Closed Universe Searches

A genetic genealogy search for a match to a crime scene profile identifies eight relatives, one of whom is almost certain to be the perpetrator. A geofence warrant returns an anonymized list of four cell phones present at a series of bank robberies. A facial-recognition software analysis of a surveillance video generates fifteen persons of interest. By now, these scenarios are commonplace. Equally as commonplace are the two strands of case law and scholarship that have unfurled around them. The first strand addresses the breadth of such searches and wrestles with whether, when, and how these indiscriminate, open-universe searches should be permitted. The second strand, propelled even more strongly forward by the Supreme Court’s decision in Carpenter v. United States, addresses the depth of technological searches and focuses on whether, when, and how the Fourth Amendment should restrain the uniquely intrusive capacity of new technologies to divulge information about a targeted person.

What is missing, however, is an account of the middle. All too overlooked is a third tapestry, woven from both of these threads. Specifically, technological searches have dramatically increased both the frequency with which law enforcement confronts a closed universe of suspects rather than a single suspicious target, as well as the probability that the true perpetrator of a crime can be found by engaging in invasive technological searches within that closed universe.

This Article is the first to identify and define closed universe searches as (1) searches of a small pool of persons connected by happenstance to a crime; (2) one of whom is almost certainly the perpetrator, but the rest of whom are equally certainly innocent; (3) using new technological tools that can identify the perpetrator from within the pool with certainty or near certainty. Applying this new concept, it then considers how Fourth Amendment doctrine has or might resolve closed-universe-search questions—questions like how intrusively police can investigate the suspects in that closed universe; how transparent investigative actions must be; and how accountable police are to rules designed to prevent abuse, misuse, or excess. By exploring the special dynamics of closed universe searches that differentiate them from either indiscriminate, universal searches or targeted, individualized searches, this Article reveals the inadequacy of applying existing analytical frameworks to this particularly vulnerable group. It thus closes by proposing that legislatures enact statutory frameworks to govern closed universe searches.

Table of Contents Show

Introduction

In Agatha Christie’s Murder on the Orient Express,[1] an unexpected snowfall traps a train on its tracks one night, and when morning comes, a passenger is dead. Renowned detective Hercule Poirot combs through the clues to deduce which one of the twelve passengers could be the perpetrator. Of course, the twist is that they all did it—what seemed to be an arbitrary and disconnected assortment of riders is in fact a conspiratorial group seeking to avenge the murder of a child years earlier. What presented as a pool of suspects was in fact an oceanic conspiracy.

Today, the distinguished detective might never have unspooled the thread that stitched the passengers together. Instead, Poirot would have just swabbed the decedent for DNA, obtained samples from the riders on the train, and identified the perpetrator as whomever made the incriminating match. Or perhaps he would have pulled the footage from a security camera and used sophisticated algorithmic pattern-matching to link the killer by facial feature or gait. Or maybe he would have sought location data about the cell phones on board, trying to pinpoint who was moving about the cabin in the wee hours of the night. Arrest made, case closed, secrets forever unspoken.

Much has been written about the capacity of new technologies to help prevent and solve crimes.[2] This literature spans a wide array of topics, tackling privacy implications, constitutional rights, civil liberties, evidentiary accuracy, and so on. Both scholars and courts have grappled with questions raised by the breadth of searches using new technologies that enable indiscriminate, mass searches or trawls of data, as well as the depth of the intrusion when such tools are used, allowing for the collection of comprehensive information that was prohibitively difficult to obtain in an analog world.

But notably missing from these breadth and depth accounts is a third story. Namely, that technological searches have dramatically increased the frequency with which law enforcement investigations generate not just a single target but a closed universe of suspects in which the perpetrator can then almost certainly be identified by deploying a technological search method. To be sure, closed universe suspect pools predate the technological innovations of the twenty-first century; in Agatha Christie’s time, the physical world of the train closed the universe of suspects to twelve possible passengers. But technology has accelerated closed universe searches in two important ways. First, the happenstance required to create a true closed universe of suspects was simply less common in the analog world, and thus, fewer such instances arose. Second, even if there was a closed universe of suspects by chance, police lacked a ready and easy technological means to identify which person within that group was the true perpetrator, and so ordinary investigation was the only way to isolate the perpetrator within the pool.

In short, technological advances have altered both parts of the equation: Searches conducted using new technologies routinely produce closed universe suspect pools, and new technologies offer a way to identify the perpetrator within those pools with efficiency and certainty. And the rapid pace of technological change suggests that these searches will continue to proliferate. To give a concrete example: Imagine a person is robbed in a back alley by a masked assailant. Having no leads, investigators turn to Google to conduct a geofence search to locate any phones in the area at the time of the robbery; in other words, to conduct a broad, indiscriminate search.[3] If the search leads to just one cell phone, then that likely gives rise to probable cause, enabling investigators to obtain a warrant authorizing deeper intrusions, such as a search of their phone or a compulsory DNA sample.

But what if the result of the indiscriminate search is six cell phones, not one? In a world without new investigative technologies, police would simply resort to analog methods such as conducting interviews or interrogations or engaging in physical surveillance. The limits of the physical world would limit the degree of the investigation’s intrusiveness (which in turn might also reflect the priority of the investigation in light of the seriousness of the crime). And existing constitutional law (whether one agrees with the doctrinal standards or not) would likely prohibit the most intrusive investigative actions (such as a search of a home) until officers could adduce probable cause by narrowing the list to one suspect.

But in a world with new technological tools, the calculus changes. What if all it would take to isolate the likely perpetrator from the pool of six suspects is a mouth swab from each person? Or a somewhat deeper dive into each of those six people’s location history? Or a quick scan of their Google Maps search histories? The allure of these new technologies is manifold. They are digitally intrusive rather than physically intrusive and thus feel less volatile and confrontational. They promise rapid and highly certain identification of a likely perpetrator. And they scale easily in small numbers. For instance, it is easier to collect six people’s location history than it is to try to piece together that history from physical evidence or surveil someone prospectively on the ground. In many cases, these searches can even be done without alerting any of the searched persons that their DNA has been taken, their data processed, or their search or location histories retrieved.

Despite the greater number of closed universe suspect pools created by new technological tools, and the allure of permitting searches of persons within that pool using these new technological tools, scant attention has been paid to their risks or parameters. Significant scholarly and judicial attention has focused on the breadth question: Should law enforcement be allowed to search all users’ location data indiscriminately to reveal whether any users were at the robbery site? Significant scholarly attention has also focused on the depth question: How intrusive is it to search a phone or compel a DNA sample from a suspect and what should existing doctrine say about if and when police can do so? But little attention has zeroed in on the intermediate question: How should we think about these new technological methods when used to search within a closed universe of only six suspects, one of whom is almost certainly the perpetrator, and when such searches are almost certain to reveal that person?

Doctrinally, the poor fit between closed universe searches and universal, indiscriminate searches on the one hand and deeply intrusive, individualized searches on the other, renders the suspects within that closed universe lingering in a vulnerable status. Because police can show that the narrow, closed universe group almost certainly includes the perpetrator, there is a modicum of suspicion as to each individual member of the group, and thus a closed universe search is not a pure indiscriminate search. At the same time, it is equally certain that most of the members of the group are innocents mixed in with the target by mere happenstance. Similarly, closed universe searches are not classic suspicion-based searches, either, because the suspicion is diffused over a collective, rather than the typical probable cause or reasonable suspicion that a particular person has committed a criminal offense.

If investigations of closed universe suspect pools simply employed conventional methods, then they would not be so troubling. The limits of constitutional doctrine and the physical world would likely suffice to safeguard innocent persons caught in the pool just as they do any other person for whom police lack individualized suspicion. But the convenience and certainty of technological search methods that can be easily deployed to isolate a perpetrator from a closed pool of suspects renders persons in that pool uniquely vulnerable. The pool is both small enough and suspicious enough to make authorizing an intrusive search method plausible and irresistible, even though there is not enough suspicion to apply conventional doctrinal safeguards like a warrant or probable cause. That is why a court that would not dream of authorizing a dragnet for DNA samples from all the men in a neighborhood might, for instance, be more willing to bend doctrinal analysis to authorize samples from three men if such sampling could conclusively establish which one committed the crime. Or why both logistics and the courts might foreclose a deep probing of the search histories of every person in a neighborhood, but why officers might seek, and courts might approve, the same intrusion when limited to four persons in a closed universe pool.

Neither Fourth Amendment doctrine nor the scholarly literature offers a ready framework for how to think through these closed universe searches. It is neither quite right to analogize closed universe searches to indiscriminate, open universe searches (think TSA security or a drunk-driving roadblock) nor to targeted, individualized searches (applying conventional definitions of probable cause). And so, the few courts that have confronted the issue directly have been inconsistent as to both whether and when closed universe searches comport with the Constitution, as well as what standard should apply in reaching that conclusion.[4]

This Article is the first to identify closed universe searches as their own distinct phenomenon, as differentiated from open universe or indiscriminate searches on the one hand and targeted, individualized searches on the other. It explains how closed universe pools are not distinctly the result of new technologies, but that closed universe searches are, because new technologies have increased both the frequency with which law enforcement confronts closed universe pools and the certainty and precision with which investigative tools can isolate the perpetrator within the closed universe. Most importantly, this Article explains how existing scholarly and doctrinal frameworks fail to account for the unique characteristics of closed universe searches, advocates for more concerted attention to the particular dynamics of such searches, and sets forth a framework for addressing them.

Part I examines existing taxonomies for new technology searches and distinguishes closed universe searches as a unique category defined as (1) searches of a small pool of persons connected by happenstance to a crime; (2) one of whom is almost certainly the perpetrator, but the rest of whom are equally certainly innocent; (3) using new technological tools that can identify the perpetrator from within the pool with certainty or near certainty. Part II canvasses the thin case law surrounding closed universe searches and independently analyzes their constitutionality under five existing doctrinal approaches. Part III explains why it is critically important to conceptualize closed universe searches as distinct from their indiscriminate or individualized counterparts. In light of those concerns, Part IV sketches a framework to govern closed universe searches. A brief conclusion summarizes the Article.

I. Searches Using New Technologies

A. Existing Taxonomies

Numerous courts and scholars, myself included, have grappled with the Fourth Amendment implications of our new era of technological tools. Whereas the traditional investigative arsenal relied on informants, confessions, and rudimentary forensic evidence, contemporary policing has at its disposal a suite of far more sophisticated methods. The newspapers are filled with stories of crimes solved through DNA typing and forensic investigative genetic genealogy, facial recognition software, license plate readers, cell-site simulators, geolocation and historical cell-site location information, and other big data searches.

Most of the judicial opinions on the use of new technologies come at the issue from two opposing points of entry. One line of case law stems from Fourth Amendment suppression motions raised post-arrest by individuals identified as suspects either by big data searches or using new and intrusive technological methods.[5] The pinnacle of such cases is a triad of high-tech Fourth Amendment opinions handed down by the Supreme Court: Carpenter, holding that seeking extended cell-site location data is a search under the Fourth Amendment;[6] Riley, holding that a warrant is required to search the cell phone of an arrestee;[7]and Jones, holding that the affixation of a GPS tracker on a car is a Fourth Amendment search.[8]

The second line of case law examines the impact of new technologies from the ex ante point of view, addressing indiscriminate searches of mass quantities of data in order to solve[9] or prevent crime.[10] In these cases, courts focus on the constitutionality of permitting broad, open universe searches that sweep the general public or large swaths of individuals for information about criminal activity. Thus, these cases likewise focus on the immense privacy implications of new technological tools, but in the context of mass searches of a collective of persons, as opposed to an intrusive search of a single person.

Legal scholarship on new technologies similarly tends to cluster around these two poles, addressing the concerns and benefits surrounding broad, indiscriminate data mining (e.g., geofencing, license plate readers, compulsory DNA collection laws, facial recognition software, and cell-site simulators) or deep, targeted intrusions (e.g., historical cell-site and location information, phenotypic or trait-based testing of DNA, and extended data searches of specific accounts). Both strands of constitutional analysis express similar concerns: the intrusiveness of new technologies and attendant assault on privacy, equity, and autonomy;[11] the shift of power between the individual and the state;[12] and the uncertain application or thin scope of existing Fourth Amendment doctrine when applied to technological searches.[13] Both strands likewise attract advocates who view technological advances as an opportunity to enhance (or at least keep pace with) public safety, improve accuracy, and diminish hostile confrontations between individuals and the police.[14]

Whether in court or in the scholarly literature, the language surrounding both broad, indiscriminate searches and deep, targeted ones is often similar. Scholars tend to use phrases designed to capture the idea of new, pervasive, totalizing surveillance. Some common language can be found among the leaders in the field. Andrew Ferguson talks about “big data policing” in his book of the same name.[15] Christopher Slobogin coined the term “suspectless search” to refer to “techniques that use technology to identify the suspected perpetrator of a crime rather than find out more about an already identified suspect.”[16] Mary Fan describes “big data searches”[17] and Barry Friedman and Danielle Keats Citron refer to “indiscriminate data surveillance.”[18]

Only a handful of scholars have focused on developing a precise typology of new technological searches, with an eye toward attaching correspondingly more precise terminology. Perhaps the most careful taxonomy was devised by Slobogin in his book Virtual Searches. Slobogin articulates five different kinds of technological (or, in his parlance, virtual) searches: “suspect-driven,” “profile-driven,” “event-driven,” “program-driven,” and “volunteer-driven” searches.[19] Of Slobogin’s five categories, “event-driven searches” comes the closest to describing the kind of searches of interest in this Article; the rest of the categories focus on only one end of the existing poles, either broad and indiscriminate or deep and targeted. Slobogin describes “event-driven searches” as “retrospective rather than prospective,” in that they “use[] a crime as the starting place and work[] from there.”[20] Although “event-driven searches” are aimed at identifying the perpetrator, “law enforcement must often obtain data about a large number of innocent people.”[21] Implicit in that idea, of course, is not just the broad, indiscriminate nature of the initial search but also some notion that the results of such a search might not pinpoint a single person. Still, while many closed universe searches start as “event-driven searches,”[22] Slobogin’s concept, as the next Section more fully explains, captures something more akin to an indiscriminate search than a closed universe search. The distinguishing characteristic of “event-driven searches” is that the event drives the use of technological tools rather than something about the search itself.

