Defining Automatic Telephone Dialing System
The dispute over the ATDS definition might surprise someone simply skimming the statute. The TCPA defines 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). Most lay readers would likely understand that a device is not an ATDS unless it uses a random or sequential number generator. But the siren song of uncapped statutory damages has prompted plaintiffs to proffer a number of atextual readings of the statute, chief among them that anything is an ATDS if it can store and then dial such numbers. Under that reading, whether equipment uses a random or sequential number generator is irrelevant.
This issue has caused deep division in the lower courts, with the Third, Seventh, and Eleventh Circuits reading the definition narrowly; the Second, Sixth, and Ninth Circuits reading it broadly; and businesses wondering how to comply with a statute whose scope varies so wildly from circuit to circuit. Compare Dominguez v. Yahoo, Inc., 894 F.3d 116, 121 (3d Cir. 2018), Gadelhak v. AT&T Servs., Inc., 950 F.3d 458 (7th Cir. 2020), and Glasser v. Hilton Grand Vacations Co., 948 F.3d 1301 (11th Cir. 2020), with Duran v. La Boom Disco, Inc., 955 F.3d 279 (2d Cir. 2020), Allan v. Pa. Higher Educ. Assistance Agency, 968 F.3d 567 (6th Cir. 2020), and Marks v. Crunch San Diego, LLC, 904 F.3d 1041 (9th Cir. 2018).
What constitutes an ATDS finally came to a head in Facebook v. Duguid, in which the Supreme Court heard argument in December 2020 and should rule by June 2021.
Using a Random or Sequential Number Generator
The ATDS definition matters because most modern callers do not use random or sequential numbers but rather use stored numbers—for example, lists of customers or borrowers. That was not so when Congress enacted the TCPA in 1991. Congress gave random and sequential number generation special treatment because they were causing special harms: random dialing was reaching unlisted numbers and emergency numbers; sequential dialing was causing the equivalent of a denial-of-service attack at hospitals and similar facilities; and both kinds of dialing were reaching wireless numbers, which at the time incurred substantial per-minute charges for incoming calls.
The TCPA was not, then, a “privacy” statute that required consent for every autodialed call. We know that because the overwhelming majority of numbers in 1991 were residential rather than wireless. And yet Congress chose not to require consent for autodialed calls to residential numbers. Instead, it required consent for calls to residential numbers only if they used an automated voice, and it allowed consumers to avoid other calls to their residences by registering on the Do-Not-Call Registry. See 47 U.S.C. §§ 227(b)(1)(B), 227(c). As counsel for Facebook explained at oral argument, if Congress had meant to restrict any device that could “store and dial numbers, its failure to protect the home front would be inexplicable.”
The FCC understood at the time that merely dialing from a stored list would not trigger the statute’s restrictions on automated dialing. One of its earliest TCPA rulings declared that “speed dialing,” “call forwarding,” and “delayed messages” do not fall within the ATDS definition because the numbers at issue “are not generated in a random or sequential fashion.” In re Rules and Reguls. Implementing the Tel. Consumer Prot. Act of 1991, 7 FCC Rcd. 8752, ¶ 47 (1992). Another early ruling found that the ATDS definition does not apply to equipment that dials “specifically programmed contact numbers” rather than “randomly or sequentially generated numbers.” In re Rules and Reguls. Implementing the Tel. Consumer Prot. Act of 1991, 10 FCC Rcd. 12391, ¶ 19 (1995). Although some subsequent statements from the FCC tried to stretch the definition to fit emerging technologies, those interpretations were roundly rejected by the D.C. Circuit in a Hobbs Act appeal in 2018. See ACA Int’l v. FCC, 885 F.3d 687, 695 (D.C. Cir. 2018). Recent rulings have hewed closer to the statute’s text. See, e.g., In re Rules and Reguls. Implementing the Tel. Consumer Prot. Act of 1991, 35 FCC Rcd. 6526, ¶ 9 (June 25, 2020) (“[O]nly technology that has the capacity to store or produce numbers to be called using a random or sequential number generator, and to dial such numbers, is deemed to be an autodialer. Whether a certain piece of equipment or platform is an autodialer turns on whether it is capable of performing those functions without human intervention, not whether it can make a large number of calls in a short time.”).
