The Yale Law Journal


Fourth Amendment Reasonableness After Carpenter

01 Apr 2019

abstract. In Carpenter v. United States, the Supreme Court held not only that the Fourth Amendment applies when the government collects certain categories of third-party data, but also that for such collection no process short of a warrant is constitutional. This Essay argues that a categorical warrant requirement for electronic surveillance is a mistake, and that, when faced with warrantless electronic surveillance, courts should consider whether such surveillance is nevertheless reasonableness, especially where it is legislatively authorized and subject to judicial oversight.


Carpenter v. United States1is one of this generation’s most important Fourth Amendment opinions.2 Commentators have highlighted—and overwhelmingly praised—the opinion’s limitation of the third-party doctrine.3 But just as important as the decision’s effect on the scope of the Fourth Amendment—under what circumstances the amendment applies—is its impact on the amendment’s content. The Supreme Court didn’t just hold that government acquisition of cell-site location information is subject to the Fourth Amendment. It also held that, even if congressionally authorized, any process short of obtaining a warrant—and thus any level of suspicion less than probable cause—would be unconstitutional. This Essay argues that Carpenter’s embrace of a categorical warrant requirement was a mistake,4 and that the Court missed the opportunity (one it will inevitably have to take up in the future) to decide when legislatively authorized warrantless electronic surveillance is reasonable.

I. carpenter and the stored communication act

Unlike many Fourth Amendment cases, Carpenter is not about a generic police practice, but rather a specific law: the Stored Communications Act (SCA).5 The heart of the SCA is 18 U.S.C. § 2703, which sets out the procedures law enforcement must follow when it compels third parties to disclose user data.6 Congress passed the SCA against a background assumption that electronic data held by third parties was not covered by the Fourth Amendment because of the third-party doctrine, which provides that government acquisition of a person’s data (especially noncontent data) from a third party is generally outside the scope of the Fourth Amendment.7 Thus, although the SCA is often framed as a grant of power to law enforcement, its main impetus was the opposite: Congress was chiefly concerned about digital privacy, and thus went to great lengths to specify workable, privacy-protecting rules governing law enforcement’s ability to access certain categories of digital information.8

The SCA provides different levels of protection depending on how sensitive Congress deemed certain categories of information. For example, the SCA requires law enforcement to secure a warrant (and thus establish probable cause) before acquiring the contents of certain categories of emails.9 On the other end of the spectrum, Congress viewed user names, addresses, and billing information as relatively nonsensitive and thus the SCA lets law enforcement acquire such information with a mere subpoena.10

Section 2703’s main innovation is to recognize a category of information between these two poles—information that is not so sensitive as to require a warrant, but sensitive enough that the government must do more than merely issue a subpoena for its production. Specifically, sections 2703(c) and 2703(d) provide that the government may acquire a noncontent “record or other information” about a user if it gets a court order (known as a “D order”) based on “specific and articulable facts showing that there are reasonable grounds to believe that the . . . records or other information sought[] are relevant and material to an ongoing criminal investigation.”11

In Carpenter, the government had used an otherwise-valid D order to obtain Timothy Carpenter’s cell-site location information (CSLI) from his cell-phone provider.12 To ensure consistent service, cell phones continuously connect to nearby cell towers, and cell phone providers record this CSLI for their own business purposes.13 CSLI can provide fairly detailed information on where a particular cell phone—and by inference its owner—has been, and is therefore very useful to law enforcement for criminal investigations. In Carpenter’s case, the government obtained two orders, which together provided over a hundred location points a day for more than a hundred days.14 The government used this information to place Carpenter at the scene of a number of armed robberies, for which he was convicted and sentenced to over one hundred years in prison.