A closer, but still not quite identical, articulation of a search in a manner akin to the closed universe searches described in this Article is offered by Jane Bambauer, who coined the phrase “filtered dragnets.” As with Slobogin’s event-driven search, “filtered dragnets” are “digital searches that identify a suspect based on the details of a crime.”[23] But the defining features of a filtered dragnet are its accuracy, precision, and ability to automate the search and screening function.[24] In her words, a filtered dragnet “provide[s] a suspect’s data to police only if (a) their data matches uniquely criminal details such that there is a high probability they have engaged in criminal conduct; and (b) their data has been pared down to provide only relevant details about the suspected crime to the police.”[25] Bambauer’s definition thus deliberately excludes searches that yield more than one suspect—she expressly states, for instance, that “geofencing and familial DNA-matching procedures often allow police today to access data about a handful of individuals, all but one of whom are necessarily innocent,” which therefore excludes them from consideration as a “filtered dragnet.”[26] In contrast, “closed universe searches,” as explained in the next Section, are meant to address precisely those circumstances.

B.   Defining Closed Universe Searches

So, if a closed universe search is neither a broad, indiscriminate search nor a deep, individualized search, what is it? By my definition, the distinguishing characteristics of a closed universe search are that it is a search (1) of a small pool of persons connected by happenstance to a crime; (2) one of whom is almost certainly the perpetrator, but the rest of whom are innocent; (3) using new technological tools that can identify that perpetrator with certainty or near certainty.

Before explaining more, let me be clear about what a closed universe search is not. A closed universe search is not mass surveillance (for instance, the use of technology to record all license plates that pass through a particular tunnel), irrespective of any subsequent search. Nor is it an open universe, indiscriminate search of surveillance data (for instance, the use of software to search through the collected license plate data), or what courts in the geofencing cases refer to as “step one” of the investigation.[27] Both of these kinds of policing activity use technology in ways that press against the limits of conventional constitutional doctrine. And, to be sure, both kinds of policing activity have helped to enable and accelerate the proliferation of closed universe searches. But neither of these methods are what I mean to call a “closed universe search.”

Rather, a “closed universe search,” as I use it, has two key components: It focuses on a discrete group of persons, not picked at random but connected by happenstance to a crime that most of them did not in fact commit, and it uses technologies capable of isolating the person within that group who likely did commit the crime with both precision and certainty (or near certainty). To be clear: Investigations of persons within a closed universe using conventional methods are not “closed universe searches” as I mean to use the term, nor are searches using new technological methods when deployed only against a single suspect for whom there is clear probable cause or reasonable suspicion.[28]

To give some examples of closed universe searches:

  A disabled woman is sexually assaulted in a group home, and the rape kit reveals the DNA profile of the perpetrator. Only six persons had access to the woman’s room during the time of the assault. Police seek to compel DNA samples from those six persons.

  Police obtain a geofence warrant to identify any cell phones within the vicinity of two separate carjackings. Four distinct phones are identified. Police want to reveal the identity of each phone’s user and to obtain location data as to each phone to isolate the suspect.

  A pole camera records a violent carjacking, and facial recognition software identifies fifteen possible matches to the perpetrator. Police want to use a cell-site simulator to obtain those persons’ mobile numbers and then use location history to determine whether any of them were at the site of the carjacking at the time it occurred.

  A house fire kills several victims. Police seek a “reverse-keyword” warrant to find any devices that searched for the address of the house around the time of the fire.[29] Two such devices are identified, and police seek to identify the users and use facial recognition software and surveillance video to determine if they were in the vicinity at the time of the fire.

  Police use a public recreational DNA database to identify the family lineage of the genetic profile of the perpetrator in a cold rape case. Through additional investigation, police determine that the perpetrator is likely one of four family members. Police seek to compel DNA samples from those four persons.[30]

  A license plate reader identifies ten possible vehicles as matching a vehicle seen speeding away from a kidnapping. Police seek to learn the identity and address of each driver and then use a pole camera to record each of their comings and goings in order to determine if they might be the kidnapper.

In each of these cases, a closed universe pool is created; in most instances, it is an indiscriminate search using new technology that creates it (such as the geofence warrant or the facial recognition search), but in the first example, it was a nontechnological circumstance (physical presence in a group home). Although this precursor act of creating the pool may raise any number of Fourth Amendment questions, those questions are not the concern of the closed universe search—except inasmuch as the increasing use of indiscriminate searches means that many more closed universe pools are being created. The key point is that the manner in which the pool was created is, for purposes of this Article, immaterial.

Notice also that in each example, the pool is just that—a pool with two or more suspects. When a mass or indiscriminate search yields a single suspect—what Jane Bambauer calls a “filtered dragnet”—that too is distinguishable from a closed universe search.[31] Part III addresses this difference at greater length, but suffice it to say now that the concerns that arise with regard to searches of a single, suspicious person are simply different in kind from searches that aim to differentiate amongst a small group of people, each of whom individually are at best only fractionally suspicious (and all but one of whom are indisputably innocent). More importantly, existing constitutional doctrine is designed to address precisely this situation, in which reasonable suspicion or probable cause attaches on an individualized basis.

What transforms a closed universe pool into a closed universe search is what happens next: the use of technological tools on the persons within that pool to identify with certainty (or near certainty) the person of interest. Officers may choose to investigate a closed universe pool using analog or conventional investigative methods—they might call the suspects in for questioning seriatim, or seek to follow them using a surveillance team, or even attempt to obtain evidence using conventional methods like a wiretap or grand jury subpoena. When police investigate closed universe suspect pools using ordinary methods of policing, ordinary doctrines of criminal procedure likewise apply without much drama. There may be good reasons to be concerned about the use of ordinary investigative methods (in-person surveillance, questioning, etc.) in a closed universe investigation, but they are not the primary concern of this Article, and they are no different whether there is one suspect or ten.

Rather, to constitute a closed universe search, the search of the pool must entail the use of an advanced technological method. That is for two reasons. First, because it is the very intrusiveness of technological searches that, when compounded with the closed universe typology of the targets, raises the profound and distinct concerns this Article aims to address. Second, because it is the very ease and attractiveness of using sophisticated and intrusive technological tools to winnow the pool to the suspect that pose a distinct threat to the rights of the innocents caught within the pool.

Part III elaborates at length on why it is important to carve out conceptual space distinguishing closed universe searches from both universal and targeted searches, but briefly, what value is there in defining this subcategory of closed universe searches? The closed universe search is critical to conceptualize now because of the dramatic effect that mass surveillance and technology have had on criminal investigations in two specific ways. New technologies have dramatically accelerated the frequency with which closed universe suspect pools are created. And new technologies have made it possible to conveniently and with certainty isolate the perpetrator from within that closed universe pool.

First, in terms of frequency: The ability of law enforcement to harness mass datasets has transformed what used to be an occasional occurrence into a common and recurring start of an investigation. Again, to be clear, nothing about my concept of “closed universe searches” requires that the suspect pool be created by technology. Closed universe searches existed in the analog world—didn’t we just use to call them “the usual suspects”? It is not uncommon that a case generates a pool of suspects, and then further investigation winnows the pool down to the actual target. The point of identifying closed universe searches is thus not to suggest that the simple concept of a closed universe pool is unique to this moment in time.

But in an analog world, the proverbial “murder on the Orient Express” problem that started this Article was relatively anomalous. A closed universe suspect pool arose only as often and to the extent that the facts of the world allowed it to, such as who had access to a storage cabinet or victim’s apartment. All too often, those facts did not exist.

Technological tools changed this reality because now, with pervasive surveillance and “big data policing” like investigative genetics or continuous location tracking, police often start an investigation by probing datasets for information. Again, to be clear, those probes are what this Article calls indiscriminate (or mass, or universal, or open universe) searches, and they are not the same as closed universe searches. Moreover, if an indiscriminate search turns up a single, precise match—say, a recreational genetic database search that pinpoints one single family member or a geofence that uncovers a solitary cell phone user at the scene of the crime—then the next step is what I would call a targeted investigation, not a closed universe search. But all too often, an open universe search (or facts in the world) does not return a single match but instead identifies some subset of possible suspects—a pool of candidate matches. In such cases, the pool constitutes a closed universe, and the next steps in the investigation entail closed universe searches.

Second, in terms of the nature of the ensuing investigation of the closed universe pool, the use of technological tools matters very much. To constitute closed universe searches for the purpose of this Article, the searches of the closed pool must be conducted using the kinds of sophisticated technological means described in Part I.A as “virtual searches” or “suspectless searches” and the like. That is because the distinct character of such searches renders them particularly threatening to the privacy interests of innocent persons. Technological tools are minimally intrusive in the conventional Fourth Amendment physical-confrontation or liberty-deprivation sense yet maximally intrusive in the informational-privacy sense. They are also typically low-cost for law enforcement and thus readily scalable to the finite number of persons within a closed universe pool. And they tend to be highly accurate or at least offer the promise of accuracy. These features combine to make closed universe searches both difficult to regulate under conventional Fourth Amendment doctrine as well as attractive to allow even when such doctrine suggests otherwise, as explored in depth in Part II.

To draw out the point with an example: Imagine an indiscriminate search conducted on an entire, diffuse population (e.g., drivers, participants in Ancestry.com, cell phone users at a festival). If a search yields only one probable suspect—as in Bambauer’s filtered dragnet example—the very solitariness of that person both curtails the scope of any additional investigation and justifies that investigation with some notion of reasonable suspicion or probable cause.

But what if an indiscriminate search identifies two people? Or ten? Or twenty-five? The mere existence of equally plausible alternate suspects diminishes the strength of a claim to reasonable suspicion or probable cause as to any one suspect. And yet the narrowness of the pool and the ready availability of an efficient, technological means of conclusive identification (e.g., a simple DNA test, a search of those persons’ phone histories) makes such intrusions seem attractive and justifiable to both law enforcement and courts alike.

The closed universe scenario also raises distinct questions about rights and obligations when innocents are caught by happenstance in a police investigation. In such situations, what burdens on privacy or liberty, if any, should each person in the pool bear in service of the social goal of crime solving? What risks or incentives materialize when law enforcement’s attention is trained on a closed universe of suspects, as opposed to either the indiscriminate masses or a single, targeted individual? What value should be accorded to the fact that technological methods typically do not involve the same kinds of physical confrontation or degree of uncertainty as conventional evidence-gathering tools like interrogations,[32] witness interviews,[33] informants,[34] or more primitive forensics?[35] What obligation does law enforcement have to dispose of or safeguard the data of innocent persons swept up into closed universe searches once a perpetrator—not them—has been identified?

The point of defining the distinct category of “closed universe searches” is to begin to build a vocabulary for addressing the particular concerns that attend these searches. Defining a distinct category clearly differentiates closed universe searches from both the indiscriminate searches from which they tend to spring, as well as the targeted searches or filtered dragnets that easily hew to existing law. Accordingly, the next part canvasses doctrine for traces of the court’s treatment of the constitutionality of closed universe searches while also independently assessing the legal status of closed universe searches.

II. Five Ways of Looking at a Closed Universe Search

Because this Article is the first to identify closed universe searches as a distinct species within the broader genus of indiscriminate technological searches, there is not yet a robust body of either scholarship or judicial opinions analyzing them with doctrinal precision. This may be due in part to the very nature of closed universe searches: Many persons swept into closed universe searches will never learn that they have been investigated, especially since little in the current doctrine requires disclosure of a search even once a suspect has been ruled out. Even if an individual learns about the search, they may have no meaningful legal recourse to assert a claim or demand the disclosure or expungement of their data—notwithstanding that closed universe searches can probe deeper into each person’s data than an indiscriminate or open universe search would.

Consider again the examples in Part I.B. For some closed universe investigations, there are search alternatives that under present doctrine would elude legal scrutiny. For instance, rather than compel DNA swabs, law enforcement can simply ask for voluntary samples or, easier yet, take them surreptitiously. The legal constraints on using cell-site simulators for short term tracking data are similarly fuzzy and ambiguous. Lastly, as Parts II.C and II.D detail, even if reasonable suspicion or probable cause is required for a closed universe search, a court may deem that standard met in some fraction of cases just by citing to whatever happenstance placed the person within the closed universe pool.

With these questions in mind, this Section maps out the variety of analytical roads that courts have taken, as well as the routes that scholars or law enforcement might travel, in assessing the constitutionality of closed universe searches. Specifically, this Section addresses (A) the inapplicability of constitutional or statutory law altogether; (B) a traditional warrant and probable cause requirement; (C) a modified warrant and probable cause requirement; (D) general reasonableness review and the “special needs” doctrine; and (E) statutory standards such as “reasonable grounds” and the subpoena requirement.

A. Not a Constitutional Issue

In many respects, closed universe investigations may proceed using methods that courts do not view as implicating the Fourth Amendment at all. To be clear, it is beyond question that some of the classic technological tools of law enforcement investigation constitute a search or seizure and therefore trigger Fourth Amendment scrutiny. Drawing from just some of the examples in Part I.B, the Supreme Court has already ruled a demand to supply a DNA sample is a “search,”[36] as is a request for extended cell-site location data.[37] But it is easy to imagine that many closed universe investigations may be undertaken using methods that implicate no constitutional or statutory rights at all.

To begin with, the constitutional status of the technological searches employed in many closed universe searches is far from established. In the words of one magistrate judge surveying the case law on methods like geofences, tower dumps, and cell-site simulators: Whether “the use of these technologies constitutes a search under the Fourth Amendment” is an issue “not settled in present law.”[38] To the extent that the investigation of the closed universe involves a targeted use of one of these technologies, such as a cell-site simulator intended only to capture data from a targeted number of phones, it is thus unclear that any Fourth Amendment intrusion has occurred at all.

