the challenges to those four aspects of the Commission's 2015 ruling in the same order.

II.

Under the Administrative Procedure Act, we assess whether the Commission's challenged actions in its 2015 order were "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law." 5 U.S.C. § 706(2)(A). We review the lawfulness of the Commission's interpretations of the TCPA using the two-step Chevron framework. That inquiry calls for examining whether "Congress has directly spoken to the precise question at issue," and, if not, whether "the agency's answer is based on a permissible construction of the statute." Chevron U.S.A. Inc. v. Nat'l Res. Def. Council, Inc., 467 U.S. 837, 842-43, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984).

To be lawful, the Commission's challenged actions must also satisfy the Administrative Procedure Act's requirement that they not be arbitrary or capricious. Arbitrary-and-capricious review includes assuring that the agency "engaged in reasoned decisionmaking." Judulang v. Holder, 565 U.S. 42, 53, 132 S.Ct. 476, 181 L.Ed.2d 449 (2011). Review of agency action for arbitrariness and capriciousness sometimes entails essentially the same inquiry as review of an agency's exercise of statutory interpretation under Chevron's second step. See id. at 52 n.7, 132 S.Ct. 476; Agape Church, Inc. v. FCC, 738 F.3d 397, 410 (D.C. Cir. 2013).

Applying those standards to petitioners' four sets of challenges to the Commission's 2015 Declaratory Ruling, we set aside the Commission's explanation of which devices qualify as an ATDS, as well as its understanding of when a caller violates the Act by calling a wireless number previously held by a consenting party but reassigned to a person who has not given consent. We sustain, however, the Commission's ruling that a party can revoke consent through any reasonable means clearly expressing a desire to receive no further calls or texts, and we also uphold the scope of the Commission's exemption for time-sensitive, healthcare-related calls.

A.

We first consider the Commission's effort to clarify which sorts of calling equipment qualify as an ATDS so as to fall subject to the general prohibition against making calls using such a device without consent. The statute defines an ATDS as "equipment which has the capacity — (A) to store or produce telephone numbers to be called, using a random or sequential number generator; and (B) to dial such numbers." 47 U.S.C § 227(a)(1). That definition naturally raises two questions: (i) when does a device have the "capacity" to perform the two enumerated functions; and (ii) what precisely are those functions? We conclude that the Commission's approach to those two questions cannot be sustained, at least given the Commission's unchallenged assumption that a call made with a device having the capacity to function as an autodialer can violate the statute even if autodialer features are not used to make the call.

1.

a. In addressing what it means for equipment to have the "capacity" to perform the autodialer functions enumerated in the statute, the Commission rejected the arguments of various parties that a device's capacity must be measured solely by reference to its "present capacity" or its "current configuration" without any modification. 2015 Declaratory Ruling, 30 FCC Rcd. at 7974 ¶ 16. The Commission instead determined that the "capacity" of calling equipment "includes its potential functionalities" or "future possibility," not just its "present ability." Id. at 7974 ¶ 16; id. at 7975 ¶ 20.

The Commission reasoned that the "functional capacity of software-controlled equipment is designed to be flexible, both in terms of features that can be activated or de-activated and in terms of features that can be added to the equipment's overall functionality through software changes or updates." Id. at 7974 ¶ 16 n.63. And the Commission found support for its "potential functionalities" approach in dictionary definitions of the term "capacity," one of which is "the potential or suitability for holding, storing, or accommodating." Id. at 7975 ¶ 19 (quoting Capacity, Merriam-Webster Dictionary Online, https://www.merriam-webster.com/dictionary/capacity (as visited May 18, 2015)).

In challenging the Commission's approach, petitioners argue that the term "capacity" in the statutory definition of an ATDS can refer only to a device's "present ability," i.e., its current and unmodified state, not its "potential ability" taking into account possible upgrades or modifications. It is far from clear, though, that labels such as "present" ability versus "potential" ability should carry dispositive weight in assessing the meaning of the statutory term "capacity." After all, even under the ostensibly narrower, "present ability" interpretation advanced by petitioners, a device that "presently" (and generally) operates as a traditional telephone would still be considered have the "capacity" to function as an ATDS if it could assume the requisite features merely upon touching a button on the equipment to switch it into autodialer mode. Virtually any understanding of "capacity" thus contemplates some future functioning state, along with some modifying act to bring that state about.

Consequently, the question whether equipment has the "capacity" to perform the functions of an ATDS ultimately turns less on labels such as "present" and "potential" and more on considerations such as how much is required to enable the device to function as an autodialer: does it require the simple flipping of a switch, or does it require essentially a top-to-bottom reconstruction of the equipment? And depending on the answer, what kinds (and how broad a swath) of telephone equipment might then be deemed to qualify as an ATDS subject to the general bar against making any calls without prior express consent?

b. Here, the Commission adopted an expansive interpretation of "capacity" having the apparent effect of embracing any and all smartphones: the device routinely used by the vast majority of citizens to make calls and send messages (and for many people, the sole phone equipment they own). It is undisputed that essentially any smartphone, with the addition of software, can gain the statutorily enumerated features of an autodialer and thus function as an ATDS. The Commission in its ruling did not question the observation of a dissenting Commissioner that "[i]t's trivial to download an app, update software, or write a few lines of code that would modify a phone to dial random or sequential numbers." 2015 Declaratory Ruling, 30 FCC Rcd. at 8075 (Comm'r Pai, dissenting). The Commission itself noted that "[d]ialing options" are now "available via smartphone apps" that enable "[c]alling and texting consumers en masse." Id. at 7970 ¶ 7.

The Commission's ruling concluded that app downloads and other software additions of that variety — and the enhanced functionality they bring about — are appropriately considered to be within a device's "capacity." The ruling states that equipment's "functional capacity" includes "features that can be added ... through software changes or updates." Id. at 7974 ¶ 16 n.63. As a result, "a piece of equipment can possess the requisite `capacity' to satisfy the statutory definition of an `autodialer' even if, for example, it requires the addition of software to actually perform the functions described in the definition." Id. at 7975 ¶ 18. The Commission reinforced the point in an example set forth in its brief in this case: "If I ask whether the Firefox browser has the `capacity' to play Flash videos, it would be natural for you to answer `Yes, if you download the Flash plug-in' — and it would be incorrect for you to answer `No.'" FCC Br. 29.

If a device's "capacity" includes functions that could be added through app downloads and software additions, and if smartphone apps can introduce ATDS functionality into the device, it follows that all smartphones, under the Commission's approach, meet the statutory definition of an autodialer. The Commission's ruling does not deny that conclusion.

To the contrary, a number of parties specifically argued to the agency "that a broad interpretation of `capacity' could potentially sweep in smartphones because they may have the capacity to store telephone numbers to be called and to dial such numbers through the use of an app or other software." 2015 Declaratory Ruling, 30 FCC Rcd. at 7976 ¶ 21. Rather than resist that contention, the Commission assumed its correctness, responding that, even if smartphones qualify as autodialers, it was unclear to the Commission that the "typical use of smartphones" would be "likely" to give rise to "unwanted calls" of a kind producing "legal action." Id. at 7977 ¶ 21. A dissenting Commissioner read that portion of the Commission's order to "acknowledge[] that smartphones are swept in under its reading," such that "each and every smartphone ... is an automatic telephone dialing system." Id. at 8075 & n.576 (Comm'r Pai, dissenting). The Commission did not disagree or suggest otherwise.

c. If every smartphone qualifies as an ATDS, the statute's restrictions on autodialer calls assume an eye-popping sweep. Recall that the statute generally bars the use of an ATDS to make any call (or send any text message) without prior express consent, and tags each violation with a minimum $500 penalty in damages for each individual recipient of each prohibited call or message. The reach of the statute becomes especially pronounced upon recognizing that, under the Commission's approach, an uninvited call or message from a smartphone violates the statute even if autodialer features were not used to make the call or send the message. Id. at 7976 ¶ 19 n.70. We explore that interpretive issue in greater depth below (infra § II.A.3); but for now, it suffices to appreciate the Commission's understanding that, as long as equipment has the "capacity" to function as an autodialer — as is true of every smartphone under the agency's view — any uninvited call or message from the device is a statutory violation.