Carpenter challenged his conviction on the grounds that it was unconstitutional for the government to use a D order—rather than a warrant—to acquire his CSLI. The Court of Appeals affirmed Carpenter’s conviction, relying (as had the other circuits that had considered the issue) on the third-party doctrine.15 Since the CSLI was noncontent data (that is, data about Carpenter’s cell phone, rather than the data Carpenter’s cell phone transmitted) and acquired from a third party (Carpenter’s cell-phone provider), the Court of Appeals held that the Fourth Amendment did not apply, and thus the D order was constitutional.16

The Supreme Court reversed. Most of the Court’s opinion—and the portion of the opinion that has received the most public and scholarly commentary—focused on the question of whether acquisition of CSLI from a third party could in fact constitute a search within the meaning of the Fourth Amendment. The Court held that, because the amount of CSLI the government acquired exposed a vast amount of private information, Carpenter had a “reasonable expectation of privacy” in that information, and thus the Fourth Amendment applied.17 Although the Court did not overrule the third-party doctrine, it substantially limited its scope, so much so that one commentator takes the decision as “declar[ing] the third-party doctrine to be almost dead.”18

Although the Court (and the four separate dissents) focused on the issue of Fourth Amendment coverage—was the government’s acquisition of Carpenter’s CSLI a “search”?—that did not resolve the issue of whether the government acted unlawfully; the Court still needed to decide whether the search was “unreasonable.” Without much analysis, the Court applied the traditional warrant requirement, whereby “warrantless searches are typically unreasonable where a search is undertaken by law enforcement officials to discover evidence of criminal wrongdoing.”19 Because the SCA allows for D orders on a showing of less than probable cause, D orders cannot qualify as warrants. And because none of the traditional warrant exceptions—namely exigency—applied, the Court held the government’s acquisition of CSLI to be unlawful. The Court thus found the SCA to be unconstitutional to the extent that it permits the government to collect certain categories of noncontent data (like a large amount of CSLI) with less than probable cause. But should it have?

II. four models of the fourth amendment

As Christopher Slobogin has pointed out, a useful way to think about the Fourth Amendment is to imagine a world in which the amendment didn’t exist, decide on core principles of investigative criminal procedure that we can all agree on, and then (with the least possible unsettling of precedent) identify how to bring Fourth Amendment doctrine in line with those principles.20 We would need to decide: (1) what sorts of government actions the rules cover, (2) what the rules require of those actions, and (3) what to do when the rules are violated. This nicely maps to how Fourth Amendment doctrine is generally divided: scope, requirements, and remedy. For our purposes, we can put the issue of remedy to the side and focus on scope and requirements. In particular, consider four different approaches: (1) minimal, (2) maximal, (3) selective, and (4) regulatory.

Law-and-order types might support a minimal criminal procedure with narrow scope and few requirements—a world in which rules of criminal procedure apply infrequently and are quite permissive even when they do apply. There aren’t many defenders of this position in the judiciary (and almost none in the academy), and it is thus an unlikely model for Fourth Amendment jurisprudence.

Civil libertarians want the opposite: they want a maximal criminal procedure that combines broad scope and a lot of requirements—a world in which the rules of criminal investigations apply to most police activities and impose substantial restrictions on them. The vast majority of academic commentary on the Fourth Amendment takes this view to one degree or another—for example, by criticizing the third-party doctrine21 or celebrating (and urging the reinvigoration of) the warrant requirement.22

The maximal approach has two disadvantages, however. The first is that a maximalist push for civil liberties will, at some point, detract from social welfare, since civil liberties must, at some point, be traded off for public safety.23 The second is that, whatever its normative merits, the maximal position is a political nonstarter—not just in terms of electoral politics, but also in the judicial politics underlying constitutional law. A variety of factors—from the small-c conservativism of the federal bench, to its demographic makeup (many judges are former federal prosecutors),24 to the United States’ exceptionalism in both crime and punishment25—mean that our criminal procedure is highly unlikely to settle into a civil-libertarian equilibrium. As the history of the Fourth Amendment shows, in the long run criminal procedure must meet criminal justice halfway.26

This brings us to the selective approach: a narrow-scope, high-requirements position that captures much of contemporary Fourth Amendment doctrine. Constitutional rules of criminal procedure apply only to certain narrow categories of government actions, but, where the rules apply, they impose substantial limitations on government practices, primarily through the requirement of ex ante judicial authorization supported by individualized, probable cause. The selective approach thus supports the once-dominant and still influential warrant requirement.27

A key flaw with the selective approach is that it withholds Fourth Amendment protections from large categories of police practices that clearly deserve it.28 Under current doctrine, the Fourth Amendment doesn’t apply when the police physically track our movements through the streets.29 But why not? Clearly such surveillance infringes on privacy, raises the specter of government abuse, and can chill liberty. An ideal code of criminal procedure should, therefore, at least regulate (even if it ultimately allows) such conduct.