Even for methods that constitute “searches” for Fourth Amendment purposes, a number of possible loopholes and alternatives in existing doctrine remain. For example, although compelling a person to give a DNA sample is a search triggering Fourth Amendment scrutiny, simply following the person around and collecting their discarded or “abandoned” cups or napkins is generally not.[39] Similarly, although requesting a person’s extended historical location data constitutes a search, obtaining data from a shorter period of time, erecting secret long pole cameras targeting the person’s home,[40] or using facial recognition software to cull their image from public surveillance cameras may not be.[41] Even the outer contours of Carpenter’s rewriting of third-party doctrine remain undefined: If law enforcement obtains historical cell-site records not from Google but from a private vendor selling such information on the open market, does Carpenter’s rule still apply?[42]

Even though closed universe searches may evade scrutiny because general constitutional doctrine does not recognize the activity as a “search” of Fourth Amendment stature, these searches should still garner concern. Some investigative methods are not practicable to deploy against the indiscriminate masses (such as collecting discarded DNA) but are easily done within the confines of a closed universe pool. Conversely, to the extent that a method is not concerning because it is feasible to deploy only against persons for whom there is already preexisting suspicion, then the predicate requirement of suspicion itself operates as some kind of limiting principle. Targeted searches by their very nature affect fewer persons, and at least some subset of those searches will come to light as a result of the arrest and charging of the person who was surveilled.

But the innocent persons in the closed universe pool have neither the protection of the crowd nor the constraining power of individualized suspicion. The members of the pool are vulnerable precisely because the pool is both small enough to investigate using the most intrusive methods, and yet they are unprotected by either individualized suspicion or judicial process. To give a concrete illustration in the context of DNA sampling, consider the practice of obtaining voluntary “rule out samples”—DNA samples intended to help police find a perpetrator. Such persons are in a group small enough to target for intrusive investigation yet protected by fewer doctrinal rules safeguarding against abuse or misuse, or even basic oversight as to the collection or retention of information, than the person actually arrested for the offense. It is perhaps no surprise that many such persons filed lawsuits after they learned that law enforcement had failed to limit their use of collected samples or comply with express terms under which the samples were collected.[43]

Lastly, closed universe searches will evade scrutiny precisely because of their very nature as closed universe searches. Even if courts might be inclined to impose restrictions on the use of closed universe search methods, those methods might never come to light. Many members of the pool will never learn that they were surveilled or investigated extensively or have any recourse or control over what happens to the data that was collected about them. It is these features of closed universe searches that Part IV seeks to remedy.

B. Warrant and Probable Cause: Contraband or Evidence of a Crime

What about closed universe search methods that clearly constitute a “search” or “seizure” under the Fourth Amendment? The canonical constitutional rule is that searches and seizures by the government require a warrant and probable cause.[44] But probable cause has proven somewhat elusive as a concept. Perhaps the most common formulation derives from Illinois v. Gates, where the Court defined probable cause as whether “given all the circumstances . . . there is a fair probability that contraband or evidence of a crime will be found in a particular place.”[45] Another iteration appears in Florida v. Harris, where the Court asked whether the available facts “‘warrant a person of reasonable caution in the belief’ that contraband or evidence of a crime is present.”[46]

How should a court apply this standard in a situation in which there is some quantum of belief that a person possesses evidence of a crime, but also other, clearly quantifiable evidence that they may not? When law enforcement seeks a warrant for a single target, the focal point of the courts is only on the amount of suspicion as to that single person. It is always possible that the law enforcement officer is wrong, or has the wrong person, but that possibility is latent and uncertain. In fact, the possibility that law enforcement has the wrong person need not just be latent; there could be actual evidence that law enforcement is wrong, but existing doctrine does not unequivocally require the disclosure of material exculpatory evidence in a warrant affidavit.[47] Thus, evidence of actual innocence is both insufficient to defeat probable cause and unnecessary for adjudicating the issuance of the warrant.

But when law enforcement seeks a warrant to compel information from a pool of persons in a closed universe search, it is inherent that each person is at least as likely not to have the evidence as to have it. It is easy to intuit this fact when you play with the numbers in the closed universe: Is there probable cause as to any one person if the closed universe contains only two possible suspects? Twenty? Two hundred? Should probable cause be assessed for the search holistically, such that the question is reframed as probable cause that this search will uncover the evidence (near 100 percent), rather than the “suspiciousness” of each specific person in light of the size of the pool (in this example, 50 percent, 5 percent, 0.5 percent)?

The Supreme Court addressed a similar argument in 1957 in Mallory v. United States,[48] long before advanced technological tools hit the scene. In Mallory, a woman was sexually assaulted in the basement of a small building while doing her laundry. The evidence narrowed the attacker to either the petitioner or one of his two nephews. All three were apprehended, and after a lengthy interrogation, the petitioner confessed.[49] Reversing the conviction, the Court noted that

[p]resumably, whomever the police arrest they must arrest on “probable cause.” It is not the function of the police to arrest, as it were, at large and to use an interrogating process at police headquarters in order to determine whom they should charge. . . .[50]

This implies that a one-in-three probability is not “probable cause.” It may also suggest that one-in-two is not sufficient, either. In the words of Tracey Maclin: “[S]ome view[] Mallory as requiring ‘evidence sufficient to charge one and only one person prior to arrest.’”[51]

Ybarra v. Illinois,[52] decided a couple decades later, seems to affirm this rigorous, and to some extent quantifiable, concept of probable cause. In Ybarra, officers obtained a search warrant for a tavern at which the bartender allegedly trafficked drugs.[53] A state statute permitted officers executing the warrant to detain and search any person found on the premises.[54] Ybarra was one of thirteen customers present at the time of the warrant’s execution, nine of whom were searched.[55] The Court held the search of Ybarra was unconstitutional and explained that “a person’s mere propinquity to others independently suspected of criminal activity does not, without more, give rise to probable cause to search that person.”[56] Notably, the Ybarra Court was referring to literal, physical proximity, but the principle could extend to the suspicion attached to any one person within a closed universe of suspects.

Following the reasoning in these cases, a person’s mere presence in a closed universe pool should fail, without more, to amount to probable cause for a search or seizure of that person. Drawing on the examples in Part I.B, if the four family members deny the request for DNA samples, police could not seek a warrant to compel them. Or if a geofence search identifies multiple phones on site at the time of the carjacking, police could not obtain a warrant to search or seize any one of them.

However, Mallory and Ybarra are both clearly cases from a different time. Current case law, including the cases discussed in Part II.C, suggests that the courts might not take the same stingy view of probable cause today. For instance, in Maryland v. Pringle,[57]the Supreme Court confronted facts similar to Ybarra—the arrest of three men in a car where contraband was found—and held the opposite. Technically, the Court distinguished rather than overruled Ybarra[58] on the ground that the facts allowed that all three might have possessed the contraband jointly. But the extent to which the remote possibility of multiple perpetrators was decisive is debatable; the analysis seemed to approve an “all or any arrest option when officers are confronted with multiple suspects.”[59]

In light of the case law, it seems conceivable that a court would hold that there is probable cause as to each person so long as there was a sufficiently closed universe of suspects, especially if that pool was created by some threshold of incriminating connection to the crime. After all, no one argues that probable cause is a certainty that a person committed a crime or has evidence of it, just some sufficient chance of it. As the Supreme Court has admonished: “Probable cause, we have often told litigants, is not a high bar.”[60] Meanwhile lower courts have never precisely quantified probable cause, and some have even seemed to depress the range of probability. Empirical work suggests that a fifth of judges would quantify probable cause as between 11 and 30 percent chance of guilt, and a third as below 40 percent.[61] Consequently, if there are only three suspects in a single-perpetrator crime, as many as one-third of judges could find it 33 percent likely that any one committed the offense and issue all three warrants on “probable cause.”[62]

On this reading, there may be probable cause as to many, even as there are doubts as to all. The existence of probable cause as to any single person in a closed universe search hinges on the size of the closed universe pool more than the existence of distinct suspicion as to any individual person within it. As one court recently explained in connection with a petition for a warrant for a canvassing cell-site simulator (CCSS), a sufficiently narrowly drawn request might suffice to meet probable cause:

[I]f a device is captured by three separate CCSS searches in sufficiently different areas where the suspect is seen or reasonably believed to be, that are sufficiently far apart such that the CCSS coverage areas are unlikely to overlap, then it is probable that the identified device belongs to the suspect. True, it is not impossible that other, uninvolved individuals will happen to be within the CCSS’s coverage area during each of the searches, but the Fourth Amendment deals in probabilities, not certainties.[63]

The judge’s reasoning refers to a single phone, found by narrowing the search to finding a device in three overlapping places. But if the Fourth Amendment “deals in probabilities, not certainties,” then probable cause is perhaps simply a matter of toggling the numbers of locations and devices to arrive at a lesser, but still viable, probability.

C. Warrant and Probable Cause: Aid in a Particular Apprehension or Conviction

Given that many closed universe searches will contain suspect pools of greater than two or three persons, the “probable cause as to each, even if doubts as to all” approach described in Part II.B will yield diminishing probabilities and is thus unlikely to resolve every case. But that does not mean that a warrant and probable cause is necessarily off the table for closed universe searches.

1.   “Aid” in the Apprehension of a Perpetrator

Case law makes clear the probable cause question has never been just about “probable cause to believe thisperson has committed a crime.” It has also been phrased as probable cause that “evidence of a crime is present,”[64] or even more generously, as “probable cause . . . to believe that the evidence sought will aid in a particular apprehension or conviction.”[65] This broader phrasing is critical to the analysis of closed universe searches because closed universe searches can be easily repackaged as evidentiary searches that aid in apprehending the perpetrator, rather than searches for the perpetrator.

This approach is evident in Bill v. Brewer, a Ninth Circuit case that addresses one type of closed universe search.[66] Brewer was a civil case brought by three police officers for deprivation of their Fourth Amendment rights in connection with an order compelling they each give a DNA sample.[67] The investigation involved the death of another officer under suspicious circumstances, and several unknown male DNA profiles were recovered from the scene.[68] Police obtained voluntary DNA samples from over one hundred persons who had entered the crime scene, including fifty police officers. But the three (officer) plaintiffs and two other officers refused to provide samples.

Investigators eventually obtained court orders compelling all five officers to give samples.[69] Although the orders were based on a statute nominally requiring only “reasonable cause,” precedent in the jurisdiction had interpreted the applicable standard as a warrant and probable cause.[70] The Ninth Circuit upheld the orders and conceded that “the orders here did not seek to obtain evidence that the plaintiffs committed a crime.”[71] In other words, application of an ordinary probable cause test—belief the person committed a crime or possessed evidence of a crime—would find probable cause lacking.[72] But the court defined the “critical element” of probable cause as “a nexus . . . between the item to be seized and criminal behavior.”[73] Accordingly, “in the case of mere evidence, probable cause for such a nexus must be examined in terms of cause to believe that the evidence sought will aid in a particular apprehension or conviction.”[74]

Applying this test to the officers’ claims, the Ninth Circuit concluded that there was “plainly . . . ‘a nexus’ between the crime under investigation and the evidence sought.”[75] Notably, by the time the samples were requested, investigators had already excluded the officers from suspicion. Consequently, it may be justifiable to reclassify the rationale in the case from a Constitution-implicating compelled search for crime-solving purposes (core Fourth Amendment fare) to a Constitution-implicating compelled search for purposes other than ordinary law enforcement (a category discussed further in Part II.D.3). But that is a stretch, as all parties conceded the reason to compel the samples was to help solve the crime under investigation.

Moreover, the Ninth Circuit deemed the fact that the plaintiffs were no longer suspects at the time of the search irrelevant.[76] Probable cause supported collection of the samples simply because they would “aid in the conviction of an eventual criminal defendant, by negating any contention at trial that police had contaminated the relevant evidence.”[77] Indeed, the Ninth Circuit noted that a few years earlier, in Messerschmidt v. Millender,[78] the Supreme Court had affirmed the idea that the “Fourth Amendment allows a search for evidence when there is probable cause . . . to believe that the evidence sought will aid in a particular apprehension or conviction.”[79] And Brewer is not alone in its interpretation. For example, the D.C. Court of Appeals applied this reasoning in In re G.B., upholding a warrant to compel a DNA sample from a victim of a stabbing. In so doing, the D.C. Court of Appeals expressly rejected the petitioner’s argument that “no valid investigative warrant may issue to forcibly take the DNA of a witness who is not suspected of the crime for which the DNA is sought.”[80]

Applying the reasoning of these cases, the relevant inquiry for probable cause in a closed universe search is not “what is the probability, given the size of the suspect pool, that this person is the perpetrator?” Rather, the question is “what is the probability that conducting the search of this person aids in finding the perpetrator?” In a closed universe search, that probability can approach 100 percent—the whole point is that taking DNA from the three suspects or checking the location history of the three phones or otherwise conducting whatever search is desired will almost definitively identify the actual perpetrator and by extension rule out the possibility that one of the other persons is the true perpetrator.