Imagine, for instance, that a person wishes to send an invitation for a social gathering to a person she recently met for the first time. If she lacks prior express consent to send the invitation, and if she obtains the acquaintance's cell phone number from a mutual friend, she ostensibly commits a violation of federal law by calling or sending a text message from her smartphone to extend the invitation. See 2015 Declaratory Ruling, 30 FCC Rcd. at 8076 (Comm'r Pai, dissenting). And if she sends a group message inviting ten people to the gathering, again without securing prior express consent from any of the recipients, she not only would have infringed the TCPA ten distinct times but would also face a minimum damages recovery against her of $5,000.

Those sorts of anomalous outcomes are bottomed in an unreasonable, and impermissible, interpretation of the statute's reach. The TCPA cannot reasonably be read to render every smartphone an ATDS subject to the Act's restrictions, such that every smartphone user violates federal law whenever she makes a call or sends a text message without advance consent.

A "significant majority of American adults" owned a smartphone even by 2013. Riley v. California, ___ U.S. ___, 134 S.Ct. 2473, 2484, 189 L.Ed.2d 430 (2014). And as of the end of 2016, nearly 80% of American adults had become smartphone owners. See 10 Facts About Smartphones as the iPhone Turns 10, Pew Research Ctr., June 28, 2017, http://www.pewresearch.org/fact-tank/2017/06/28/10-facts-about-smartphones (last visited Dec. 18, 2017). That figure will only continue to grow, and increasingly, individuals own no phone equipment other than a smartphone. See id.; Wireless Substitution: Early Release of Estimates From the National Health Interview Survey, January-June 2017, Nat'l Ctr. for Health Statistics 1 (Dec. 2017), https://www.cdc.gov/nchs/data/nhis/earlyrelease/wireless201705.pdf.

It is untenable to construe the term "capacity" in the statutory definition of an ATDS in a manner that brings within the definition's fold the most ubiquitous type of phone equipment known, used countless times each day for routine communications by the vast majority of people in the country. It cannot be the case that every uninvited communication from a smartphone infringes federal law, and that nearly every American is a TCPA-violator-in-waiting, if not a violator-in-fact.

In that regard, it is notable that Congress, in its findings setting forth the basis for the statute, found that some "30,000 businesses actively telemarket goods and services to business and residential customers" and "[m]ore than 300,000 solicitors call more than 18,000,000 Americans every day." 47 U.S.C. § 227 note, Pub. L. No. 102-243, § 2(2)-(3), 105 Stat. 2394, 2394. Those sorts of predicate congressional findings can shed substantial light on the intended reach of a statute. See Sutton v. United Air Lines, Inc., 527 U.S. 471, 484-87, 119 S.Ct. 2139, 144 L.Ed.2d 450 (1999).

Of course, there is no expectation that a statute's reach necessarily will precisely match Congress's findings about a problem it aims to address, and Congress might well fashion a statute's operative provisions with built-in flexibility to accommodate expansion of the concerns animating the legislation over time. But a several-fold gulf between congressional findings and a statute's suggested reach can call into doubt the permissibility of the interpretation in consideration.

That is what happened in Sutton. There, the Supreme Court rejected an interpretation of the term "disability" in the Americans with Disabilities Act that would have treated some 160 million persons as disabled in the face of congressional findings contemplating the population of disabled persons as numbering only 43 million. See id.; id. at 494-95, 119 S.Ct. 2139 (Ginsburg, J., concurring). (After Sutton, Congress amended the statutory findings and the statute to allow for an expansive application. See ADA Amendments Act of 2008, Pub. L. No. 110-325, § 2, 122 Stat. 3553, 3554.)

Here, as in Sutton, the Commission's expansive understanding of "capacity" in the TCPA is incompatible with a statute grounded in concerns about hundreds of thousands of "solicitors" making "telemarketing" calls on behalf of tens of thousands of "businesses." The Commission's interpretation would extend a law originally aimed to deal with hundreds of thousands of telemarketers into one constraining hundreds of millions of everyday callers.

The Commission's capacious understanding of a device's "capacity" lies considerably beyond the agency's zone of delegated authority for purposes of the Chevron framework. As we have explained, "even if the [statute] does not foreclose the Commission's interpretation, the interpretation [can] fall[] outside the bounds of reasonableness" at Chevron's second step. Goldstein v. SEC, 451 F.3d 873, 880-81 (D.C. Cir. 2006). That is because an "agency['s] construction of a statute cannot survive judicial review if a contested regulation reflects an action that exceeds the agency's authority." Id. (quoting Aid Ass'n for Lutherans v. United States Postal Serv., 321 F.3d 1166, 1174 (D.C. Cir. 2003)).

In Aid Ass'n, for example, we examined Postal Service regulations that excluded nonprofit organizations' use of certain reduced postage rates. We found the regulations to be incompatible with congressional intent. The regulations, we said, "constitute an impermissible construction of the statute under Chevron Step Two because the interpretation is utterly unreasonable in the breadth of its regulatory exclusion." 321 F.3d at 1178.

In this case, similarly, the Commission's interpretation of the term "capacity" in the statutory definition of an ATDS is "utterly unreasonable in the breadth of its regulatory [in]clusion." Id. Nothing in the TCPA countenances concluding that Congress could have contemplated the applicability of the statute's restrictions to the most commonplace phone device used every day by the overwhelming majority of Americans.

The Commission suggested in its ruling that, unless "capacity" reached so broadly, "little or no modern dialing equipment would fit the statutory definition." 2015 Declaratory Ruling, 30 FCC Rcd. at 7976 ¶ 20. But Congress need not be presumed to have intended the term "automatic telephone dialing system" to maintain its applicability to modern phone equipment in perpetuity, regardless of technological advances that may render the term increasingly inapplicable over time. After all, the statute also generally prohibits nonconsensual calls to numbers associated with a "paging service" or "specialized mobile radio service," 47 U.S.C. § 227(b)(1)(A)(iii), yet those terms have largely ceased to have practical significance.

In any event, the Commission retains a measure of authority under the TCPA to fashion exemptions to the restrictions on use of autodialers to call wireless numbers. Id. § 227(b)(2)(C). The agency presumably could, if needed, fashion exemptions preventing a result under which every uninvited call or message from a standard smartphone would violate the statute.

d. In its briefing before our court, the Commission now submits that its order in fact did not reach a definitive resolution on whether smartphones qualify as autodialers. As we have explained, however, a straightforward reading of the Commission's ruling invites the conclusion that all smartphones are autodialers: the ruling explained that a number of parties specifically raised the issue; and it responded, not by disputing the parties' concerns that smartphones would be covered by the statutory definition under the agency's approach, but instead by accepting that conclusion and then questioning whether uninvited calls in fact would be made and lawsuits in fact would be brought.

It is highly difficult to read the Commission's ruling to leave uncertain whether the statutory definition applies to smartphones. And any uncertainty on that score would have left affected parties without concrete guidance even though several of them specifically raised the issue with the agency, and even though the issue carries significant implications — including the possibility of committing federal law violations and incurring substantial liability in damages — for smartphone owners.