So why have courts largely adopted the selective model? It’s unlikely that the Justices on the Supreme Court actually think that detailed public surveillance doesn’t implicate Fourth Amendment values.30 Rather, courts are selective in where they apply the Fourth Amendment because they perceive any broad-scope position as incompatible with the realities of modern policing—as long as courts impose the high requirements of individualized, probable cause in all cases in which the Fourth Amendment applies.31 As Anthony Amsterdam recognized in his famous lectures on the Fourth Amendment, an “all-or-nothing approach to the amendment puts extraordinary strains upon the process of drawing its outer boundary lines.”32

This dynamic is apparent in how lower courts have applied Carpenter: generally cabining Carpenter’s cabining of the third-party doctrine.33 Carpenter’s holding only applies to a particular set of records: more than seven days of reasonably precise CSLI.34 Thus, courts not wanting to subject third-party surveillance to the Fourth Amendment’s requirements could choose to read Carpenter narrowly. For example, one district court has held that Carpenter did not apply to grand jury subpoenas sent to an internet service provider (ISP) and an email provider for subscriber information associated with an ISP account and an email address: “The privacy interest in this type of identifying data . . . simply does not rise to the level of the evidence in Carpenter such that it would require law enforcement to obtain a search warrant.”35 Courts have similarly found Carpenter inapplicable to fixed video monitoring,36 location-revealing bank records,37 and online-shopping histories.38 Other courts have emphasized that Carpenter only discussed historical, rather than real-time, collection of CSLI.39 Still others have emphasized Carpenter’s seven-day timeframe as a potential lower bound for when Fourth Amendment protections kick in.40

The stakes for this jurisprudential choice are high, especially when applied to emerging “data driven,” “big data,” and “predictive” policing.41 Because these investigative practices rely on accumulating and analyzing large data sets, they cannot operate if the government is required to establish probable cause that a particular individual is tied to a particular offense before the government can collect or analyze that person’s data. But if the Fourth Amendment doesn’t apply at all to such programs—for example, because, under the third-party doctrine, individuals have no privacy interests in data generated by others about them—then highly intrusive and privacy-threatening government activity will go unchecked. If courts have to choose between hamstringing police and allowing privacy intrusions to go unchecked, they will likely choose the latter. But the public loses out either way.

Fortunately, this choice—between a selective, narrow-scope criminal procedure and a maximal, broad-scope one—is false. Both alternatives assume that, whenever the rules of criminal procedure are implicated (that is, whenever there is a “search” under the Fourth Amendment), the full warrant requirement—ex-ante authorization by an independent magistrate, supported by probable cause—applies.

But this need not be the case. A regulatory model of criminal procedure would have criminal procedure rules apply to almost all government investigatory conduct, but would tailor the requirements to the invasiveness of the government action, the public-safety interests at stake, and the costs and benefits of different levels of ex-ante authorization and predication.42 Criminal procedure would then be no different than any other of the many regulatory domains in which government action must always meet some minimal bar of evidence, logic, and reasoned elaboration.43 Doctrinally, this view emphasizes that “[t]he touchstone of the Fourth Amendment is reasonableness,” and that reasonableness is ultimately a question of balancing: “assessing, on the one hand, the degree to which it intrudes upon an individual’s privacy and, on the other, the degree to which it is needed for the promotion of legitimate governmental interests.”44

Such a response to Carpenter would fully accept not just the letter but also the spirit of the case: that the third-party doctrine, by making a reasonable expectation of privacy (and thus Fourth Amendment coverage) rise and fall based on the mere fact of third-party access, is hopelessly flawed and should be broadly cut back, if not abandoned entirely. Whatever the ultimate ground(s) of the Fourth Amendment—privacy, security, liberty, etc.—the third-party doctrine is both so over- and under-inclusive that its drawbacks outweigh its doctrinal benefits.