With that said, it is useful to observe that the court in Brewer did not give its unconditional approval to the grant of a warrant to compel the genetic material of known nonsuspects. In a particularly opaque passage, the court commented that even a search authorized by a judicial warrant could become unreasonable “by intruding on personal privacy to an extent disproportionate to the likely benefits from obtaining fuller compliance with the law.”[81] This resembles the proportionality approach championed by Slobogin.[82] For its part, the court drew its support from the line of cases addressed to searches of bodily cavities and other physical intrusions, such as Winston v. Lee, which found compelled surgery to recover a bullet from a robbery suspect unreasonable under the Fourth Amendment.[83]

Ultimately, however, the Brewer court waved away any such concerns in compelling the DNA samples. It reasoned that (1) a cheek swab for DNA is a minimal intrusion (both in terms of the brief seizure of the person and the search represented by the DNA on the swab); (2) there was reason to trust the investigators’ assertion that the sample would be destroyed immediately after comparison; and (3) public servants like police officers have diminished privacy interests.[84]

The lesson of Brewer for analyzing closed universe searches is that the simple fact that the persons within the closed universe were finite and ascertainable (at least to some extent) reflexively drew the court’s concern away from the common law instinct to strike down anything resembling the Founders’ hated “general warrants.”[85] The easiest way to illustrate the special treatment given to the closed universe search is to imagine the other extremes. Would the investigators in Bill v. Brewer have dared approach a court for an order to compel one hundred officers to give DNA samples? Or dared to ask for a warrant to search each of the five officers’ homes? Even the indiscriminate search cases, like the geofence or real-time cell-site requests, that have persuaded courts to grant warrants in seeming contradiction of the “general warrant” prohibition have typically been ones in which search parameters are defined narrowly.[86] In this way, the very narrowness of the closed pool and the perceived low risk/high reward of the technological search differentiates the closed universe search from its cousins: the universal, indiscriminate search and the targeted, suspicion-based search.

More significantly, the approach taken in Brewer mirrors (without citation, for the most part) that of the handful of other courts that have addressed orders to compel biological evidence from persons not suspected of the crime, including victims[87] and uninvolved third parties.[88] For purposes of thinking about the constitutionality of closed universe searches, Brewer is especially pointed because the court felt comfortable compelling DNA samples for a crime-solving purpose even when the officers were concededly not suspects in the crime, and thus the “nexus” or “aid” was pure process-of-elimination. It is hard to predict how the analysis would change had the perpetrator clearly been within the suspect pool; the “nexus” or “aid” is both stronger (because the search would definitively find the perpetrator) but also arguably more constitutionally troubling (because the target is among those searched).

2.   The Line Between Narrowly Drawn, Indiscriminate Searches and Closed Universe Searches

The general principles underlying the Brewer approach also lurk in the background of various courts’ analyses of the constitutionality of indiscriminate or open universe searches, particularly when such searches are narrowly drawn. Searches using geofences, canvassing cell-site simulators, or keywords all invert the usual warrant framework: Instead of seeking the data of a suspect, they seek suspects from the data. Courts are all over the map on the constitutionality of indiscriminate, open-universe searches, including the question of whether the Fourth Amendment applies at all.[89] The most developed case law is with respect to geofence warrants, which is presently pending before the Supreme Court.[90]

Lower courts that have applied Fourth Amendment scrutiny to these types of indiscriminate, open-universe searches have tended to endorse a two- or three-step process, which has also been endorsed by the service providers that hold the data.[91] This process is roughly that (1) the court authorizes an initial universal search that returns anonymized matching data; (2) law enforcement scrutinizes that data to narrow the list of candidates; and (3) once sufficiently narrowed, law enforcement receives the pertinent deanonymized data.[92] By breaking each disclosure into steps, courts can retain oversight of the amount and degree of information ultimately disclosed,[93] even going so far as to mandate data-handling terms such as destroying the nonpertinent data or restricting the scope of the use of data to the pertinent case.[94] But not every court requires judicial oversight of each step, and neither law enforcement nor the supplier of the data always follow the requirements laid out.

Most judicial opinions address the constitutionality of the initial warrant, or the “step-one” indiscriminate search. Fewer cases analyze step two or three, but some clues can be gleaned. Some courts have found a step-one warrant request sufficiently narrow or procedurally tailored to authorize not just the initial search but also the disclosure of any pertinent subscriber information without additional process.[95] Other courts have imposed a minimal threshold for additional disclosures, such as requiring a showing of “relevance”[96] for more detailed information at the step-two stage, although the data typically remains anonymized at that point and courts have reserved judgment on the showing necessary before the final step of obtaining deanonymized data.[97] Still other courts have affirmed that “probable cause”[98] or a “fair probability”[99] is required for each device to be deanonymized, yet do not always specify exactly how “probable cause” works in relation to multiple “suspects.”[100] And some courts have simply indicated additional process is necessary without clearly articulating what standards apply.[101]

In practice, adherence to this multistep process seems erratic. For instance, in United States v. Smith, law enforcement sought a geofence warrant for phones within a 98,192 square meter rural area within a one-hour window.[102] The court granted the warrant, with terms that mirrored the three-step process announced by Google: an initial return of anonymized data, further winnowing, and deanonymization.[103] However, although the warrant required “further legal process” after step one,[104] police did not apply for an additional order but instead simply sent a request for more information to Google, which complied.[105] After further winnowing, police sent a request to deanonymize the data, again without any further legal process, and Google again complied. Ultimately, without paying special attention to the legal standards applicable in steps two and three, the Fifth Circuit held the entire process unconstitutional, ruling that geofence warrants (“step one”) constitute impermissible “general warrants.”[106]

Conversely, in Jones v. State, the police sought a warrant for devices within one hundred meters of the victim’s home during a four-hour time span in the middle of the night.[107] The court granted the warrant, which, unlike the warrant in Smith, did not include language requiring further legal process. Yet investigators did ultimately return to court to seek a probable cause warrant to deanonymize a device they found suspicious, and both warrants were ultimately upheld by the appellate court.[108]

These cases also make plain that the line between an indiscriminate, universal search and a closed universe search can occasionally be fuzzy. The “aid in apprehension” cases tend to cluster more toward allowing searches of persons not suspected of the crime (even if that fact is not a basis for upholding the search), and the indiscriminate geofence cases cluster toward requiring that the search narrowly center around the perpetrator of the crime. Indeed, the more demanding that courts are in narrowing the parameters of a universal search, the more they begin to resemble closed universe searches.[109]

For instance, in a recent opinion by Magistrate Judge Beth Jantz denying a warrant application for a canvassing cell-site simulator, the court rejected as “too broad” the argument made by the government that “it only need establish probable cause that ‘use of the CCSS will lead to evidence of a particular criminal offense.’”[110] The court also rejected the contention that by limiting its search to devices shown to overlap or appear multiple times, the CCSS net would be more likely to catch the perpetrator and thus satisfy probable cause.[111] At the same time, however, the court acknowledged that a sufficiently defined group in terms of time and place could satisfy probable cause even if it invariably swept in some innocent persons.[112] Similarly, a D.C. District Court approved a geofence warrant, finding it “appropriately limited in scope,” in part by reasoning that “constitutionally permissible searches may infringe on the privacy interests of third persons.”[113] Other courts have upheld warrants on the same theory.[114]

Mary Fan has defined the concept of a narrowly drawn but still indiscriminate search as authorized by “digital probable cause,” or “probable cause that arises from data parameters sufficiently specified to give rise to a fair probability that evidence of a crime is contained within the data boundaries, as defined.”[115] The key feature is, again, how narrowly the indiscriminate search is drawn. In her words,

[w]hether there is the sort of fair probability required for digital probable cause depends on the circumstances of the crime, such as whether it was committed in a time and place with few people other than the perpetrator, and the nature of the data search parameters, such as whether the keyword search terms are likely to be used only by the perpetrator of a crime in planning and executing it. The greater the number of potential devices and users likely to be captured in the data search parameters, the less likely there is digital probable cause because gathering a large haystack of user data does not give a fair probability of finding the needle in that haystack.[116]

Applying Fan’s standard of “digital probable cause,” an indiscriminate search is rendered constitutional by the imposition of sufficiently narrow parameters, even if those parameters may not suffice to isolate only a single perpetrator.

The question, then, is whether Fan’s more elastic concept of digital probable cause should apply not just when the net is not blindly cast, as in the case of the indiscriminate searches she focuses on, but also when the searches are to be done only of those within a pool of known persons, as in the case of a closed universe search. As explained more in Part III, the critical distinction between an indiscriminate search and a closed universe search is not its narrowness or breadth; it is that an indiscriminate search casts a line into the anonymous open universe at inception, whereas a closed universe search already has a pool of identified targets. It is this distinction that should matter most to the constitutional analysis, and which may also partly explain why some courts seem to embrace a generous standard of “fair probability” when it comes to “step one” trawls but require particularity when it comes to subsequent deanonymization. At the same time, if a court finds that sufficiently narrow search standards are sufficient to satisfy probable cause for the search of an open universe of persons, it may also find probable cause satisfied with respect to a closed universe search in which each person in the pool is already somewhat suspicious.

D. General Reasonableness, Terry, and the “Special Needs” Doctrine

Another way to think about the constitutionality of closed universe searches is to ignore the warrant and probable cause default altogether, and instead review police practices according to the standard of general reasonableness. The longstanding debate among scholars as to how to read the Constitution’s Warrant and Reasonableness Clauses is beyond the scope of this Article,[117] but suffice it to say that current case law far from resolves the question.

In conventional doctrinal terms, once police action has been deemed a “search” or “seizure,” the default presumption is that the action requires a warrant and probable cause. But there are significant carveouts to this presumption: The Court has approved searches and seizures that require probable cause but not a warrant,[118] a warrant but not individualized probable cause,[119] and lesser intrusions based only on reasonable suspicion.[120] Traditionally, however, deviations from the warrant and probable cause requirement for searches have been authorized by the Court only in special circumstances, most pertinently in the “special needs” cases, such as when the government is acting other than with a “primary purpose” of “detect[ing] evidence of ordinary criminal wrongdoing.”[121]

1.   Terry and Reasonableness

The canonical exception to the requirement of a warrant and probable cause for ordinary criminal investigations is Terry v. Ohio, which endorsed brief, investigative stops and superficial searches (or “frisks”) on the grounds of reasonable, articulable suspicion.[122] Terry could be analogized to closed universe searches in that some intrusions could arguably be cast as brief and minimal (in large part because they are technological, rather than physical, in nature). If so, each person’s presence in the pool might amount to “reasonable suspicion” even if it does not meet the standard of probable cause.

That said, Terry stands as a fairly singular exception to the warrant and probable cause requirement even as the special needs and administrative search case law (discussed below in Parts II.D.2 and II.D.3) has blossomed. Moreover, although Terry dispensed with the requirement of a warrant and lowered the standard of probable cause, it still maintained a focus on individualized suspicion. The closed universe search, with its pool of suspects, seems a poor fit.

Perhaps more importantly, closed universe searches are more like conventional searches for evidence, deserving of full-blown Fourth Amendment protection, rather than a series of superficial, digital Terry stops. In fact, at least some of the methods used to isolate the perpetrator have been adjudicated constitutionally significant “searches” in Fourth Amendment case law. For instance, courts have clearly declared a DNA swab a “search,”[123] as well as obtaining extended location history data.[124] As such, it is highly unlikely that a court would carve a categorical doctrine for closed universe searches that would entail distinguishing identical intrusions based on whether the intrusion was motivated by individualized suspicion, collective suspicion such as in a closed universe search, or indiscriminate mass surveillance.

2.   Administrative Searches for Law Enforcement Purposes

More likely, a “reasonableness” analysis could justify relaxing the standard of probable cause even while maintaining the warrant requirement by following the example of cases loosely grouped under the heading of “special needs” or the “administrative search doctrine.” As 2011, Eve Brensike Primus observed that “[a]s science and technology advance, the frequency and scope of administrative searches will only expand.”[125] The geofence cases and statutes have already gestured in the direction of administrative search doctrine, as has the scholarly literature grappling with how conventional ideas of criminal procedure should map onto the distinct characteristics of new technologies.

Unfortunately, the administrative search doctrine has been described as “incoherent,” “abysmal,” “devoid of content,” and “a conceptual and doctrinal embarrassment of the first order.”[126] This lack of clarity and prominent contradictions make analytical precision difficult. Presciently, Primus observed that some of the precedential disorder arose from “the Supreme Court’s conflation of two distinct types of searches,” namely “dragnet searches” (or what this Article calls indiscriminate searches) and “special subpopulation searches,” which are searches of special subcategories of persons (like students, probationers, or government employees).[127] But within this taxonomy, it is already easy to see how closed universe searches fit, as they are a little bit of both: a narrow dragnet of special subpopulation.

For purposes of this Article, the most apt precedent may unexpectedly derive from an opinion issued half a century ago. In Davis v. Mississippi,[128] the Court decided a case in which police detained at least twenty-four young Black men over a period of ten days, during which each man was questioned and fingerprinted.[129] The Court, primarily focusing on the mass nature of the detentions and the length and scope of the investigation, held that fingerprints obtained from the man ultimately charged with the offense should be suppressed.

But, significantly for our purposes, the Court explicitly reserved the possibility that “the requirements of the Fourth Amendment could be met by narrowly circumscribed procedures for obtaining, during the course of a criminal investigation, the fingerprints of individuals for whom there is no probable cause to arrest.”[130] As support for that idea, the Court analogized to the special needs line of cases.[131] Davis, using “reasonableness” as the touchstone, thus contemplates the acceptance by the Supreme Court of the constitutionality of a kind of closed universe search: the compelled taking of fingerprints from a pool of persons under suspicion but for whom there is no probable cause solely for law enforcement purposes.

What is more, the Court in Davis offered this tacit approval in the context of an (assertedly) highly probative (if primitive) technological tool: fingerprinting.[132] In this respect, its rationale was telling. Although the Court expressly rejected the idea of mass suspicionless detention for purposes of interrogation, the Court left open the option for fingerprinting, reasoning that “fingerprinting is inherently more reliable and effective crime-solving tool than eyewitness identifications or confessions and is not subject to such abuses as the improper line-up and the ‘third degree.’”[133]

In other words, high-tech searches of closed universe suspects may be constitutionally reasonable precisely in part because they are reliable and less susceptible to abuse. It is not hard to imagine a future court quoting that exact line, only replacing “fingerprinting” with “DNA testing,” “historical cell-site location information,” or “cell-site simulators.” To the extent that a statutory scheme bolsters this conclusion, it seems even more likely to withstand constitutional scrutiny.