At any rate, even assuming the Commission's ruling could be conceived to leave room for concluding that smartphones do not qualify as autodialers, that result itself would be unreasonable and impermissible. The Commission's order, in that event, would not constitute reasoned decisionmaking and thus would not satisfy APA arbitrary-and-capricious review. See United States Postal Serv. v. Postal Regulatory Comm'n, 785 F.3d 740, 754 (D.C. Cir. 2015).

Administrative action is "arbitrary and capricious [if] it fails to articulate a comprehensible standard" for assessing the applicability of a statutory category. Id. at 753. If a "purported standard is indiscriminate and offers no meaningful guidance" to affected parties, it will fail "the requirement of reasoned decisionmaking." Id. at 754. That will be the case if an agency cannot satisfactorily explain why a challenged standard embraces one potential application but leaves out another, seemingly similar one. See id. at 754-55.

That would be precisely the situation here if, as the Commission now contends in its briefing before us, its order in fact left open the possibility that smartphones fail to meet the statutory definition of an ATDS. In the same briefing, the Commission, as noted, simultaneously maintained that the Firefox browser has the "capacity" to play Flash videos because the Flash plug-in can be downloaded. Precisely the same logic seemingly should compel concluding that smartphones have the "capacity" to function as autodialers because apps carrying the requisite features can be downloaded. If the Commission believes smartphones nonetheless do not meet the definition of an autodialer, there is no explanation of "this differential treatment of seemingly like cases." Id. at 755 (internal quotation marks omitted).

The Commission did say in its order that "there must be more than a theoretical potential that the equipment could be modified to satisfy the `autodialer' definition." 2015 Declaratory Ruling, 30 FCC Rcd. at 7975 ¶ 18. But that ostensible limitation affords no ground for distinguishing between a smartphone and the Firefox browser. In light of the ease of downloading an app to a smartphone, there is no evident basis for concluding that the Firefox browser has more than a mere "theoretical potential" to play Flash videos by downloading a plug-in, but a smartphone nonetheless has only a "theoretical potential" to function as an autodialer by downloading an app.

The point is fortified by the sole example of a mere "theoretical potential" set forth by the Commission in its order. That example involves a traditional rotary-dial phone (which by now is approaching obsolescence): the Commission observed that "it might be theoretically possible to modify a rotary-dial telephone to such an extreme that it would satisfy the definition of `autodialer,' but such a possibility is too attenuated ... to find that a rotary-dial phone has the requisite `capacity' and therefore is an autodialer." Id. A rotary phone has no relevant similarity to a smartphone. To the contrary, whereas a smartphone and the Firefox browser substantially resemble one another in their amenability to an upgrade via the addition of software, they substantially differ in that regard from a rotary-dial phone, which has no such capability.

In the end, then, the Commission's order cannot reasonably be understood to support the conclusion that smartphones fall outside the TCPA's autodialer definition: any such reading would compel concluding that the agency's ruling fails arbitrary-and-capricious review. The more straightforward understanding of the Commission's ruling is that all smartphones qualify as autodialers because they have the inherent "capacity" to gain ATDS functionality by downloading an app. That interpretation of the statute, for all the reasons explained, is an unreasonably, and impermissibly, expansive one.

2.

Recall that the statutory definition of an ATDS raises two sets of questions: (i) when does a device have the "capacity" to perform the functions of an autodialer enumerated by the statute?; and (ii) what precisely is the content of those functions? The impermissibility of the Commission's interpretation of the term "capacity" in the autodialer definition is compounded by inadequacies in the agency's explanation of the requisite features. Having addressed the first issue, we now turn to the second one.

a. As a threshold matter, the Commission maintains that the court lacks jurisdiction to entertain petitioners' challenge concerning the functions a device must be able to perform. The agency reasons that the issue was resolved in prior agency orders — specifically, declaratory rulings in 2003 and 2008 concluding that the statutory definition of an ATDS includes "predictive dialers," dialing equipment that can make use of algorithms to "assist[] telemarketers in predicting when a sales agent will be available to take calls." 2015 Declaratory Ruling, 30 FCC Rcd. at 7972 ¶ 10 n.39; see also In re Rules and Regulations Implementing the Telephone Consumer Protection Act of 1991 (2008 Declaratory Ruling), 23 FCC Rcd. 559 (2008); 2003 Order, 18 FCC Rcd. 14,014. According to the Commission, because there was no timely appeal from those previous orders, it is too late now to raise a challenge by seeking review of a more recent declaratory ruling that essentially ratifies the previous ones. We disagree.

While the Commission's latest ruling purports to reaffirm the prior orders, that does not shield the agency's pertinent pronouncements from review. The agency's prior rulings left significant uncertainty about the precise functions an autodialer must have the capacity to perform. Petitioners covered their bases by filing petitions for both a declaratory ruling and a rulemaking concerning that issue and related ones. See, e.g., Prof'l Ass'n for Customer Engagement, Inc. Pet. 3-4; ACA Int'l Pet. 6; GroupMe, Inc. Pet. 3; Glide Talk, Ltd. Pet. 13. In response, the Commission issued a declaratory ruling that purported to "provid[e] clarification on the definition of `autodialer,'" and denied the petitions for rulemaking on the issue. 2015 Declaratory Ruling, 30 FCC Rcd. at 8039 ¶ 165 & n.552. The ruling is thus reviewable on both grounds. See 5 U.S.C. § 554(e); Biggerstaff v. FCC, 511 F.3d 178, 184-85 (D.C. Cir. 2007).

b. The statutory definition says that a device constitutes an ATDS if it has the capacity to perform both of two enumerated functions: "to store or produce telephone numbers to be called, using a random or sequential number generator"; and "to dial such numbers." 47 U.S.C. § 227(a)(1)(A)-(B). The role of the phrase, "using a random or sequential number generator," has generated substantial questions over the years. The Commission has sought to address those questions in previous orders and did so again in the 2015 Declaratory Ruling we consider here.

The Commission's most recent effort falls short of reasoned decisionmaking in "offer[ing] no meaningful guidance" to affected parties in material respects on whether their equipment is subject to the statute's autodialer restrictions. Postal Regulatory Comm'n, 785 F.3d at 754. A basic question raised by the statutory definition is whether a device must itself have the ability to generate random or sequential telephone numbers to be dialed. Or is it enough if the device can call from a database of telephone numbers generated elsewhere? The Commission's ruling appears to be of two minds on the issue.

In certain respects, the order conveys that equipment needs to have the ability to generate random or sequential numbers that it can then dial. The order twice states that, to "meet[] the TCPA's definition of `autodialer,'" the equipment in question must have the capacity to "dial random or sequential numbers." 2015 Declaratory Ruling, 30 FCC Rcd. at 7972 ¶ 10; see also id. at 7974 ¶ 15. And it is clear from context that the order treats the ability to "dial random or sequential numbers" as the ability to generate and then dial "random or sequential numbers."

To see why, it is helpful to understand that the ruling distinguishes between use of equipment to "dial random or sequential numbers" and use of equipment to "call[] a set list of consumers." Id. at 7972 ¶ 10. Anytime phone numbers are dialed from a set list, the database of numbers must be called in some order — either in a random or some other sequence. As a result, the ruling's reference to "dialing random or sequential numbers" cannot simply mean dialing from a set list of numbers in random or other sequential order: if that were so, there would be no difference between "dialing random or sequential numbers" and "dialing a set list of numbers," even though the ruling draws a divide between the two. See id. at 7973 ¶¶ 13, 14. It follows that the ruling's reference to "dialing random or sequential numbers" means generating those numbers and then dialing them.