Without the third-party doctrine, courts would have to decide whether people have reasonable expectations of privacy in the many categories and vast quantities of digital information that are currently excluded from the Fourth Amendment. Freed from the concern that finding government activity a search would require the onerous imposition of the full warrant requirement, courts would almost certainly widely expand Fourth Amendment coverage. This would finally bring Fourth Amendment doctrine in line with general intuitions of what deserves Fourth Amendment regulation: not just physical intrusions like searches of homes and persons, but also digital intrusions of information—location, communication, etc.—held by third parties.

What would a post-Carpenter, reasonableness-centered, regulatory jurisprudence look like? The Seventh Circuit provided a hint in Naperville Smart Meter Awareness v. City of Naperville.45 Plaintiffs, a civil-society group, sued Naperville, Illinois to prevent it from installing electricity “smart meters.” Smart meters, unlike ordinary electricity meters, collect vast quantities of granular information on domestic electricity consumption, which can “reveal[] . . . the happenings inside a home.”46 The plaintiffs argued that the city’s collection of such data would violate the Fourth Amendment’s prohibition on unreasonable searches. Naperville responded that the third-party doctrine applied, since residents entered a “‘voluntary relationship’ to purchase electricity” and thus “sacrifice[d] their expectation of privacy in smart-meter data.”47 In rejecting the city’s third-party-doctrine argument, the Seventh Circuit relied on Carpenter:

[I]n this context, a choice to share data imposed by fiat is no choice at all. If a person does not—in any meaningful sense—“voluntarily ‘assume the risk’ of turning over a comprehensive dossier of physical movements” by choosing to use a cell phone, it also goes that a home occupant does not assume the risk of near constant monitoring by choosing to have electricity in her home.48

But what sort of Fourth Amendment protection would be called for? Crucially, the court did not follow Carpenter in assuming that only a warrant would render the government search “reasonable” under the Fourth Amendment. Instead, the court analyzed the data-collection program under a reasonableness balancing test. It concluded, based on the minimal intrusiveness of the smart-meter program, the noninvestigatory purpose of the program, and the substantial government interest in electrical-grid modernization, that the government action, while a warrantless search, was constitutional.49

Naperville is a start, but more doctrinal work needs to be done to combine Carpenter’s cutting back on the third-party doctrine with a reasonableness test that looks beyond the warrant requirement. Courts will need to extend Naperville’s reasonableness approach to surveillance conducted for the “ordinary enterprise of investigating crimes.”50 It remains to be seen whether lower courts will be able to read Carpenter creatively to do so51 or whether the Supreme Court will itself have to clarify the stringency of the warrant requirement.

III. reasons for reasonableness

The intuition behind reasonableness balancing is, well, that it’s reasonable. After all, how else could we approach complicated policy areas like law enforcement and government surveillance than by pragmatically trying to balance costs and benefits? Nevertheless, my proposal is, like all others of its type, subject to the “traditional critique of reasonableness”: (1) that it is insufficiently protective of privacy and security; and (2) that, as compared to the “rule” requiring warrants, the reasonableness-balancing “standard” is too difficult to administer.52

A. Reasonableness Can Protect Fourth Amendment Values

The longstanding concern with departing from a strict warrant requirement is that the results undermine the Fourth Amendment’s protections of security and privacy. For example, in Terry v. Ohio the Supreme Court held that, although a stop and frisk is a search and seizure within the scope of the Fourth Amendment, the police can conduct them with less than probable cause.53 Terry has been widely criticized as both a betrayal of the Fourth Amendment’s core values54 and an invitation to programmatic police harassment that falls disproportionality on racial minorities without providing substantial public-safety benefits.55 And the Supreme Court’s administrative and special-needs cases—where the Court explicitly engages in freestanding reasonableness balancing—have been similarly criticized, with commentators characterizing them as using highly deferential rational basis review in which the government interest—however attenuated or invented after the fact—inevitably trumps individual privacy and security.56 Even defenders of reasonableness balancing concede that the Supreme Court has failed to specify precisely what constitutes “reasonableness”57 for a warrantless search.