Several scholars have expressly argued for a “reasonableness” standard when it comes to Fourth Amendment review of new technologies, noting that it might afford greater flexibility to adjust to the nuances of searches in the technological context. For instance, in a white paper published in 2012, Peter Swire and I argued that “conventional approaches to the Fourth Amendment fail to constrain discretion when it comes to new technologies . . . .”[134] Building from the case law around checkpoints, loitering, and inventory searches, we argued for a four-step procedural approach that would substitute a substantive inquiry in favor of specific guidelines designed to safeguard rights.[135] Slobogin has argued for a “proportionality” analysis, which would require that “the justification for a search or seizure . . . be roughly proportionate to its intrusiveness.”[136] And Barry Friedman has pointed out how the pressure to require a warrant and probable cause had prompted courts to simply label intrusive police practices “not a search” rather than risk not allowing useful investigative methods.[137] Other scholars have similarly argued for various restrictions on technology-based searches, citing “reasonableness.”[138]

But, it must be said that while scholars have urged a “reasonableness” approach, and Davis v. Mississippi may have left the door open to one, the Supreme Court has thus far generally rejected it. Most recently, in Carpenter v. United States, the Court effectively doubled down on the warrant and probable cause standard as the proper default once conduct is deemed a constitutionally significant search conducted for ordinary law enforcement purposes.[139] In Carpenter, the Court addressed an order issued by the government for more than three months of the defendant’s historical cell-site location information (CSLI) pursuant to the “reasonable grounds” standard authorized by the Stored Communications Act.[140] Although the Court primarily occupied itself with resolving the “is this a search” question, the end of the opinion also addressed the question of what was required to render it constitutional.

The Court first recognized that “[h]aving found that the acquisition of Carpenter’s CSLI was a search, we also conclude that the Government must generally obtain a warrant supported by probable cause before acquiring such records.”[141] Although the “ultimate measure of the constitutionality of a government search is ‘reasonableness,’”[142] the Court noted that “our cases establish that warrantless searches are typically unreasonable where ‘a search is undertaken by law enforcement officials to discover evidence of criminal wrongdoing.’”[143] Even the Court’s explicit nod to the idea that “case-specific exceptions . . . may support a warrantless search” remained within the confines of established doctrine—such as the exception for exigent circumstances.[144] And although the Court expressly reserved judgment on indiscriminate, open universe practices like real time CSLI or “tower dump” data searches, it is as easy to imagine that those might not be deemed “searches” at all (as discussed further in Part II.A) as it is that they would be deemed searches allowable on something less than probable cause.[145]

3.   Administrative Searches for Purposes Other Than Law Enforcement

Lastly, it is possible that courts might affirm the general principle that searches for a law enforcement purpose require a warrant and probable cause, but then simply recast the purpose of the closed universe search as something other than “ordinary law enforcement.” By doing so, the courts could relieve any warrant or probable cause requirements.[146] A similar approach is evident in the line of cases that approve Fourth Amendment intrusions for the ostensible purpose of locating witnesses.[147]

Most pertinent to this analysis is Maryland v. King,[148] which upheld the programmatic DNA testing of felony arrestees. Applying a “reasonableness” standard, the Court in King described the true purpose of collection as to ascertain identity, not to solve crimes—a somewhat laughable premise given that the DNA collection program was not restricted to searches of the database of known contributors used to establish identity, but rather authorized searches of the databases of crime samples used to match known persons to criminal activity.[149] A court might similarly craft a distinction for closed universe searches, arguing that such searches are not for crime-solving but “to prevent unnecessary investigations of innocent persons” or (as in Brewer) to “exclude other suspicious persons from erroneous accusation” or even, in some cases, “to find witnesses.”

In the end, if general reasonableness has yet to clearly prevail as regards open universe searches, or in any context involving ordinary law enforcement, then it seems unlikely to succeed in a context somewhere between a targeted search (as in Carpenter) and an indiscriminate, open universe search (as in the geofence cases). Rather, it seems doctrinally more consistent, as explained in the Sections above, for courts to turn to elastic ideas of probable cause or hold such searches outside the scope of Fourth Amendment scrutiny altogether.

E.  Statutory Standards: Reasonable Grounds and Subpoenas

As Part I.B explains, one of the most important differences between universal, indiscriminate searches and closed universe searches is the finite and identifiable nature of the persons in a closed universe pool. Courts and scholars have struggled with how to analyze indiscriminate searches of large-scale databases (such as “Google, give me the number of any cell phone found within a ten-block radius of this address at this time”),[150] and deeply intrusive, targeted inquiries (“Google, list every location this specific phone has been in for the past six months.”). Closed universe searches complicate things further by lying in the in-between. Practically speaking, however, it may be that these inquiries have more in common with targeted searches inasmuch as law enforcement starts its investigation with a pinpointed set of identifiable persons. Especially since persons can, for instance, be subpoenaed or asked to give evidence on the basis of “reasonable grounds.”

Again, precedent points in contradictory directions. On the one hand, the primary limitation on subpoenas and statutorily authorized orders has historically been that they must be “relevant to a legitimate purpose; fishing expeditions are not permitted.”[151] Thus one line of cases about subpoenas focuses on those that ask too much from a person.[152] A case like Carpenter is a version of that: Subpoenas that probe too deeply may run afoul of the warrant and probable cause requirement.

But on the other hand, another line of precedent, the one most pertinent to closed universe searches, examines subpoenas and orders issued too broadly—i.e., to persons with at best tangential connections to the case. In In re McVane,[153] the court addressed subpoenas for personal financial and other records from “individuals who are married to or are immediate family members of the targets” of the investigation.[154] Applying a “more exacting scrutiny” to these subpoenas to those “whose connection to the investigation consists only of their family ties,” the court held that the agency had not made a sufficient showing of need.[155] It added that an “agency is not automatically entitled to obtain all material that may in some way be relevant to a proper investigation.”[156]

Subpoenas and court orders within the context of closed universe searches could be viewed in a similar light: as targeting persons with too peripheral of a connection to the investigation. But it seems far more likely that the direct, even if slight, connection between the crime and those within the closed universe pool would suffice. Certainly, subpoenas and orders are routinely issued to persons based on scant connections derived from the physical world, such as passing access to a crime scene. So why would that not be true with respect to a connection identified as a result of a universal, indiscriminate search?

By way of analogy, grand juries routinely subpoena multiple persons for the purpose of giving nontestimonial exemplars. Although such exemplars generally do not trigger Fourth Amendment scrutiny,[157] the calling of the person to the grand jury does. And yet, the mere fact that numerous people are ordered to appear is generally not deemed unreasonable or excessive, and courts typically do not worry too much ex ante about just how useful any particular person is likely to be or how clearly connected the person is to the case. Typical is the language of the Supreme Court in United States v. Dionisio describing the “historically grounded obligation of every person to appear, and give his evidence before the grand jury.”[158] Court orders and subpoenas thus might work to compel the persons in a closed universe pool to appear and provide evidence, or even to compel physical evidence like DNA samples or short-term location data.

But again, even this workaround only goes so far. The Supreme Court’s decision in Carpenter v. United States[159] upended reliance on orders and subpoenas as the primary means by which law enforcement obtains at least some digital information.[160] Carpenter did not just hold the request for extended cell-site location information a “search,” it also thwarted the use of the subpoena power as a loophole around the constitutional requirement of a warrant and probable cause when it came to records held by a third party in which the suspect “has a legitimate privacy interest.”[161] As the Court put it: “If the choice to proceed by subpoena provided a categorical limitation on Fourth Amendment protection, no type of record would ever be protected by the warrant requirement.”[162]

So at least inasmuch as the information sought from a closed universe suspect would require a warrant were it sought from a single target, then Carpenter may likewise require a warrant and probable cause inquiry akin to that undertaken in Parts II.B and II.C. But so long as the material sought falls outside that still nebulous core, then the mere fact that multiple persons are suspected, as opposed to one single target, likely poses little bar on the use of “reasonable grounds” orders or “relevance” subpoenas.

III. Conceptualizing the Distinct Interests Entailed in the Closed Universe Search

Part I expounded the concept of a “closed universe search” as distinct from universal, indiscriminate searches on the one extreme and targeted, individualized searches at the other. Part II canvassed the doctrinal approaches that courts have taken or could take in response to government requests to use technological tools to conduct closed universe searches. This Section argues that, regardless of their legal status, it is critically important to distinguish closed universe searches from both their broader and narrower counterparts.

What is the value of distinguishing the closed universe search from both its indiscriminate and targeted counterparts? Fundamentally, as Bambauer notes, the endeavor is worthwhile if for no other reason than that “Fourth Amendment theory and reasoning is just starting to find its legs in digital search cases.”[163] As such, and as illustrated by Part II, “[d]ragnets of every sort . . . still suffer from analytical chaos because of value judgments and predictions that too often stay latent in the scholarship.”[164] To the extent that scholars and courts are wrestling with digital searches, they often do so with underlying assumptions about the character of those searches that may not be true in every case. By surfacing the value judgments and predictions specific to closed universe searches, we can better debate, if not determine, the proper limit and scope of them. Thus, the rest of this Section identifies aspects of closed universe searches that differentiate them from either their universal or targeted counterparts.

A. Diminished Legal and Political Constraints Due to Blurred Lines

Perhaps the single most important difference between open universe, indiscriminate searches and closed universe searches is the inapplicability of any assumption that the universality of the search operates as a political check against excess or abuse. As noted in Part II, scholars (and courts) have broadly disagreed on whether indiscriminate searches implicate the Fourth Amendment at all, require warrants and probable cause, should stand on some lesser form of reasonableness review, or should be altogether forbidden.

Across these doctrinal threads, however, one theme that has consistently emerged is the notion that there is some degree of safety in numbers. And that is true in two ways. First, if everyone is subject to the same surveillance, the reasoning goes, then it is both inherently more fair and the democratic process is better equipped to respond with appropriate levels of protection or safeguards.[165] Indeed, the very universality of indiscriminate searches means that even judges or law enforcement may have incentives to exercise restraint, knowing that such searches may eventually comb through their own data. Second, the bigger the pool, the less likely that officers will be able to abuse their power or probe impermissibly deeply. Although some of this equation may change with the assimilation of artificial intelligence into such tools, at least at present there are simple logistical obstacles to deeply probing indiscriminate numbers of persons or storing and using their data for other purposes later. In this view, courts can relax their Fourth Amendment anxieties, secure in the knowledge that the political process and the high cost of data storage rates will pick up where constitutional law leaves off.[166]

But whereas there is arguably the “safety in numbers” backstop when it comes to indiscriminate searches, that assumption is diminished or nonexistent when the suspect pool is finite and defined. Everyone is a suspect when an indiscriminate search is conducted, but at the point that a closed universe search occurs, the scrutiny falls only on a discrete and politically vulnerable pool. At the same time, unlike a targeted search, it is also definitively clear that the search will impact innocent persons, and depending on the size of the pool, potentially many of them.[167]

The political valence of power thus shifts when the search moves from the open to the closed universe. Mere placement in the pool raises a specter of guilt that may only be augmented by any effort to resist investigation. If you are innocent, why won’t you just give a DNA sample? Or your location or search history on the day of the homicide? Legal authorities may be more inclined to approve intrusions on persons within the closed universe than they would for a person at random, or to approve a particularly intrusive technological method at all. Brewer exemplifies this point: Again, it is improbable that a court would find probable cause to compel one hundred police officers to give a DNA sample to “aid in the investigation” of the case, or for search warrants to retrieve DNA samples from physical objects within their homes, and yet the Brewer court easily found probable cause to order three persons to submit a DNA swab. Intrusions on a closed universe of persons using technological tools simply may feel inherently less like a “general warrant” than intrusions on a universal, indiscriminate mass—even if the only meaningful difference is numerical and technological.

In short, to the extent that political power may help patrol the use of open universe, indiscriminate searches, that power operates quite differently with respect to closed universe searches. The will of the masses is more likely to pressure those in the pool to disclose, rather than to resist, suspicionless intrusions. Doctrinal or scholarly arguments that rest on the idea that the breadth of technological searches alone serves as a political safeguard must therefore account for the contrary political dynamic when such searches neither pinpoint one suspect nor treat everyone as a suspect indiscriminately but rather isolate a discrete group to target with closed universe searches.

B. Heightened Risk of Overreach, Bias, and Abuse

A second common defense of indiscriminate or mass search methods is that their universality minimizes or eliminates the risk of abuse.[168] The idea here is that when an investigative method applies population-wide, or indiscriminately, it is less amenable to overreach, bias, and abuse. There are simply too many people in the pool to probe each too intimately, to harass its members unjustifiably, or to concentrate on any one piece of data well enough to misuse it.

There are reasons to question the strength of this argument as it relates to mass search methods,[169] but it has no traction whatsoever when it comes to closed universe searches. It may be true that the creation of a closed universe pool comes about by indiscriminate searching or happenstance—who had access to the building, whose phones were recorded near the robbery. But once created, the closed universe itself is not indiscriminate. At that point, in fact, the pool is most vulnerable, and the risk of abusive police practices is magnified. Rather than facing a universal morass of possible suspects or a singular target for whom there is individualized suspicion, police confront a short list of people to concentrate on, at least one of whom they have good reason to believe is the target.

Viewed in this way, searches of innocent persons within a closed universe pool arguably deserve maximalist protection. To make the point more concrete by way of illustration: Imagine a court permits a geofence order to identify any cell phones present at the location of two separate robberies.[170] That initial indiscriminate search is less likely to overreach or be susceptible to bias precisely because of its universality; to get any meaningful results, law enforcement is incentivized to narrow the generalized search as much as possible and without preconceptions as to the likely result. Conversely, if the search returns only one cell phone, then that alone provides individualized suspicion; depending on the facts of the incident, and any marginally additional facts, it might even amount to probable cause.[171] Further investigation of that person is thus well governed by existing doctrine and can proceed in a narrow, constitutionally conventional way. Even if that investigation revealed the person’s innocence, the collateral impact of the intrusion would be minimal—one person, one set of intrusions, one set of data.