The Commission's prior declaratory rulings reinforce that understanding. In its 2003 ruling addressing predictive dialers, the Commission observed that, "[i]n the past, telemarketers may have used dialing equipment to create and dial 10-digit telephone numbers arbitrarily." 2003 Order, 18 FCC Rcd. at 14,092 ¶ 132 (emphasis added). But the industry had "progressed to the point where" it had become "far more cost effective" instead to "us[e] lists of numbers." Id. Again, the Commission suggested it saw a difference between calling from a list of numbers, on one hand, and "creating and dialing" a random or arbitrary list of numbers, on the other hand. Or as the Commission has elsewhere said, numbers that are "randomly or sequentially generated" differ from numbers that "come from a calling list." In re Implementation of the Middle Class Tax Relief and Job Creation Act of 2012, 27 FCC Rcd. 13,615, 13,629 ¶ 29 (2012) (quoted in 2015 Declaratory Ruling, 30 FCC Rcd. at 8077 (Comm'r Pai, dissenting)).

While the 2015 ruling indicates in certain places that a device must be able to generate and dial random or sequential numbers to meet the TCPA's definition of an autodialer, it also suggests a competing view: that equipment can meet the statutory definition even if it lacks that capacity. The Commission reaffirmed its 2003 ruling insofar as that order had found predictive dialers to qualify as ATDSs. 2015 Declaratory Ruling, 30 FCC Rcd. at 7972-73 ¶¶ 12-14. And in the 2003 order, the Commission had made clear that, while some predictive dialers cannot be programmed to generate random or sequential phone numbers, they still satisfy the statutory definition of an ATDS. 2003 Order, 18 FCC Rcd. at 14,091 ¶ 131 n.432; id. at 14,093 ¶ 133. By reaffirming that conclusion in its 2015 ruling, the Commission supported the notion that a device can be considered an autodialer even if it has no capacity itself to generate random or sequential numbers (and instead can only dial from an externally supplied set of numbers). The 2015 ruling correspondingly expresses that "predictive dialers" can differ from other "dialers that utilize random or sequential numbers instead of a list of numbers." 2015 Declaratory Ruling, 30 FCC Rcd. at 7973 ¶ 14.

So which is it: does a device qualify as an ATDS only if it can generate random or sequential numbers to be dialed, or can it so qualify even if it lacks that capacity? The 2015 ruling, while speaking to the question in several ways, gives no clear answer (and in fact seems to give both answers). It might be permissible for the Commission to adopt either interpretation. But the Commission cannot, consistent with reasoned decisionmaking, espouse both competing interpretations in the same order.

The choice between the interpretations is not without practical significance. Petitioners and various amici describe calling equipment that they wish to use to call set lists of cellular numbers without any generation of random or sequential numbers. See ACA Int'l Reply Br. 21; Am. Bankers Ass'n Amicus Br. 29-30. And at least some predictive dialers, as explained, have no capacity to generate random or sequential numbers.

The uncertainty in the 2015 ruling, moreover, does not stop with the question of whether a device must be able to generate random or sequential numbers to meet the statutory definition. The ruling is also unclear about whether certain other referenced capabilities are necessary for a dialer to qualify as an ATDS.

For instance, the ruling states that the "basic function" of an autodialer is the ability to "dial numbers without human intervention." 2015 Declaratory Ruling, 30 FCC Rcd. at 7973 ¶ 14; id. at 7975 ¶ 17. Prior orders had said the same. 2003 Order, 18 FCC Rcd. at 14,092 ¶ 132; 2008 Declaratory Ruling, 23 FCC Rcd. at 566 ¶ 13. That makes sense given that "auto" in autodialer — or, equivalently, "automatic" in "automatic telephone dialing system," 47 U.S.C. § 227(a)(1) — would seem to envision non-manual dialing of telephone numbers.

But the Commission nevertheless declined a request to "clarify[] that a dialer is not an autodialer unless it has the capacity to dial numbers without human intervention." 2015 Declaratory Ruling, 30 FCC Rcd. at 7976 ¶ 20. According to the Commission, then, the "basic function" of an autodialer is to dial numbers without human intervention, but a device might still qualify as an autodialer even if it cannot dial numbers without human intervention. Those side-by-side propositions are difficult to square.

The Commission further said that another "basic function[]" of an ATDS is to "dial thousands of numbers in a short period of time." Id. at 7975 ¶ 17. But the ruling imparts no additional guidance concerning whether that is a necessary condition, a sufficient condition, a relevant condition even if neither necessary nor sufficient, or something else. Nor does it indicate what would qualify as a "short period of time." Again, affected parties are left in a significant fog of uncertainty about how to determine if a device is an ATDS so as to bring into play the restrictions on unconsented calls.

In short, the Commission's ruling, in describing the functions a device must perform to qualify as an autodialer, fails to satisfy the requirement of reasoned decisionmaking. The order's lack of clarity about which functions qualify a device as an autodialer compounds the unreasonableness of the Commission's expansive understanding of when a device has the "capacity" to perform the necessary functions. We must therefore set aside the Commission's treatment of those matters.

3.

We briefly note an additional statutory provision affecting the scope of the TCPA's restrictions on autodialer calls to cell numbers — a provision we ultimately have no occasion to examine because of the way the case has been presented to us. Two TCPA provisions work together to establish the reach of the general prohibition against making autodialer calls without prior consent. The first provision, as we have seen, defines the equipment — viz., "automatic telephone dialing system" — subject to the statutory prohibition. 47 U.S.C. § 227(a)(1). The second provision then incorporates that definition in setting out the scope of the prohibition: "It shall be unlawful for any person ... to make any call (other than a call made for emergency purposes or made with the prior express consent of the called party) using any automatic telephone dialing system... to any telephone number assigned to a... cellular telephone service[.]" Id. § 227(b)(1)(A)(iii) (emphases added).

Petitioners have confined their challenge to the Commission's understanding of the first of those provisions, the statutory definition of an autodialer, and our analysis has been focused on that issue. Petitioners have raised no challenge to the Commission's understanding of the second provision — i.e., to the agency's interpretation of what it means to "make any call using any" ATDS. In particular, in the case of a device having the "capacity" both to perform the autodialer functions set out in the statutory definition and to perform as a traditional phone, does the bar against "making any call using" an ATDS apply only to calls made using the equipment's ATDS functionality? Or does the bar apply to all calls made with a device having that "capacity," even ones made without any use of the equipment's autodialer capabilities? Or does the bar apply to calls made using certain autodialer functions, even if not all of them?

The Commission's ruling endorsed a broad understanding under which the statute prohibits any calls made from a device with the capacity to function as an autodialer, regardless of whether autodialer features are used to make a call. 2015 Declaratory Ruling, 30 FCC Rcd. at 7975 ¶ 19 n.70. A dissenting commissioner, by contrast, read the pertinent statutory phrase, "make any call," to mean "that the equipment must, in fact, be used as an autodialer to make the calls" before a TCPA violation can be found. Id. at 8088 (Comm'r O'Rielly, dissenting in part and approving in part).

The dissenting commissioner's interpretation would substantially diminish the practical significance of the Commission's expansive understanding of "capacity" in the autodialer definition. Even if the definition encompasses any device capable of gaining autodialer functionality through the downloading of software, the mere possibility of adding those features would not matter unless they were downloaded and used to make calls. Under the dissent's understanding of the phrase, "make any call," then, everyday calls made with a smartphone would not infringe the statute: the fact that a smartphone could be configured to function as an autodialer would not matter unless the relevant software in fact were loaded onto the phone and were used to initiate calls or send messages.