That the task is challenging, however, does not mean it is impossible. A growing movement, the “new administrativist turn,”58 looks to administrative law and theory as a model for regulating government investigations and surveillance. For example, Daphna Renan proposes using cost-benefit analysis to measure “programmatic efficacy.”59 Barry Friedman and Maria Ponomarenko suggest that local police departments promulgate rules through notice and comment before acting.60 Slobogin argues for a general “proportionality principle” for law-enforcement activities,61 and, in the case of broad surveillance programs, appropriate levels of democratic participation.62

Matters should be substantially easier where the Fourth Amendment activity is pursuant to legislative authorization, as was the case with the SCA and the CSLI acquisition in Carpenter. In such cases, courts should at least seriously consider whether the legislature’s judgment can shed useful light on the difficult tradeoffs inherent in any surveillance scheme. Legislatures have several structural advantages when it comes to surveillance policymaking: Congress can legislate more comprehensively, it has access to better information and expertise, and, most importantly, its representative nature means that it enjoys a legitimacy that the courts do not, at least when it comes to making tradeoffs between otherwise reasonable policy choices. As Judge Wilkinson has observed with respect to the SCA’s treatment of CSLI: “Faced with a term literally crying out for balance between the competing interests of individual privacy and societal security, it is appropriate to accord some degree of deference to legislation weighing the utility of a particular investigative method against the degree of intrusion on individuals’ privacy interests.”63

None of this is to suggest that courts should uncritically let legislatures preempt the field of criminal procedure.64 Nor does it suggest that a court, after performing reasonableness balancing, could never conclude that a warrant is in fact necessary.65 But we should recognize that the judiciary has no monopoly on reasonableness. We should also recognize the need to incentivize legislatures (not to mention law-enforcement and surveillance agencies themselves) to develop detailed and binding regulations in contexts where the Fourth Amendment does apply, and not, as is currently the case, only where the Fourth Amendment doesn’t apply.66 The less deference the political branches can expect regarding what constitutes reasonableness under the Fourth Amendment, the less they’ll try to bring their representative processes to bear on the issue.

Given all this, it’s notable how cursory—dismissive, even—was the Supreme Court’s treatment of the SCA. In a single paragraph, the Court disparaged the applicable legal standard—whether the CSLI was “relevant and material to an ongoing investigation”—as “fall[ing] well short of the probable cause required for a warrant.”67 The Court read the standard expansively, such that “law enforcement need only show that the cell-site evidence might be pertinent to an ongoing investigation—a gigantic departure from the probable cause rule.”68

The Court’s insistence on portraying the legislative standard as meaningless led it to a mischaracterization. The court “decline[d] to grant the state unrestricted access to a wireless carrier’s database of physical location information.”69 But D orders permit no such thing: as noted above, the D-order standard is drawn from the intermediate requirements of Terry, and provides far greater protection than a mere grand-jury subpoena, which can issue with essentially no suspicion at all.70

A useful case study of how courts can analyze the constitutionality of complex, legislatively authorized surveillance programs can be found in section 702 of the Foreign Intelligence Surveillance Act (FISA).71 Section 702, which has been reauthorized by Congress multiple times over the past decade, allows the government to conduct warrantless domestic surveillance of electronic communications on “persons reasonably believed to be located outside the United States to acquire foreign intelligence information.”72 Critically, section 702 does not require the government to obtain a warrant before engaging in such collection; rather, the program sets up a complex system of ex ante judicial oversight (of targeting and minimization procedures) and ex post internal oversight (through multiple layers of compliance review, both within the intelligence agency itself and from the Department of Justice) to minimize incidental impacts on the privacy of U.S. persons.73 There is no question that interception of information under section 702 is a “search” under the Fourth Amendment. But those courts that have evaluated the constitutionality of the program have upheld it on the grounds that its institutional features render it reasonable.74 The future of law-enforcement electronic surveillance lies in FISA-like regulatory regimes that innovatively combine judicial oversight with rigorous internal compliance mechanisms, not slavish adherence to a judicially constructed, one-size-fits-all warrant requirement.