In contrast, imagine now that the initial indiscriminate search does not return one suspect, but six. At this point, law enforcement has a closed universe of suspects. As to each person, there is some modicum of suspicion, but the mere existence of the pool as opposed to a single person means that the suspicion as to any single person falls shy of reasonable suspicion or probable cause. The obvious next step is to investigate the members of the pool. But, applying conventional Fourth Amendment doctrine, the lack of individualized suspicion stymies the use of methods requiring a warrant or suspicion, even as the very existence of a small group of suspects makes more intrusive investigations feasible and desirable.

So instead, law enforcement is incentivized to pressure those persons for a DNA sample, or trick them into giving one, or even extract one from discarded bits of trash. They may surveil the individuals, probe digital data obtainable on less than probable cause, or engage in the kinds of intrusive technological searches (like cell-site simulators) that as of now fall outside of constitutional scrutiny. Just as the suspicious-but-not-suspicious-enough closed universe pool invites investigation outside of constitutional strictures, so too does the certainty promised by technological methods most invite courts to endorse. The very challenge of showing probable cause or reasonable suspicion as to any single suspect in a closed universe pool means that law enforcement must resort to methods that do not invoke Fourth Amendment scrutiny, or courts must approve tactics on grounds other than a warrant and probable cause.[172] To the extent those methods may seem intrusive—say, surreptitiously gathering DNA from ten closed universe pool members, or using a cell-site simulator to track their movements—courts are less likely to be troubled given the connection, however tenuous, of each person to the crime. Courts may be enticed to stretch the meaning of probable cause or to issue warrants or orders to compel information, precisely because the closed universe provides a modicum of suspicion and the allure of the certainty of technology is strong. Yet, even once the target is identified, there is no clear set of rules governing how police may retain or use the dossiers they compiled about each of the remaining five innocent persons.

This is why the concerns about overreach, bias, and abuse arise with respect to closed universe searches more so than with either targeted or universal searches. Targeted searches are well governed by the presence of individualized suspicion and existing doctrine. And indiscriminate searches are protected by their numbers; it is both politically and logistically challenging to surveil everyone, much less harass them or retain their data for use later. But it is possible to do those things with a closed universe of persons. Moreover, whereas intrusive methods that affect the masses may prompt blowback or political action, those same practices perpetrated against a closed universe of suspects may not. In this way, it is not the initial pool creation that raises the greatest threat to the values of privacy and liberty that the Fourth Amendment aims to protect, or even a targeted search that arises if a single suspect is identified, but rather the searches of innocents within a closed universe pool. Whereas the political process and logistical constraints discourage law enforcement overreach when it comes to indiscriminate searches, and conventional Fourth Amendment doctrine is geared to address persons targeted directly with suspicion, those safeguards are compromised in the case of the closed universe search.

To be clear, the point at this juncture is not to pass judgment on whether police should do these things—it of course is the very objective of policing to identify and apprehend persons who commit crimes. The point is only that the dialogue and legal analysis surrounding technological searches often slips between the universal and the targeted, ignoring the special vulnerability of the closed universe pool in the constitutional order.

C. Tunnel Vision, Confirmation Bias, and Error

A related danger of closed universe searches that is lacking with regard to indiscriminate searches is the risk of tunnel vision, confirmation bias, and error. As explained, a defining feature of the closed universe search is the use of technological tools to isolate the likely perpetrator within the pool with high certainty. But although that certainty is part of the seductiveness of the closed universe search, technological search tools are not infallible, and even accurate searches can induce cognitive deficits that in turn lead to wrongful investigation or conviction.

Consider the case of Joao Monteiro.[173] Monteiro was accused of murdering a child in 1988 based on a partial DNA profile recovered from the child and tested in 2018.[174] That profile purportedly matched Monteiro’s son, whose DNA was in a department of health database,[175] but he was too young to have perpetrated the crime. As a result, the detective in the case focused on Monteiro and collected a DNA sample from him “because he was the first male she found related to” the son.[176] A miscommunication between the crime scene analyst and the detective led the detective to erroneously believe the samples matched when in fact they did not,[177] and the detective’s affidavit for arrest and a confirmatory DNA sample contained numerous instances of either intentionally misleading statements or unintentional exaggerations driven by the belief that the DNA proved Monteiro’s guilt.[178] It was only after the prosecutor reviewed the evidence six months after the case was filed that the errors were revealed and the case dismissed.[179]

The Monteiro case illustrates the risk that overreliance on technology perceived to be infallible induces motivated reasoning that ultimately undermines the integrity of an investigation. The literature on wrongful convictions, drawing from work within the social sciences, is replete with examples of known phenomena that distort cognition.[180] For instance, tunnel vision is the natural tendency to “focus on a suspect, select and filter the evidence that will ‘build a case’ for conviction, while ignoring or suppressing evidence that points away from guilt.”[181] Confirmation bias is a related type of “selective thinking” in which “our inclination is to confirm rather than to refute” a hypothesis once formed, and thus we “tend to look for supporting information, interpret ambiguous information as consistent with our beliefs, and minimize any inconsistent evidence.”[182] One study even showed that tunnel vision and confirmation bias are the two most common factors in wrongful convictions.[183]

The danger of a closed universe search is that it invites precisely the kind of motivated reasoning that studies show can undermine the integrity of an investigation and lead to wrongful accusations or conviction. Faced with a closed universe pool, investigators may narrow in on otherwise coincidental or marginally culpable information as evidence of guilt and ignore or dismiss contrary information.

Moreover, it is critical to remember that technological tools are not infallible. To be sure, in some cases, the closed universe search methods may operate as a counterweight to cognitive deficits, such as if a DNA test comes back as clearly nonmatching. But even highly reliable technologies are not impervious to error.[184]

Recall the opening example of the murder on the Orient Express. One lesson from the counterfactual example, in which Poirot simply face-IDed the killer, is that the technology, although not wrong in identifying the suspect, would have missed all the others. In the same way, results may be more nuanced than they first appear, inferences from the data may be contestable or misunderstood (as in Monteiro), or seemingly inculpatory data may have exculpatory explanations. And as closed universe searches become more popular, true perpetrators may do more to affirmatively obscure their identity. The pool may have five phones that were in the area at the time of the crime, but the actual perpetrator may have known better than to bring along a trackable device.

D. Publicity, Privacy, and Transparency

Lastly, closed universe searches are distinguishable from indiscriminate searches in two additional ways, each an alternate of the other and both related to their public nature. First, by effectively treating everyone as a potential wrongdoer, universal searches remove any stigma from being branded a potential wrongdoer. No one walks through TSA security or passes by a license plate reader and thinks “everyone is looking at me like a criminal” or “I might get fired if my boss hears about this.” But closed universe searches do the exact opposite: By focusing only on a discrete group of persons, the closed universe search taints each by association. Even if a single person is ultimately identified as the actual perpetrator, the other persons’ mere presence in the pool of suspects has the potential to haunt their reputation within their community.

But that is assuming, of course, that the persons in the closed universe pool were identified or investigated in a manner that made their identities as suspects known to their associates or the general public.[185] It is equally as likely—and this is the second way that universal searches may differ markedly from closed universe searches—that the closed universe searches never come to light at all.

Universal searches, by their very universality, are hard to keep secret. Some are transparent by their very nature—a universal DNA program would be hard to keep secret, even if done via established practices like newborn blood spot testing.[186] Similarly, as soon as a technology like facial recognition software or the admission that a person was found via a geofence warrant is made public in a criminal trial, the public learns that universal searches can and do take place under certain circumstances. At least in theory, that allows for generalized debate about the desirability or scope of such searches.

In contrast, the balance between publicity, privacy, and transparency is far more complex when it comes to closed universe searches. The fact and extent of many if not most closed universe searches are much more likely to remain shrouded in secrecy, especially those derivative of an initial universal search. There is no general obligation for law enforcement to disclose all the persons they investigated that did not commit the crime or all the collected data unrelated to the defendant on trial. There is no incentive for police to reveal that, on the road to identifying the charged individual, they also collected enormous troves of intrusive data on six other persons. So, in many cases, such information will never come to light—whether to the individuals investigated or to the public at large.[187] And thus there will be no debate or democratic exchange about the justification for such searches.

That said, it is also desirable that closed universe searches take place in complete secrecy. The very fact of labeling an individual as a person of interest in a crime can have tremendous stigmatizing effect, not to mention collateral impacts on the person’s interpersonal, professional, and financial future. In today’s information environment, it is paramount to protect the identity of innocent persons placed in a pool by happenstance so that their name is not forever tarnished by association.

How, then, to balance these two competing concerns? Allowing intrusive closed universe searches to unfold in total secrecy and without any democratic oversight is not consistent with the values of liberty and autonomy embodied in the Fourth Amendment. But publicizing closed universe searches is also not without its perils. Without public debate, important questions go unresolved; individual rights are left without a means of vindication. What happens to the DNA surreptitiously collected from the innocent third parties in closed universe searches? Or the data exposed from probing their cell phone patterns or personal records? Who monitors whether law enforcement practices conform with law, much less are free from inaccuracy or abuse, when it comes to those not charged with the crime? Conventional search methods tend to reveal themselves; a person knows when they are being interrogated, physically searched, or their belongings seized. But technological methods can probe deeply without any notice to their subjects at all.

It makes little sense to afford the target of an investigation more rights to challenge search practices via a Fourth Amendment motion, or to expunge wrongfully obtained data once revealed through an investigation, than are available to the innocent persons swept up in an investigation. Yet the innocent persons in a closed universe pool will in many cases be unaware that they were ever even tracked, surveilled, or surveyed, and so have neither the information nor opportunity, or perhaps even legal basis, to challenge either law enforcement’s activities or its retention of the data. Again, our system is designed for a world in which the closed universe search pool was an oddity—dependent on the accident of physical proximity or fortuity, which alone might lend a patina of guilt by association. It is not designed for a world that regularly produces closed universe pools by technological happenstance and then enables deeply intrusive searches into the data of those persons within the group, with little recourse for persons wishing to challenge the collection, retention, or later use of that data.

IV. Formalizing the Closed Universe Search Framework

Part I conceptually distinguished closed universe searches from both universal, indiscriminate searches and targeted searches. Part II analyzed the application of conventional constitutional doctrine to closed universe searches. Part III identified concerns specific to the closed universe context that are either lacking or less salient when it comes to universal or targeted searches. This Section now closes by recommending that legislatures pass closed universe search statutes that specifically address the specific context and concerns raised by these searches.

To reiterate, this Article takes no position on the constitutionality of the indiscriminate searches that often precede and create closed universe pools, which is still a nascent legal question over which courts are divided.[188] Rather, this Section starts from the proposition that closed universe pools exist, no matter how they came into being, and endeavors how to think of what rules govern the authority to search or seize each member of the pool.

The easy answer is to say that closed universe pools deserve no special scrutiny or oversight, because the ordinary rules of criminal procedure more than adequately address an individual’s rights whether they are the lone suspect or one of three (or more). The problem with this view is that it fails to acknowledge the particular vulnerability of closed universe search suspects, as explained in Part III. It is also not enough to say that the initial focus on the pool-creation step addresses any concerns that may arise at the closed universe search stage, again for the reasons stated in Part III.

Moreover, as Part II explained, there are too many ways in which closed universe searches can fall through the doctrinal cracks in the Fourth Amendment. Courts may be willing to stretch concepts like “probable cause” or “reasonableness” when it comes to closed universe searches, even though they would not do the same in other circumstances because otherwise there is no avenue for obtaining such information lawfully. Or existing Fourth Amendment doctrine may incentivize workarounds and loopholes (such as interpretations of the “reasonable expectation of privacy” that read out whole categories of data, or broad application of the abandonment or third-party doctrines) that invite deceptive or coercive investigative tactics. While such risks have always been present in an investigation, closed universe searches substantially heighten the probability that such tactics will be deployed against innocent persons, based on nothing more than a happenstance connection to a crime.

The best way to protect innocent persons in a closed universe search is by legislative intervention.[189] Doctrinally, a properly drawn statute would comport with the Fourth Amendment both because it governs conduct altogether outside of law enforcement protection, and because it adequately circumscribes law enforcement discretion enough to satisfy a “reasonableness” standard. Practically, a closed universe suspect statute would help usher closed universe investigations away from the shadows and toward the light of judicial oversight.

The core objectives of any regime to govern closed universe searches must be threefold. First, a statutory regime must ensure judicial supervision over the scope and extent of investigations of persons in the closed universe, even when the Fourth Amendment may not clearly apply to that activity when undertaken in either an indiscriminate or targeted context. In other words, a statutory scheme must offer additional protection to innocent persons ensnared by happenstance in a criminal investigation even when the methods used otherwise do not implicate a reasonable expectation of privacy. It must also limit the use of workaround or loophole methods for conducting searches that are simply not feasible on the mass or indiscriminate scale (such as surreptitious DNA collection) but attractive and viable given only a closed universe of suspects. Second, a statutory regime must circumscribe such investigations with specific parameters, such as case type, degree of intrusion, and scope of affected persons. Lastly, a statutory regime must clearly provide for the protection and destruction of data collected about persons later determined to have no connection to the criminal offense.

This Section proposes a “material evidence” framework for addressing closed universe searches. Ideally, this framework would be implemented through legislative enactment. But even in the absence of legislation, this framework might guide courts applying the Fourth Amendment in situations where the analysis of individualized suspicion and probable cause is complicated by a closed universe of suspects.

A. Closed Universe Search Framework: The Material Evidence Warrant

A statutory scheme governing closed universe searches should address the three critical stages of such investigations: (1) narrowing the pool; (2) searching the pool; and the (3) aftermath of the search. Although the specific contours of the statutory provisions are rightly left to each jurisdiction’s development, this Section offers some general guidance for legislators crafting these terms, and courts reviewing them.

1.   Narrowing the Pool

A statutory regime should start by specifying what precisely constitutes a closed universe search, as any proposed category raises many questions. Suppose the search of license plate reader data returns one hundred drivers as the pool of interest. Is the next step in the investigation another “universal” search or a “closed universe search”? I can say with absolute confidence that a return of three drivers satisfies my conception of a “closed universe” and five hundred does not. Similarly, does a cell-site simulator trawling a huge public crowd constitute an indiscriminate search, but a cell-site simulator trawling a ten-unit apartment building constitute a closed universe search? What role does anonymity play?