Petitioners, however, raise no challenge to the Commission's understanding of the statutory words, "make any call using" an ATDS, and the parties therefore have not presented arguments on the issue in their briefing before us. Our consistent practice in such a situation is to decline to address (much less resolve) the issue. See, e.g., U.S. Telecom Ass'n v. FCC, 825 F.3d 674, 697 (D.C. Cir. 2016). We "sit to resolve only legal questions presented and argued by the parties." Id. (internal quotation marks omitted). We nonetheless note the issue in light of its potential interplay with the distinct challenges petitioners do raise. The agency could choose to revisit the issue in a future rulemaking or declaratory order, and a party might then raise the issue on judicial review.

B.

We now turn to the Commission's treatment of circumstances in which a consenting party's cell number has been reassigned to another person. While there is no consensus about the exact numbers of reassignments, there is no dispute that millions of wireless numbers are reassigned each year. In the event of a reassignment, the caller might initiate a phone call (or send a text message) based on a mistaken belief that the owner of the receiving number has given consent, when in fact the number has been reassigned to someone else from whom consent has not been obtained.

Does a call or message in that situation violate the statutory bar against making autodialer calls without prior consent? The Commission's answer is yes, apart from a one-call, post-reassignment safe harbor. We set aside the Commission's interpretation on the ground that the one-call safe harbor is arbitrary and capricious.

1.

The pertinent statutory language generally renders it unlawful "to make any call (other than a call made for emergency purposes or made with the prior express consent of the called party) using any automatic telephone dialing equipment or prerecorded voice." 47 U.S.C. § 227(b)(1)(A) (emphasis added). The Commission, in its ruling, initially addressed who is properly considered the "called party" when a consenting party's number is reassigned to another person: does "called party" refer to the person the caller expected to reach (whose consent had previously been obtained), or does it refer to the person actually reached, the wireless number's present-day subscriber after reassignment (whose consent has not been obtained)?

The Commission adopted the latter interpretation. 30 FCC Rcd. at 7999-8001 ¶¶ 72-73. The result is that the reassignment of a wireless number extinguishes any consent given by the number's previous holder and exposes the caller to liability for reaching a party who has not given consent. An alternative approach, the Commission reasoned, would "effectively require consumers to opt out of such calls when the TCPA clearly requires the opposite — that consumers opt in before they can be contacted." Id. at 8004 ¶ 80.

The agency also refused to "place any affirmative obligation" on new subscribers to inform callers that a wireless number now belongs to someone else. Id. at 8011 ¶ 95. The ruling thus expressly contemplates that a new subscriber could "purposefully and unreasonably" refrain from informing a good-faith caller about a number's reassignment "in order to accrue statutory penalties." Id. (formatting modified). In that regard, the Commission described a reported case in which the new, post-reassignment subscriber waited to initiate a lawsuit until after having received almost 900 text alerts that were intended for the previous subscriber. Id. at 8011 ¶ 94 & n.324.

The Commission acknowledged that even the most careful caller, after employing all reasonably available tools to learn about reassignments, "may nevertheless not learn of reassignment before placing a call to a new subscriber." Id. at 8009 ¶ 88. The Commission observed that it nonetheless "could have interpreted the TCPA to impose a traditional strict liability standard on the caller: i.e., a `zero call' approach under which no allowance would have been given for the robocaller to learn of the reassignment." Id. at 8009 ¶ 90 n.312. But the Commission declined to interpret the statute "to require a result that severe." Id. Rather, the Commission read the statute to "anticipate[] the caller's ability to rely on prior express consent," which the Commission interpreted "to mean reasonable reliance." Id. (internal quotation marks omitted).

The Commission effectuated its "reasonable reliance" approach by enabling a caller who lacks knowledge of a reassignment "to avoid liability for the first call to a wireless number following reassignment." Id. at 8009 ¶ 89. For that first call, the caller can continue to rely on the consent given by the "previous subscriber." Id. at 8003 ¶ 78. The Commission did "not presume that a single call to a reassigned number will always be sufficient for callers to gain actual knowledge of the reassignment." Id. at 8009 ¶ 90 n.312. But it believed that "[o]ne call represents an appropriate balance between a caller's opportunity to learn of the reassignment and the privacy interests of the new subscriber." Id. at 8009 ¶ 90.

2.

In challenging the Commission's resolution, petitioners first contend that the statutory reference to the consent of the "called party" refers to the expected recipient of a call or message, not the actual recipient. When a wireless number is reassigned without the caller's awareness, petitioners' interpretation would mean that a caller would avoid liability for a post-reassignment call because the "called party" — the former owner of the number — had given consent. In petitioners' view, the Commission's contrary interpretation of "called party" to refer to the new (post-reassignment) subscriber is foreclosed by the statute. We disagree.

Another court of appeals has examined the meaning of the term "called party" in the same statutory provision, 47 U.S.C. § 227(b)(1)(A), and in the same situation of a reassigned wireless number formerly belonging to a consenting party. Soppet v. Enhanced Recovery Co., 679 F.3d 637 (7th Cir. 2012). The Seventh Circuit explained that the phrase "called party" appears throughout the broader statutory section, 47 U.S.C. § 227, a total of seven times. 679 F.3d at 640. Four of those instances "unmistakably denote the current subscriber," not the previous, pre-reassignment subscriber. Id. Of the three remaining instances, "one denotes whoever answers the call (usually the [current] subscriber)," and the other two are unclear. Id. By contrast, the court observed, the "phrase `intended recipient' does not appear anywhere in § 227, so what justification could there be for equating `called party' with `intended recipient of the call'?" Id. For those and other reasons, the court concluded "that `called party' in § 227(b)(1) means the person subscribing to the called number at the time the call is made," not the previous subscriber who had given consent. Id. at 643; see also Osorio v. State Farm Bank, F.S.B., 746 F.3d 1242, 1250-52 (11th Cir. 2014).

We find the Seventh Circuit's analysis persuasive insofar as it supports concluding that the Commission was not compelled to interpret "called party" in § 227(b)(1)(A) to mean the "intended recipient" rather than the current subscriber. The Commission thus could permissibly interpret "called party" in that provision to refer to the current subscriber.

3.

Petitioners next argue that the Commission's one-call safe harbor is arbitrary. On this score, we agree with petitioners.

When a caller is unaware that a consenting party's wireless number has been reassigned, the Commission chose to allow the caller to make one (and only one) post-reassignment call without incurring liability. For that one call, the Commission understood the statutory term "prior express consent" to refer to the consent given by the previous subscriber. 30 FCC Rcd. at 8001 ¶ 73 & n.265; id. at 8003 ¶ 78.

The Commission allowed for that one liability-free call, rather than impose "a traditional strict liability standard," because it interpreted a caller's ability under the statute to rely on a recipient's "prior express consent" to "mean reasonable reliance." Id. at 8009 ¶ 90 n.312. And when a caller has no knowledge of a reassignment, the Commission understandably viewed the caller's continued reliance on the prior subscriber's consent to be "reasonable."

Elsewhere in the Declaratory Ruling, the Commission echoed the same "reasonable reliance" understanding of the statute's approval of calls based on "prior express consent." The ruling accepts that a caller can rely on consent given by a wireless number's "customary user" ("such as a close relative on a subscriber's family calling plan"), rather than by the subscriber herself. Id. at 8001 ¶ 75. That is because the "caller in this situation cannot reasonably be expected to divine that the consenting person is not the subscriber." Id. at 8001-02 ¶ 75. The Commission reiterated in that regard that, in "construing the term `prior express consent' in section 227(b)(1)(A), we consider the caller's reasonableness in relying on consent." Id. at 8001 ¶ 75.