Wouldn’t This Statute Make a . . . Criminal of Us All?
The argument in Facebook v. Duguid pitted noted oral advocate Paul Clement (for Facebook) against noted grammarian and lexicographer Bryan Garner (for Duguid). The argument addressed a broad range of grammatical, syntactical, and historical topics. Perhaps the most important was whether Duguid’s reading would cause the statute to apply to smartphones—which, even with their default factory settings, can both store and automatically dial numbers.
Several justices noted the absurdity of such a result. For example, Justice Gorsuch asked whether Duguid would “make a criminal of us all.” Justice Breyer likewise observed that Duguid’s reading “produces a very peculiar or weird result” by making it “unlawful for a person to use a cell phone” to call stored emergency numbers. Justice Sotomayor seemed convinced of that, stating that “almost all phones have the ability to store and dial” and that the “logical consequence” of Duguid’s reading is that “every cell phone owner would be subject to the harsh criminal and civil penalties of the TCPA.” And she was not at all comforted by the notion that such a lawsuit “hasn’t happened yet.” “I know for sure that there will be lawsuits against individuals that will follow,” she said.
Duguid’s response was that smartphones could not dial numbers without some degree of “human intervention,” which would take them outside the scope of the ATDS definition. This was something of a role reversal, for it has historically been defendants that have emphasized a device’s inability to dial numbers without human intervention. That is because the second half of the ATDS definition limits its scope to “equipment which has the capacity . . . to dial . . . numbers.” 47 U.S.C. § 227(a)(1). That language has caused the FCC to say that equipment is not an ATDS unless it is the equipment—as opposed to a human—that “dials” numbers. See In re Rules & Reguls. Implementing the Tel. Consumer Prot. Act of 1991, 18 FCC Rcd. 14014, ¶¶ 132–33 (July 3, 2003); In re Rules & Reguls. Implementing the Tel. Consumer Prot. Act of 1991, 27 FCC Rcd. 15391 ¶ 2 n.5 (Nov. 29, 2012); In re Rules & Reguls. Implementing the Tel. Consumer Prot. Act of 1991, 30 FCC Rcd. 7961, ¶ 17 (July 10, 2015); In re Rules and Reguls. Implementing the Tel. Consumer Prot. Act of 1991, 35 FCC Rcd. 6526, ¶ 9 (June 25, 2020). Courts have generally agreed.
But several justices seemed unconvinced by either the legal argument (i.e., that “human intervention” is even a part of the definition) or factual argument (i.e., that smartphones cannot dial without such intervention). Justice Alito wondered whether a “human intervention” test is workable considering that “everything that computers do requires at some point some degree of human intervention.” Justice Barrett echoed that, noting that her smartphone’s autoreply feature does not require her to be “pressing a button each time” and asking “how far back” in the process the human intervention could go and still take a device outside the scope of the statute—a keen question that has been asked by many businesses that rely on vendors to make calls on their behalf. Justice Gorsuch was more direct, accusing Duguid of “putting a lot of words” in the statute simply to avoid the “parade of horribles” that results from his own reading of the statute.
Notably, Facebook picked up on this and argued in rebuttal that “the adverb ‘automatically,’ the adverbial phrase ‘without human intervention,’ and the adverbial phrase ‘en masse,’ none of those phrases is in the statute.” In Facebook’s view, the thing that gives an ATDS “a sense of automaticness is using a random or sequential number generator.”
It may be, then, that the Court ends up reading the first half of the definition narrowly but the second half of the definition broadly.
An Ill Fit for Current Technology
Several justices noted that this case concerns applying an old statute to new technology—a not uncommon phenomenon in class-action litigation. Justice Thomas observed that “we make great effort to interpret a statute that really wasn’t intended for the universe in which we are operating now” and asked, “[A]t what point do we just simply say that?” Indeed, he began by asking whether text messages—which did not even exist when the TCPA was enacted in 1991—should even be considered “calls” such that they would be covered by the statute. (The answer is that they really should not, but the FCC has said otherwise, and courts have so far deferred to that interpretation.) There appeared to be a general consensus that the TCPA is an “ill fit” for modern times.