FISA is hardly uncontroversial, and civil libertarians may view my argument—that domestic, law-enforcement electronic investigations use foreign-intelligence surveillance as a model—as taking surveillance law in precisely the wrong direction.75 But the realistic alternative is not a maximal Fourth Amendment, but rather the limping along of the selective one. In other words, either Carpenter’s rigid emphasis on the warrant requirement will have to give way to some less demanding requirement, or the third-party doctrine will live on to the extent law enforcement needs it to, as courts cabin the effect of Carpenter.

B. Reasonableness Is Administrable

As Jamal Greene notes, “[t]he most significant theoretical concern with importing proportionality jurisprudence into U.S. courts stems from the nature of the Supreme Court as an apex court within a system of highly decentralized constitutional jurisdiction.”76 In particular, “[i]nsofar as proportionality”—or, in the Fourth Amendment context, reasonableness review—”relies heavily on case-by-case adjudication, it gives less guidance to other courts, public officials, and citizens than does a categorical approach”—that is, a strict warrant requirement.77 This in turn can raise rule-of-law fears, as the “freewheeling ‘reasonableness’ standard . . . suffers from the concerns about official arbitrariness that rules are meant to combat.”78

This is an important critique of any reasonableness-focused interpretation of the Fourth Amendment, but (as Greene notes) it is not fatal. First, the increased decision costs associated with reasonableness (as with standards generally) may nevertheless be smaller than the accuracy costs associated with categorical rules.79 Given the manifest lack of fit between a blanket warrant requirement and modern data-driven police and surveillance practices, any administrability costs of a reasonableness regime seem to be dwarfed by the accuracy costs of continuing to abide by a strict warrant requirement—both in terms of the practices that will be unnecessarily precluded, and those practices that should be covered by the Fourth Amendment but will not be.

Second, the inadministrability of a reasonableness standard is exaggerated. Standards, including in the Fourth Amendment context, do not remain open-ended and amorphous for long; as they are fleshed out, they naturally become more codified and rule-like.80 And even where some residual uncertainty remains, it is hardly fatal. For example, the probable-cause standard is famously vague,81 yet its backers include some of the most diehard supporters of the warrant requirement.

No doubt a full description of what constitutes “reasonableness” for digital searches under the Fourth Amendment will be a complicated, sometimes messy, affair. But a warrant-requirement jurisprudence will be just as complex, as evidenced by the complicated series of tests courts currently use to decide whether a particular government action counts as a “search.”82 Digital searches are not exempt from this problem. For example, leading Fourth Amendment scholar Orin Kerr devotes dozens of pages to operationalizing Carpenter’s holding on scope to the many varieties of real-world electronic investigations.83 This is an invaluable service to lower courts, and I generally agree with Kerr’s conclusions, insofar as they assume the continuation of the selective (narrow-scope/high-requirement) paradigm for electronic surveillance.

But Kerr’s method does not strike me as substantially less complicated than Fourth Amendment doctrine would be under a regulatory (high-scope/variable-requirement) paradigm. As Slobogin notes, “even seemingly ‘bright-line’ rules usually become blurred as the police and the adversarial process test their outer limits. The grail of ‘rule-oriented’ jurisprudence is as mythical as King Arthur’s.”84 And crucially, doctrinal work to flesh out the reasonableness requirement would more directly engage with what I take to be the heart of the Fourth Amendment: an optimization of the inevitable tradeoffs between privacy and security. That is, the question of “reasonableness.”


If the Supreme Court is serious (as I hope it is) about rolling back the third-party doctrine—if, in other words, it means Carpenter to be the beginning of a Fourth Amendment revolution, not just a one-off case—it will sooner or later have to repudiate Carpenter’s suggestion that warrants (and thus individualized, probable cause) are required for all electronic surveillance. In the meantime, a critical research agenda in Fourth Amendment scholarship must be to develop an account of what substitutes for warrants are reasonable in a digital age.

Visiting Assistant Professor of Law, University of Minnesota Law School. For helpful comments I thank Richard Frase, Jill Hasday, Aziz Huq, Orin Kerr, Anna Lvovsky, Justin Murray, Shava Nerad, Shalev Roisman, Christopher Slobogin, and participants in the Public Law Workshop at the University of Minnesota Law School. For terrific research assistance I thank Sam Cleveland and Paul Dimick. And for excellent editorial work I thank the editors of the Yale Law Journal.