The Commission thus consistently adopted a "reasonable reliance" approach when interpreting the TCPA's approval of calls based on "prior express consent," including as the justification for allowing a one-call safe harbor when a consenting party's number is reassigned. The Commission, though, gave no explanation of why reasonable-reliance considerations would support limiting the safe harbor to just one call or message. That is, why does a caller's reasonable reliance on a previous subscriber's consent necessarily cease to be reasonable once there has been a single, post-reassignment call? The first call or text message, after all, might give the caller no indication whatsoever of a possible reassignment (if, for instance, there is no response to a text message, as would often be the case with or without a reassignment).

The Commission outlined a number of measures callers could undertake "that, over time, may permit them to learn of reassigned numbers." Id. at 8007 ¶ 86. But the Commission acknowledged that callers "may nevertheless not learn of reassignment before placing a call to a new subscriber," and that the first post-reassignment call likewise might give no reason to suspect a reassignment. Id. at 8009 ¶¶ 88, 90 n.312. In that event, a caller's reasonable reliance on the previous subscriber's consent would be just as reasonable for a second call.

To be sure, the Commission stated that it found "no basis in the statute or the record before [it] to conclude that callers can reasonably rely on prior express consent beyond one call to reassigned numbers." Id. at 8009-10 ¶ 90 n.312. But the Commission did not elaborate on — or otherwise support — its conclusory observation to that effect. And the statement is hard to square with the Commission's concession that the first call may give no notice of a reassignment, or with the Commission's disavowal of any expectation that a caller should "divine from the called consumer's mere silence the current status of a telephone number." Id. (brackets omitted). In that light, no cognizable conception of "reasonable reliance" supports the Commission's blanket, one-call-only allowance.

At times, the Commission indicated that its one-call safe harbor intends to give callers additional "opportunity" to find out about a possible reassignment. E.g., id. at 8009 ¶ 89; id. at 8010 ¶ 91. There is no indication, though, that the interest in giving callers such an opportunity is independent of the interest in giving effect to a caller's reasonable reliance. After all, a caller also has an opportunity to learn of a reassignment before the first call. The reason to allow even one, liability-free, post-reassignment call — the reason the Commission cared about affording an opportunity to learn about reassignment at all — is in order to give effect to a caller's reasonable reliance on the previous subscriber's consent.

Indeed, the Commission's one-call safe harbor applies "over an unlimited period of time." Id. at 8000 ¶ 72 n.257. If the goal were simply to provide an expanded opportunity to learn about a reassignment, the Commission presumably would have allowed for a given period of time. It declined to do so, id. at 8009 ¶ 89, opting instead to permit a single call regardless of whether it occurs within minutes or months of a reassignment.

For substantially the same reasons, the Commission's one-call-only approach cannot be salvaged by its suggestion that callers rather than new subscribers should bear the risk when calls are made (or messages are sent) to a reassigned number. Id. at 8009-10 ¶ 90 n.312. That consideration would equally support a zero-call, strict-liability rule. But the Commission specifically declined to adopt "a result that severe." Id. Having instead embraced an interpretation of the statutory phrase "prior express consent" grounded in conceptions of reasonable reliance, the Commission needed to give some reasoned (and reasonable) explanation of why its safe harbor stopped at the seemingly arbitrary point of a single call or message. The Commission did not do so.

The Seventh Circuit's decision in Soppet, discussed earlier, is not to the contrary. There, the court assumed that "any consent previously given ... lapses when [a] [c]ell [n]umber is reassigned." 679 F.3d at 641. The court, though, did not have before it an agency interpretation under which the previous subscriber's consent does not lapse with reassignment: the premise of the Commission's one-call safe harbor is that a caller can continue to rely on the previous subscriber's consent. The question we face is, why should that necessarily stop with a single call? Soppet does not speak to that question, and so does not cast doubt on our conclusion that the Commission failed to give it a satisfactory answer.

Finally, the Commission's failure in that regard requires setting aside not only its allowance of a one-call safe harbor, but also its treatment of reassigned numbers more generally. When we invalidate a specific aspect of an agency's action, we leave related components of the agency's action standing only if "we can say without any `substantial doubt' that the agency would have adopted the severed portion on its own." Am. Petroleum Inst. v. EPA, 862 F.3d 50, 71 (D.C. Cir. 2017) (per curiam) (internal quotation marks omitted).

Here, we have no such assurance. If we were to excise the Commission's one-call safe harbor alone, that would leave in place the Commission's interpretation that "called party" refers to the new subscriber. And that in turn would mean that a caller is strictly liable for all calls made to the reassigned number, even if she has no knowledge of the reassignment.

We cannot be certain that the agency would have adopted that rule in the first instance. Significantly, the Commission said that it "could have interpreted the TCPA to impose a traditional strict liability standard," i.e., "a `zero call' approach." 30 FCC Rcd. at 8009 ¶ 90 n.312. But the agency declined to "require a result that severe," opting instead for a one-call safe harbor. Id. We cannot say without any substantial doubt that the agency would have embraced the "severe" implications of a pure, strict-liability regime even in the absence of any safe harbor. As a result, we must set aside the Commission's treatment of reassigned numbers as a whole.

Notably, the Commission is already on its way to designing a regime to avoid the problems of the 2015 ruling's one-call safe harbor. The Commission recently sought comment on potential methods for "requir[ing] service providers to report information about number reassignments for the purposes of reducing unwanted robocalls." In re Advanced Methods to Target and Eliminate Unlawful Robocalls, Second Notice of Inquiry, 32 FCC Rcd. 6007, 6010 ¶ 9 (2017). Most of its proposals envision creating a comprehensive repository of information about reassigned wireless numbers. See id. at 6012-13 ¶¶ 15-19. The Commission is also considering whether to provide a safe harbor for callers that inadvertently reach reassigned numbers after consulting the most recently updated information. See id. at 6012 ¶ 14. Those proposals would naturally bear on the reasonableness of calling numbers that have in fact been reassigned, and have greater potential to give full effect to the Commission's principle of reasonable reliance.

C.

It is undisputed that consumers who have consented to receiving calls otherwise forbidden by the TCPA are entitled to revoke their consent. See 2015 Declaratory Ruling, 30 FCC Rcd. at 7996 ¶ 62. The statute, however, does not elaborate on the processes by which consumers may validly do so. The Commission sought to resolve the matter in its Declaratory Ruling.

The Commission had been petitioned to clarify that callers can unilaterally prescribe the exclusive means for consumers to revoke their consent. It explicitly denied that request. Allowing "callers to designate the exclusive means of revocation," the Commission believed, could "materially impair" the "right of revocation." Id. at 7997 ¶ 66.

The Commission instead concluded that "a called party may revoke consent at any time and through any reasonable means" — orally or in writing — "that clearly expresses a desire not to receive further messages." Id. at 7989-90 ¶ 47; id. at 7996 ¶ 63. In assessing whether a revocation request meets the "reasonable means" standard, the Commission said it would consider "the totality of the facts and circumstances." Id. at 7996 ¶ 64 n.233. One relevant factor is "whether the caller could have implemented mechanisms to effectuate a requested revocation without incurring undue burdens." Id. Another consideration is "whether the consumer had a reasonable expectation that he or she could effectively communicate his or her request ... in that circumstance." Id.

Petitioners challenge the Commission's treatment of revocations on various grounds, none of which we find persuasive. Petitioners' chief objection is that the Commission's approach is arbitrary and capricious in eschewing the establishment of standardized revocation procedures in favor of an unduly uncertain, any-reasonable-means standard. Without the certainty of standardized procedures, petitioners fear, they will be able to ward off TCPA liability only by "tak[ing] exorbitant precautions." ACA Int'l Br. 57.