One answer might be to say that a closed universe search targets a defined population, whereas an indiscriminate search does not. But such a test itself is inadequate and puts the category at the mercy of linguistic gymnastics. TSA may search everyone who goes through airport security, but that does not mean they search everyone who “travels” (say by bus or boat or train). In this way, many indiscriminate searches are not in fact truly universal or indiscriminate.

Another take might try to define the category numerically. But closed universe searches could have as few as two “suspects” and as many as one thousand. In most cases, the closed universe search is only likely to be attractive if it can be done against a tractable number of “suspects”—that one thousand is likely to be winnowed down simply because the alternative of searching all of them is too cumbersome. But if the case is important enough, or the investigators diligent enough, then a one-thousand-person pool might nonetheless still qualify as a closed universe search.

Ultimately, the precise number is not what makes a closed universe search; it is its substantive composition. Specifically, this Article began by defining a closed universe search as a search (1) of a small pool of persons connected by happenstance to a crime; (2) one of whom is almost certainly the perpetrator, but the rest of whom are innocent; (3) using new technological tools that can identify that perpetrator with certainty or near certainty. Applying this definition, the key is that the targeted population in a closed universe search is one already marked with some suspicion, even though some in the group are unquestionably innocents. An analogous idea known as “k-anonymity” is found in computer science.[190] To put it simply, k-anonymity is a measure for the existence of data privacy. It is present when “the information for each person contained in the release [of group data] cannot be distinguished from at least k−1 individuals whose information also appears in the release.”[191] Roughly applying that idea here, it might be said that an indiscriminate search is one in which the databased persons maintain k-anonymity, whereas a closed universe search is one in which the pool lacks such anonymity.[192]

The key point for purposes of this Article is that a legislator drafting a statute or a court confronting police practices should attend to the dynamics that distinguish universal (or, for that matter, targeted) searches from closed universe searches and craft doctrinal standards appropriate to each category of search. Not because there is a rigid numerical point at which a search becomes a closed universe search, but because approving an indiscriminate search may lead to the creation of a closed universe search pool that requires additional judicial intervention responsive to the unique concerns raised by closed universe investigations. That also means that universal or indiscriminate searches may permissibly iterate multiple times before creating a closed universe pool.

2.   Searching the Pool: Carefully Articulated Standards

Once the closed universe pool of suspects is defined, the next step is to determine whether and when additional investigative measures within that group are permissible. Here, the statutory regime does its most important work, and a comprehensive statutory regime must address three issues.

a.   Governance of All Sophisticated Technological Methods Regardless of Whether They Constitute a Fourth Amendment “Search” or “Seizure”

The regime must broadly encompass all investigative methods deployed against the suspect pool to avoid inadvertently endorsing the Fourth Amendment’s cramped notion of “search” or “seizure.”[193] And it must expressly provide that the statutory framework is the only permissible means for conducting such searches (other than a traditional warrant and probable cause) to avoid incentivizing workarounds including softly coercive or deceptive behavior. What is the point of a right to refuse DNA sampling if police can simply take DNA surreptitiously or with softly coerced “consent”?[194] And whereas it may not implicate the Fourth Amendment for police to collect one person’s short-term historical location data, is it so clear that law enforcement gathering twenty peoples’ short-term historical data should be given lesser protection than long-term data gathered on one person?[195]

The very point of the statutory regime is to provide the exclusive means by which closed universe investigations take place, precisely because the closed universe pool contains mostly innocents (as in the indiscriminate search), yet those innocents are vulnerable because they are not just one among the masses. As opposed to the indiscriminate search context, in the closed universe search context there is an unmistakable adversarial law enforcement motive and a manageable population to direct it against. Safeguarding closed universe searches ensures that the investigative ease of new technological methods does not result in secretive, intrusive searches of cohorts of persons connected only by happenstance to a criminal act.

b.   Articulation of Clear Legal Thresholds for Each Intrusion with Continued Judicial Oversight

The statutory regime should set forth clear legal standards defining the procedures necessary before the closed universe may be searched and provide a mechanism for continued judicial oversight (whether by a “warrant” or another judicial instrument). To be clear, if a method used for a closed universe search is clearly covered by the Fourth Amendment on its own terms (such as compulsory DNA sampling or retrieving long-term historical cell-site information), then of course no statutory regime can override the Constitution’s dictates. But if the approach is unprotected by the Fourth Amendment (such as a “voluntary” or surreptitious DNA sample, consent-based methods, or short-term location data), then the statute’s terms would control.

This Article does not provide a comprehensive statute; it is better that jurisdictions have a measure of flexibility in prescribing the precise set of standards to apply. But some basic principles are worth elaborating.

First, a closed universe statute should narrow the types of crimes for which such searches are permissible. Put plainly, to the extent that technology makes suspects of us all, and enables profound invasions of personal privacy, the trade-off of public safety ought to be worth it. The state is treading on shaky constitutional ground when it authorizes closed universe searches; these searches go against the foundational understanding of the Fourth Amendment as premised on individualized suspicion. To the extent that modern times call for modern methods, must the state really collect people’s DNA or cell-site histories or other intrusive information in service of solving petty or minor offenses? Again, each jurisdiction may have its own idea of what constitutes an eligible offense, but there should be some constraints on law enforcement given that the ordinary constraints of resources and manpower are so greatly diminished by technological methods. At the very least, the degree of authorized intrusion should toggle to match the seriousness of the crime under investigation.

Second, where applicable, a statutory regime should take pains to distinguish the step of deanonymizing the persons identified as within the closed universe pool from any further steps undertaken to investigate those persons, whether through conventional methods or a closed universe search. This will not be necessary in every case; some indiscriminate searches (such as a genetic genealogical search or facial recognition search) are inherently deanonymized. But for indiscriminate methods that create pools without simultaneously revealing identity to investigators (such as a geofence or reverse keyboard search), the deanonymization step should be considered its own independent moment, warranting appropriate tailoring and justification, even if to a lesser extent than would any ensuing intrusive closed universe search.

Third, a statutory regime ought to specify specific standards of suspicion for particular intrusions, whether probable cause, reasonable grounds, or some other standard, and moreover further specify suspicion as to what. We have seen the difficulty in assessing the quantum of suspicion meant by a phrase like “probable cause” (how probable is “probable”?), as well as the malleability in defining the nature of the suspicion (will this search find the perpetrator of the crime or simply produce evidence that would aid in solving the crime?). By definition, a closed universe search is not one supported by probable cause that this specific person committed the crime. Rather, a search is based on an idea of collective suspicion and conclusive resolution: that this person is among a defined number of persons for whom there is a strong indication that one among them committed the crime, and it is all but certain that a search will confirm or relieve that suspicion.

Notice, then, that in place of individualized suspicion are two analytically significant factors that the statute should endeavor to quantify: the degree to which membership within the closed universe group itself supports the suspicion and the degree to which a technological method can firmly prove or disprove that suspicion. The first factor is likely to hinge on the size of the closed universe group, how it came about, and the probability that inclusion in the group supports an inference of culpability. The second factor turns on how certain the proposed search is to confirm or relieve suspicion because the search method is highly accurate and efficient, or the nexus between the method and the inference of culpability is particularly strong. Put plainly, it is one thing to authorize the taking of DNA samples from three persons whose cell phone data placed them in the private location where the crime occurred to determine whether there is a match to an intimate swab from the crime scene victim. It is another thing to authorize the taking of DNA samples from fifteen persons who happen to pass through a public area to see if any may match to a forensic sample that simply affirms that the person was in fact in that place.

Fourth, a statutory regime should expressly include requirements for minimization and tailoring. Searches should not exceed their necessary scope, and incremental steps may be appropriately imposed to help winnow the pool even within the closed universe. For instance, investigations should proceed without deanonymization for as long as possible and should presumptively authorize use of intrusive methods only sequentially, not concomitantly. At each stage, law enforcement should show they have exhausted all reasonable measures to continue to narrow the pool before a more intrusive step is taken.[196]

Lastly, the regime must provide for continued oversight by courts as a closed universe search progresses. Monitoring ensures both that authorized searches do not exceed their permissive scope and that each new intrusion is as narrowly tailored and justified as possible. Judicial oversight also ensures compliance with retention and reuse rules, as detailed in the next Section.

By way of closing example, consider Maryland’s 2021 law comprehensively regulating genetic genealogical searches.[197] The first step of such searches is to trawl recreational genetic databases looking for matches; step two involves the creation of trees of suspects from those searches and publicly available information; and step three entails obtaining DNA samples from possible perpetrators. As with all closed universe searches, the applicability of the Fourth Amendment to such investigations is unclear. The trawl of the public database and composition of suspect trees likely does not constitute a Fourth Amendment “search” at all, and any subsequent investigation within the closed universe may likewise be done without invoking constitutional doctrine (such as by secretly obtaining DNA by collecting discarded trash).

Troubled by this implication, the Maryland legislature enacted what is effectively a DNA-specific closed universe search statute. It sets out a governance scheme addressing genetic privacy but also accuracy, public policy concerns, and quality control. For purposes of this Article, however, its provisions addressing the closed universe search aspects of an investigation are most pertinent. The scheme expressly limits such searches to the most serious criminal offenses.[198] It contains a dedicated provision regarding the collection of DNA samples from persons identified by such searches as either potential suspects or relatives whose DNA might aid the investigation—i.e., closed universe suspects.[199] And its terms include a requirement of informed consent and a prohibition on covert or surreptitious collection (which would otherwise be unregulated by the Fourth Amendment) except with court approval.[200] It also contains provisions addressed to data preservation, retention and use, which leads to the final component of a statutory regime.

3.   The Aftermath of Searches: Restrictions on Data Disclosure, Preservation, Retention, and Use

The last piece of a statutory regime governing closed universe searches should address the disclosure, preservation, retention, and use of obtained data.

First, with regard to disclosure, a statute should both ensure that persons investigated in a closed universe search are aware that their data has been accessed, absent special circumstances, but also restrict the sharing of that information beyond those persons. Disclosure is essential as a safeguard against abuse: It provides a route for accountability if such searches become excessive or fail to adhere to statutory standards. At the same time, implicating innocents in what may be serious criminal activity should not be done lightly. Being lumped in a pool of suspects, even by happenstance, can harm a person’s professional, interpersonal, or social relationships. So, care should also be taken to ensure that any disclosures in the closed universe phase be limited and no greater than necessary for the investigative purpose, until a suspect is confirmed. A scheme can provide a limited mechanism for surreptitious investigations based on a showing of true need, such as a safety risk if a potential perpetrator were to learn of the investigation.

For the same reason, a statutory mandate should require public reporting of aggregate data on the use and success of closed universe search methods. Such data should track the type of crime investigated, the number of persons identified within the closed universe, the methods used to identify a possible perpetrator, and whether that person was ultimately charged or convicted. Such data can help the public assess whether the privacy and liberty interests of closed universe searches, or even of particular methods used to winnow pools to a probable suspect, are justified for reasons of public safety. Other accountability mechanisms (such as civil relief and damages) should also be part of closed universe search statutes to bolster enforcement of the regime.

Second, a statutory scheme should directly address the preservation, retention, and use of data rather than focus solely on the constitutional question at the point of acquisition or collection. The asymmetrical attention given to collection, rather than continued use, has long been a subject of critique even for targeted searches. Over a decade ago, I argued that

rather than follow an industrial age model reliant upon physical acquisition, constitutional doctrine [should] transition to an information age approach based on knowledge, creation, and dissemination. Such attentiveness would offer more effective safeguards around the creation and utilization of databases, and be responsive to concerns about insufficient auditing structures and function creep.[201]

Other scholars have similarly called for courts or legislatures to take a more proactive supervisory role over data searches, whether in general[202] or in specific contexts.[203] As courts continue to abdicate any role for the Fourth Amendment in safeguarding lawfully acquired data, it becomes increasingly essential for legislators to pick up the slack.

In the context of upholding or approving indiscriminate or closed universe warrants, courts have already required or noted approvingly of restrictions on the subsequent retention and use of data collected from persons ultimately proven to be innocent.[204] These admonitions are important when it comes to universal searches, but they are most indispensable with regard to closed universe searches. The persons swept into a closed universe investigation not only have their privacy and liberty interests threatened by invasive technological searches, but there are also no guarantees that the information gathered in the investigation will not be kept, used, or misused in perpetuity. A statutory scheme must therefore include terms that restrict the unnecessary retention or collateral use of lawfully collected data.

B. A Final Caveat: Robust Judicial Review

Although this Section argues for legislators to clearly restrict the scope and terms of closed universe searches, particularly since such searches might otherwise evade constitutional scrutiny altogether, this does not mean that there is no role for courts. There is a common call from both judges and scholars to pin Fourth Amendment protections more directly to legislative prescriptions.[205] In this view, the dynamic nature of high-tech searches is better suited to legislative, not judicial, governance and oversight. That may make good sense when it comes to universal, indiscriminate searches,[206] but the balance may shift for closed universe searches.

First, and most pertinently, is the political problem identified in Part III. Whereas all people with a cell phone may be a politically powerful constituency, the discrete group of individuals identified near a bank at the time of the robbery is not. But second, the Court that is entrusted with reinforcing the fundamental values of the Constitution—values like the presumption of innocence, freedom from unreasonable search and seizure, and the like—does so in the face of the powerful and constant pressure to compromise those values in the name of public safety. Another way of saying this is that fundamental rights should not come down to a vote,[207] although of course that is not an uncontested view among either political theorists or constitutional scholars. To the extent that the closed universe of suspects is vulnerable, Fourth Amendment protections should be robust and available to intercede. Ideally, the Fourth Amendment would supplement and reinforce the legislative branch in this regard—but it also should be ready to thwart the will of the majority in service of individual rights if warranted.