Justice Sotomayor also asked whether the absurdity of Duguid’s reading has more to do “with the TCPA being outdated,” and asked whether it is for Congress or the Court “to update the TCPA.” Justice Thomas called the statute’s autodialing restriction “almost anachronistic, if not vestigial,” and asked whether it could be stricken due to its senescence. Justice Alito noted that although academics argue that “courts should have the power to declare statutes obsolescent” and although “this statute might be a good candidate” for that, the Court hasn’t “claimed it so far.”
In Vogue in Certain Grammarian Circles
The justices did, of course, have several questions about the rules of grammar and syntax and the canons of construction. Justices Alito and Gorsuch, for example, asked whether it might be more natural to read using a random or sequential number generator as modifying only produce, as it is not immediately apparent how a number “generator” can “store” a number. And Justice Kagan asked how to interpret a hypothetical statute that made it illegal “to stab or shoot another person using a firearm.” Facebook responded that the punctuation in the hypothetical statute would matter and that, in any event, there is no “logical inconsistency or linguistic impossibility” with reading using a random or sequential number generator as modifying both produce and store.
Facebook also addressed the parties’ superfluity arguments by explaining that “Congress does not put an elephant hole next to a mouse hole,” meaning that Duguid’s interpretation makes the word store so broad that the words produce and using a random or sequential number generator would serve no purpose. Duguid’s counsel responded that Facebook’s reading was “viperine” because it would “read the statute into oblivion,” as modern callers have largely abandoned random or sequential number generators. But Facebook and the government both noted that was not the case in 1991 and that this is “actually evidence of the TCPA’s success” in “eradicating that specific technology.” In other words, the “presumption against ineffectiveness” has no relevance here, as the statute was highly effective at doing what it was supposed to do.
But perhaps the most interesting line of questioning concerned the extent to which traditional rules and canons even matter. Several justices asked about Duguid’s reliance on “synesis”—the idea that, as his counsel put it, courts should “look at the semantic content of the words” rather than treat them as “fungible morphemes and say once you hit the comma, everything before it gets modified.” Justice Alito picked up on that, noting that lay readers “ask about the sense of it before they get to all this arcane stuff” like “treatises on grammar, syntax, usage, or interpretation.” And Chief Justice Roberts asked whether we should “assume that the ordinary speaker is applying these canons or rules of syntax at all.” Indeed, the chief justice asked—perhaps rhetorically—whether “the most useful rule of construction” would be “to take a poll of 100 ordinary speakers of English and ask them what it means.” Counsel for Duguid—whose stock and trade are words and rules of grammar—agreed that this would be “a useful datum.” The chief justice added that “most people’s first blush would suggest that” Facebook’s reading “makes the most sense.”
For his part, counsel for Facebook preemptively disagreed with any reliance on synesis, dismissing it as an idea that is “in vogue in certain grammarian circles” but isn’t “featured in this Court’s cases.” Justice Barrett asked Duguid whether synesis fits “literary language” better than “legal” language and if it is even a “legal concept” that is “appropriately applied in the law.” Having clerked for Justice Scalia—with whom Duguid’s counsel frequently collaborated—she wondered whether synesis would be “contrary to the method of interpretation that you’ve endorsed in other contexts.” Duguid’s counsel conceded that he “does not endorse it broadly” as it “commonly appears” in some academic writing.
The Case Is Submitted
Tea leaves are unreliable. But it does seem that, for reasons both grammatical and practical, a majority of justices are poised to adopt Facebook’s reading of the statute. The justices had incisive questions for both sides, as one would expect with any issue that has so divided the lower courts. But several seemed uncomfortable with reading the ATDS definition in a way that would apply to any smartphone and unconvinced that Duguid had found a limiting principle that would avoid that absurdity.
The outcome will impact thousands of pending cases, many of which have been stayed pending a decision. Some TCPA claims will no doubt survive no matter what the ruling. But the law’s autodialing restriction has been read in a way that Congress could not have imagined, let alone intended. And the ensuing explosion of litigation has created not only an invisible tax on services but also a substantial chilling effect on speech—neither of which is good for anyone.