We think petitioners' concerns are overstated. The Commission's ruling absolves callers of any responsibility to adopt systems that would entail "undue burdens" or would be "overly burdensome to implement." 30 FCC Rcd. at 7996 ¶ 64 & n.233. In light of that assurance, callers would have no need to train every retail employee on the finer points of revocation. And callers will have every incentive to avoid TCPA liability by making available clearly-defined and easy-to-use opt-out methods. If recipients are afforded such options, any effort to sidestep the available methods in favor of idiosyncratic or imaginative revocation requests might well be seen as unreasonable. The selection of an unconventional method of seeking revocation might also betray the absence of any "reasonable expectation" by the consumer that she could "effectively communicate" a revocation request in the chosen fashion. Id.

Petitioners observe that the Commission's ruling itself dictates particular opt-out mechanisms for certain types of time-sensitive banking- and healthcare-related calls that the Commission exempted from the TCPA's consumer consent requirements. Id. at 8028 ¶ 138; id. at 8032 ¶ 147. If the Commission prescribed specific opt-out methods for those types of calls, petitioners ask, then why not similarly set out standardized means of revocation for all calls?

The Commission was not required to treat the two situations in a parallel manner. For the banking- and healthcare-related calls, the Commission found that the communications were sufficiently important to warrant an exemption from the otherwise-applicable obligation to obtain prior consent. Id. at 8023 ¶ 125. As a result, the default rule for those calls is that they should be allowed (without regard to consent), such that the availability of an opt-out can be conditioned on adhering to specific procedures. By contrast, the default rule for non-exempted calls is that they are disallowed (absent consent), such that the availability of an opt-out naturally could be broader. In that context, the Commission could reasonably elect to enable consumers to revoke their consent without having to adhere to specific procedures.

Finally, petitioners object to the Declaratory Ruling insofar as it might preclude callers and consumers from contractually agreeing to revocation mechanisms. The Commission correctly concedes, however, that the ruling "did not address whether contracting parties can select a particular revocation procedure by mutual agreement." FCC Br. 64 n.16. The ruling precludes unilateral imposition of revocation rules by callers; it does not address revocation rules mutually adopted by contracting parties. Nothing in the Commission's order thus should be understood to speak to parties' ability to agree upon revocation procedures.

D.

The last set of challenges before us, brought by petitioner Rite Aid, concerns the scope of the Commission's exemption of certain healthcare-related calls from the TCPA's prior-consent requirement for calls to wireless numbers. The Commission is statutorily authorized to exempt from that requirement "calls to a telephone number assigned to a cellular telephone service that are not charged to the called party, subject to such conditions as the Commission may prescribe as necessary in the interest of the privacy rights this section is intended to protect." 47 U.S.C. § 227(b)(2)(C).

The Commission was petitioned to exempt from the consent requirement "certain non-telemarketing, healthcare calls" alleged to "provide vital, time-sensitive information patients welcome, expect, and often rely on to make informed decisions." 2015 Declaratory Ruling, 30 FCC Rcd. at 8030 ¶ 143. The agency acknowledged the "exigency and public interest" in various types of healthcare-related calls, including ones "regarding post-discharge follow-up intended to prevent readmission, or prescription notifications." Id. at 8031 ¶ 146. But it was "concerned that these policy arguments are not true" for other types of healthcare calls. Id. Specifically, the Commission "fail[ed] to see the same exigency and public interest in calls regarding account communications and payment notifications." Id.

Consequently, the Commission granted the requested exemption but "restrict[ed] it to calls for which there is exigency and that have a healthcare treatment purpose, specifically: appointment and exam confirmations and reminders, wellness checkups, hospital pre-registration instructions, pre-operative instructions, lab results, post-discharge follow-up intended to prevent readmission, prescription notifications, and home healthcare instructions." Id. The exemption would not cover calls "that include telemarketing, solicitation, or advertising content, or which include accounting, billing, debt-collection, or other financial content." Id.

Petitioner Rite Aid challenges the Commission's exemption for select healthcare-related calls on the grounds that it conflicts with another federal statute (the Health Insurance Portability and Accountability Act, or HIPAA) and is arbitrary and capricious. Rite Aid's arguments misunderstand the relevant statutory terrain, and we reject them.

1.

At the outset, we must satisfy ourselves that we have jurisdiction to entertain Rite Aid's challenge. Rite Aid has been styled a petitioner here, but it did not formally petition the Commission in the proceedings before the agency. The petition granted by the Commission in part was filed by the American Association of Healthcare Administrative Management (the Association). Rite Aid expressed "support" for the Association's petition for a declaratory ruling and exemption, and it also asked the Commission to "address certain additional issues." Comments of Rite Aid, Joint App'x 850. But it participated only by commenting on the Association's petition rather than filing one of its own. As a result, with respect to relief that only Rite Aid sought, the Commission "decline[d] to fully address th[at] request for clarification ... raised in a comment to a pending Petition." 2015 Declaratory Ruling, 30 FCC Rcd. at 8028-29 ¶ 141 n.471. The Association did not appeal the FCC's partial denial of its requested exemption. Instead, Rite Aid has petitioned the court to review that denial.

Direct review of final FCC orders is governed by the Hobbs Act, under which "[a]ny party aggrieved by [a] final order" of the Commission may petition for review of that order. 28 U.S.C. § 2344. We have consistently held that the phrase "party aggrieved" requires that petitioners have been parties to the underlying agency proceedings, not simply parties to the present suit who are aggrieved in a constitutional (Article III) sense. See Simmons v. ICC, 716 F.2d 40, 42 (D.C. Cir. 1983). The question here is whether commenting on a petition in agency proceedings that resulted in a declaratory ruling suffices to confer "party aggrieved" status on a litigant whose position the agency rejected.

We find it does. For agency proceedings that do not require intervention as a prerequisite to participation, our decisions have recognized that "party aggrieved" means a party who has "made a full presentation of views to the agency." Water Transp. Ass'n v. ICC, 819 F.2d 1189, 1193 (D.C. Cir. 1987). Rite Aid fulfilled that requirement. Just as "submitting comments" confers "party aggrieved" status in the context of a rulemaking (assuming an adverse outcome), Prof'l Reactor Operator Soc'y v. U.S. Nuclear Regulatory Comm'n, 939 F.2d 1047, 1049 n.1 (D.C. Cir. 1991), one who comments on another's petition for a rulemaking or declaratory ruling has "present[ed] its view to the agency [so as] to qualify as a `party,'" S. Pac. Transp. Co. v. ICC, 69 F.3d 583, 588 (D.C. Cir. 1995) — at least insofar as the issues appealed were also taken up by the petitioner below (as they were here). Rite Aid afforded the Commission an opportunity to consider its position on the Association's exemption request. We therefore proceed to the substance of Rite Aid's challenge.

2.

Rite Aid contends that, "[b]y restricting otherwise permissible HIPAA communications," the Declaratory Ruling "conflicts with another federal law." Rite Aid Br. 12 (quoting NextWave Pers. Commc'ns, Inc. v. FCC, 254 F.3d 130, 149 (D.C. Cir. 2001)). It essentially argues that any partial exemption of healthcare-related communications would have been unlawful, because HIPAA — the exclusive source of federal law on the disclosure of protected health information — operates of its own force to supersede any TCPA prohibition on healthcare calls. Rite-Aid is incorrect. There is no obstacle to complying with both the TCPA and HIPAA; "[t]he two statutes provide separate protections." Mais v. Gulf Coast Collection Bureau, Inc., 768 F.3d 1110, 1125 (11th Cir. 2014).

Under HIPAA regulations, covered entities and their business associates presumptively "may not use or disclose protected health information." 45 C.F.R. § 164.502(a). But they are generally permitted to use or disclose that information "for treatment, payment, or health care operations." Id. § 164.506(a). Rite Aid complains that the partial exemption granted in the Declaratory Ruling conflicts with HIPAA because it stops short of exempting billing- and account-related communications — i.e., ones "for ... payment." Id. But all that § 164.506(a)'s exclusion does is to carve out an exception to civil and criminal liability for using or disclosing protected health information. See 42 U.S.C. §§ 1320d-5, 1320d-6. It says nothing about the Commission's authority to exempt (or refrain from exempting) certain kinds of calls from the TCPA's consent requirement.

In confining the use of its exemption authority, the Commission did not restrict communications that HIPAA requires be permitted to flow freely. It simply declined to make certain exchanges even less burdensome than they would have been by default. If Rite Aid were correct, healthcare providers could use ATDS equipment to bombard nonconsenting wireless users with calls and texts concerning outstanding charges without incurring TCPA liability. Nothing in HIPAA commands such a result, and we see no basis to interpret it to frustrate the TCPA in that way.

3.

Finally, Rite Aid contends that the Declaratory Ruling's exemption for certain healthcare calls is arbitrary and capricious. Neither of its suggested grounds is persuasive.

a. Rite Aid first argues that the Commission failed to explain its purported departure from its earlier practice of exempting HIPAA-protected communications. In addition to its restrictions on calls to wireless numbers, the TCPA also forbids the use of an ATDS "to initiate any telephone call to any residential telephone line using an artificial or prerecorded voice to deliver a message without the prior express consent of the called party," unless one of three exceptions applies. 47 U.S.C. § 227(b)(1)(B) (emphasis added).

In a 2012 Order, the Commission exempted from that consent requirement "prerecorded health care-related calls to residential lines, which are already regulated by" HIPAA. In re Rules and Regulations Implementing the Telephone Consumer Protection Act of 1991 (2012 Order), 27 FCC Rcd. 1830, 1837 ¶ 18 (2012). Some parts of the Order suggested that its exemption reached no further than the one granted in 2015's Declaratory Ruling for calls to wireless numbers. Exempted calls were described as "promot[ing] important communications ... such as prescription refills and immunization reminders," id. at 1855 ¶ 63 n.192, and "concern[ing] consumers' health, not the purchase of a good or service," id. at 1856 ¶ 63 n.195. But the Order elsewhere characterized its exemption as covering "all prerecorded health care-related calls to residential lines that are subject to HIPAA." Id. at 1852 ¶ 57 (emphases added).

The 2012 Order's exemption was codified in 47 C.F.R. § 64.1200(a)(3)(v). That regulation did not use the phrase "health care-related call[]," but instead referred to "`health care' message ... [as] defined in the HIPAA Privacy Rule, 45 C.F.R. § 160.103." 47 C.F.R. § 64.1200(a)(3)(v). Likewise, § 160.103 does not mention the term "health care message." But it does define "health care" as "care, services, or supplies related to the health of an individual." 45 C.F.R. § 160.103. That term includes, among many other things, "[s]ale or dispensing of a drug, device, equipment, or other item in accordance with a prescription." Id. A "`health care' message" is presumably a message pertaining to any of the topics that "health care" is defined to include. We assume for present purposes that some calls concerning the "[s]ale ... of a drug ... in accordance with a prescription" would relate to "billing," which the 2015 Declaratory Ruling did not exempt from the consent requirement.

Rite Aid is therefore correct that, in one sense, the 2012 exemption swept more broadly than the 2015 version. We also accept that the 2012 Order cited a number of "technology-agnostic justifications" for exempting all prerecorded healthcare-related calls subject to HIPAA and made to residential lines. Rite Aid Br. 5. For example, the Commission believed that such calls "ensure continued customer access to health care-related information" and would not lead to "coercive or abusive" interactions. 2012 Order, 27 FCC Rcd. at 1853-54 ¶¶ 59-60.

The relevant question is whether the Commission acted arbitrarily and capriciously in affording a narrower exemption for healthcare-related calls made to wireless numbers. We find that it did not. Even if one might hypothesize "important reasons for treating residential and wireless telephone lines the same," Rite Aid Br. 9, the TCPA itself presupposes the contrary — that calls to residential and wireless numbers warrant differential treatment.

Unlike with the autodialer restrictions on calls to wireless numbers, callers are free to use ATDS equipment to dial residential lines as long as no "artificial or prerecorded voice" is used. 47 U.S.C. § 227(b)(1)(B). The statute itself contemplates that calls to wireless numbers "tread [more] heavily upon ... consumer privacy interests." 2012 Order, 27 FCC Rcd. at 1855 ¶ 63. That concern directly informed the 2015 exemption's scope: the Commission concluded that messages "not critical to a called party's healthcare ... do not justify setting aside a consumer's privacy interests." 2015 Declaratory Ruling, 30 FCC Rcd. at 8031 ¶ 146.

In short, there is nothing inherently contradictory about easing restrictions on certain kinds of calls to landlines, but not to cellular phones. And Rite Aid fails to mention another variable that confounds direct comparisons between the two exemptions. As codified, the 2012 exemption applies only to calls that "us[e] an artificial or prerecorded voice to deliver a message," 47 C.F.R. § 64.1200(a)(3); the Declaratory Ruling's exemption is not so limited. We therefore reject Rite Aid's first arbitrary-and-capricious challenge.

b. Lastly, Rite Aid argues that the Commission acted arbitrarily by failing to recognize that all healthcare-related calls satisfy the TCPA's "emergency purposes" exception to the consent requirement. As used in the Act, "[t]he term emergency purposes means calls made necessary in any situation affecting the health and safety of consumers." 47 C.F.R. § 64.1200(f)(4). But Rite Aid identifies no calls satisfying that exception that were not already subject to the 2015 exemption. It would be implausible to conclude that calls concerning "telemarking, solicitation, or advertising content, or which include accounting, billing, debt-collection, or other financial content" are made for "emergency purposes." 2015 Declaratory Ruling, 30 FCC Rcd. at 8031 ¶ 146. Even if accounting systems are in some sense "necessary" to the continued provision of healthcare, "[t]imely delivery of these types of messages is not critical" to that goal. Id. (emphasis added).

In marked contrast, the Commission recently exempted calls concerning certain time-sensitive risks to students' health and safety in the school setting. That list of scenarios included "weather closures, fire,... threats," "dangerous persons, health risks (e.g., toxic spills), and unexcused absences." In re Rules and Regulations Implementing the Telephone Consumer Protection Act of 1991, 31 FCC Rcd. 9054, 9061 ¶ 17, 9063 ¶ 21 (2016). In declining a request to interpret the emergency-purposes exception far more expansively, we are guided by its role in the statutory scheme. Consumers may find themselves wholly unable to stave off calls satisfying the exception. That is because, by definition, such calls fall outside the TCPA's consent framework; callers can make them even if recipients are known to object. Advertisements, solicitations, and post-treatment financial communications do not arise from the sorts of "emergencies" that would justify suspending the TCPA's consent regime.

The Commission was empowered to draw the distinction it did, and it adequately explained its reasons for doing so. We therefore reject Rite Aid's arbitrary-and-capricious challenge.

* * * * *

For the foregoing reasons, we grant in part and deny in part the petitions for review.

So ordered.