That means that although legislatures should have the opportunity to craft statutory schemes with varying specific terms, courts should not presume those schemes constitutional, or worse yet abdicate any supervisory role at all, without closer scrutiny. Just as the Carpenter court found the Electronic Stored Communication Act’s provisions constitutionally sufficient to safeguard Fourth Amendment rights in long-term cell-site data, so too should courts aggressively review closed universe suspect statutes to ensure that their provisions adequately enforce constitutional standards.

Conclusion

New technologies have brought tremendous opportunities to enhance reliable and efficient crime solving and public safety. But, as many courts and scholars have noted, they also threaten to fundamentally shift the balance of power between the individual and the state and expose the intricacies of private life to police through an ever-growing, pervasive network of surveillance. Ensuring that the Fourth Amendment, whether enforced by courts or instantiated by the legislative branch, continues to serve its role as a bulwark against unreasonable searches and seizures requires close attention to the particular nature and dynamics of police investigations in a high-tech world. This Article aims to aid in that conversation by identifying and defining closed universe searches as a distinct and important site of constitutional confrontation—different from open universe, indiscriminate searches on the one hand and individualized, targeted searches on the other—and by promoting a statutory solution to ensuring that such searches occur in a manner consonant with the Fourth Amendment.

Copyright © 2026 Erin Murphy, Norman Dorsen Professor of Civil Liberties, NYU School of Law. Thank you to my incredible research assistants Jordan Waller, Cindy Chang, Micah Musser, and Sarah Emmerich. I am also grateful for the thoughtful feedback given to me on earlier drafts by Chris Slobogin, Eve Brensike Primus, Barry Friedman, and Stephen Schulhofer, and to Jeff Bellin and the participants at the Vanderbilt Law School Criminal Justice Roundtable for their helpful comments and close engagement with this Article.

       [1]. Agatha Christie, Murder on the Orient Express (1934).

       [2]. For just a slice of recent work, see, for example, Barry Friedman & Danielle Keats Citron, Indiscriminate Data Surveillance, 110 Va. L. Rev. 1351 (2024) [hereinafter Friedman & Citron, Indiscriminate Data Surveillance]; Christopher Slobogin, Suspectless Searches, 83 Ohio St. L.J. 953, 954 (2022) [hereinafter Slobogin, Suspectless Searches]; Andrew Guthrie Ferguson, Persistent Surveillance, 74 Ala. L. Rev. 1, 13–14 (2022); Barry Friedman, Lawless Surveillance, 97 N.Y.U. L. Rev. 1143, 1143–45 (2022) [hereinafter Friedman, Lawless Surveillance]; Christopher Slobogin, Virtual Searches: Regulating the Covert World of Technological Policing 11 (2022) [hereinafter Slobogin, Virtual Searches]; Gregory Brazeal, Mass Seizure and Mass Search, 22 U. Pa. J. Const. L. 1001 (2020).

       [3]. Geofencing is a method wherein law enforcement requests “historical location information for Google-connected devices for which location data is being collected and that appeared in a given geographical area in a given time period.”Beth W. Jantz, Simulating More Particularity: Ideas for Approaching Search Warrants for Geofences, Tower Dumps, and Cell-Site Simulators, 16 Fed. Cts. L. Rev. 9, 12 (2024) (describing methods including geofencing, tower dumps, and cell-cite simulators).

       [4]. Compare, e.g., Bill v. Brewer, 799 F.3d 1295, 1301 (9th Cir. 2015), cert. denied, 577 U.S. 1194 (2016) (upholding warrant to compel three persons to give DNA samples to exclude them from sample from crime scene, finding probable cause that their samples would aid in the investigation of the crime), with State v. Contreras-Sanchez, 5 N.W.3d 151, 168 (Minn. Ct. App. 2024) (leaving open what standard would apply to deanonymization of cell phones identified by geofence warrant, having authorized additional searches of a subset of phones identified in initial indiscriminate search on basis of “relevance” showing).

       [5]. See, e.g., United States v. Smith, 110 F.4th 817, 838 (5th Cir. 2024) (holding that acquisition of location history constitutes a search, and geofence warrant violated Fourth Amendment). The suppression motions made in this line of cases make both breadth (indiscriminate searches are unconstitutional) and depth (a particular search was unjustifiably intrusive) claims.

       [6]. Carpenter v. United States, 138 S. Ct. 2206, 2221 (2018) (holding that government request for multiple months of location information from cell phone provider constitutes a “search” under the Fourth Amendment).

       [7]. Riley v. California, 573 U.S. 373, 403 (2014) (holding that the search incident to arrest exception does not allow for warrantless search of arrested suspect’s cell phone).

       [8]. United States v. Jones, 565 U.S. 400, 430–31 (2012) (holding that attaching a GPS tracker to suspect’s car constituted a Fourth Amendment search); see also, e.g., Commonwealth v. McCarthy, 142 N.E.3d 1090, 1095 (Mass. 2020) (upholding use of automatic license plate reader to track movements of man suspected of drug distribution).

       [9]. See, e.g., In re Search of Information that is Stored At Premises Controlled by Google, No. 2:22-mj-01325, 2023 WL 2236493, at 1 (S.D. Tex. Feb. 14, 2023) (granting geofence warrant application for cell phone data that would identify phones in vicinity of multiple fraudulent transactions); In re Use of a Cell-Site Simulator to Identify a Cellular Device in a Narcotics Trafficking Case, 623 F. Supp. 3d 888, 897 (N.D. Ill. 2022) (granting warrant to use cell-site simulator); In re Search of: Information Stored at Premises Controlled By Google, As Further Described in Attachment A, No. 20 M 297, 2020 WL 5491763, at 1 (N.D. Ill. 2020) (denying geofence warrant). See generally Jantz, supra note 3, at 10 (citing cases).

     [10]. Maryland v. King, 569 U.S. 435 (2013) (upholding statute categorically requiring DNA samples from felony arrestees without any showing of suspicion); Leaders of a Beautiful Struggle v. Baltimore, 2 F.4th 330, 348 (4th Cir. 2021) (striking down aerial surveillance program as unreasonable under the Fourth Amendment).

     [11]. See, e.g., Andrew Guthrie Ferguson, The Rise of Big Data Policing: Surveillance, Race, and the Future of Law Enforcement 8–14 (2017) (describing how big data tools shape law enforcement and amplify threats to civil right and civil liberties); Erin Murphy, Databases, Doctrine, and Constitutional Criminal Procedure, 37 Fordham Urb. L.J. 803, 834 (2010) (tracing doctrine around databasing and urging rethinking of Fourth Amendment in response to big data threat to liberty).

     [12]. See, e.g., Jeffrey Bellin, Fourth Amendment Textualism, 117 Mich. L. Rev. 233, 271 (2019) (contemplating rule that would treat some data as personal to the consumer and thus remove third-party concerns); David Alan Sklansky, Too Much Information: How Not to Think About Privacy and the Fourth Amendment, 102 Calif. L. Rev. 1069, 1120 (2014) (expressing concern about failing to safeguard space from police surveillance).

     [13]. See, e.g., Slobogin, Suspectless Searches, supra note 2, at 954 (taxonomizing big data searches to find suspects); Barry Friedman & Danielle Keats Citron, Indiscriminate Data Surveillance, 110 Va. L. Rev. 1351, 1355 (2024) (criticizing mass data collection for police surveillance); Barry Friedman, Lawless Surveillance, 97 N.Y.U. L. Rev. 1144, 1144 (2022) (urging more robust oversight of contemporary surveillance practices); Emily Berman, When Database Queries Are Fourth Amendment Searches, 102 Minn. L. Rev. 577, 612 (2017) (arguing that database queries are searches under the Fourth Amendment).

     [14]. See, e.g., Jane R. Bambauer, Filtered Dragnets and the Anti-Authoritarian Fourth Amendment, 97 S. Cal. L. Rev. 571, 579 (2024) (endorsing constitutionality of big data searches precise enough to identify a single suspect); Orin S. Kerr, An Equilibrium-Adjustment Theory of the Fourth Amendment, 125 Harv. L. Rev. 476, 488–89 (2011) (arguing that doctrine equalizes to threats); Orin S. Kerr, The Mosaic Theory of the Fourth Amendment, 111 Mich. L. Rev. 311, 346–47 (2012) (rejecting notion that the Fourth Amendment should consider the aggregate impact of surveillance practices); D.H. Kaye & Michael E. Smith, DNA Identification Databases: Legality, Legitimacy, and the Case for Population-Wide Coverage, 2003 Wis. L. Rev. 413, 459 (2003) (arguing that population-wide compulsory DNA databases best promote both public safety and privacy values).

     [15]. Ferguson, supra note 11, at 2.

     [16]. Slobogin, Suspectless Searches, supra note 2, at 954. Gregory Brazeal refers to a similar idea as “digital mass surveillance.” Brazeal, supra note 2, at 1003.

     [17]. Mary D. Fan, Big Data Searches and the Future of Criminal Procedure, 102 Tex. L. Rev. 877, 880 (2024) (describing the use of “corporately held big data troves to crack cold cases via geofence warrants and keyword warrants”).

     [18]. Friedman & Citron, Indiscriminate Data Surveillance, supra note 2, at 1355. In earlier work, Barry Friedman described the “almost entirely unregulated” and “growing network” of surveillance. Friedman, Lawless Surveillance, supra note 2, at 1143.

     [19]. Slobogin, Virtual Searches, supra note 2, at 11. Suspect-driven searches are aimed at producing evidence about a previously identified suspect. Id. at 78. Profile-driven searches use algorithms to comb data for predictive policing insights. Id. Program-driven searches are a form of indiscriminate mass-surveillance in which law enforcement, acting without any suspicion, “‘fuse’ for police use financial, rental, utility, vehicular, and communication data from federal, state, and local public databases, law enforcement files, and private company records.” Id. at 149–50. Lastly, “volunteer-driven” searches involve actions by private parties to support law enforcement investigations using private technologies. Id. at 12.

     [20]. Id. at 128.As in a closed universe search, a precipitating event sparks the ensuing investigation. But Slobogin’s term encompasses a wider array of search activity, including, for instance, a broad geofence to identify the perpetrator of a homicide, the use of drone footage to track the path of known suspects, the use of facial recognition software to locate a suspect or identify a person in surveillance video, and a search of a DNA database for a match between a crime scene sample and a known person. Id. at 127–29.

     [21]. Id. at 129.

     [22]. That is, indiscriminate or “event-driven” searches may generate closed universe searches but do not necessarily do so. For instance, suppose that facial recognition software is used to generate a list of possible suspects based on video surveillance of the perpetrator of three separate robberies. Such a search is innately indiscriminate—it searches through the entire trove of facial data to isolate candidates. But it may turn up no leads, or a single clear suspect. Neither of those situations gives rise to a closed universe search. A closed universe search issue arises only if the search reveals a discrete pool of suspects, as distinct from searches of either a single suspect or the indiscriminate masses.

     [23]. Bambauer, supra note 14, at 571.

     [24]. Bambauer gives examples of precursors of such searches, such as police dogs sniffing out contraband or checks of fingerprints against a known file. Id. at 579.

     [25]. Id. at 580 (emphasis added). Bambauer describes filtered dragnets as a “subset” of suspectless searches. Id.

     [26]. Id. at 581 (“To meet the definition of a filtered dragnet for my purposes, police will remain ignorant to details and identities until there is a high probability that the information identifies and pertains to the perpetrators and no one else.” (emphasis added)).

     [27]. Seeinfra Part II.C.2.

     [28]. Again, neither of these variations are included because as explained supra, in both instances conventional Fourth Amendment doctrine and the limits of the physical world (including policing resources, capacity, etc.) are adequate safeguards of the individuals’ interests.

     [29]. See, e.g., People v. Seymour, 536 P.3d 1260, 1267 (Colo. 2023) (finding that defendant had a reasonable expectation of privacy in his search history, and suggesting that warrant may have been constitutional but deciding case by applying good faith exception).

     [30]. See, e.g., State v. Vannieuwenhoven, 8 N.W.3d 63, 67 (Wis. Ct. App. 2024) (genetic genealogist narrowed suspect list to eight members of the same family and covertly collected DNA from several before using a ruse to obtain a voluntary sample from the defendant); State v. Burns, 988 N.W.2d 352, 356 (Iowa 2023) (narrowing list to three brothers in similar case).

     [31]. See, e.g., Bambauer, supra note 14, at 582–87 (giving examples of technology uses that would or would not constitute a “filtered dragnet”).

     [32]. See generally Richard A. Leo, Police Interrogation and American Justice (2008) (recounting history of interrogation tactics and exposing risk of false confessions).

     [33]. See generally Elizabeth F. Loftus, Eyewitness Testimony (1996) (examining the risks of eyewitness evidence).

     [34]. See generally Alexandra Natapoff, Snitching: Criminal Informants and the Erosion of American Justice (2009) (criticizing the role of informants in the U.S. criminal system).

     [35]. See generally Erin Murphy, Forensic Evidence, in 3 Reforming Criminal Justice 171 (Erik Luna ed., 2017) (surveying critiques of forensic science and promising proposals for reform).

     [36]. Maryland v. King, 569 U.S. 435, 446 (2013) (“It can be agreed that using a buccal swab on the inner tissues of a person’s cheek in order to obtain DNA samples is a search.”).

     [37]. Carpenter v. United States, 138 S. Ct. 2206, 2221–23 (2018).

     [38]. Jantz, supra note 3, at 10–11 (citing cases).

     [39]. See, e.g., Erin E. Murphy, Inside the Cell: The Dark Side of Forensic DNA 168–79 (2015) (describing “sneak” sampling); Elizabeth E. Joh, Reclaiming “Abandoned” DNA: The Fourth Amendment and Genetic Privacy, 100 Nw. U. L. Rev. 857, 860–68 (2006) (describing and decrying lack of constitutional scrutiny for collecting abandoned DNA). But see United States v. Davis, 690 F.3d 226, 246 (4th Cir. 2012) (holding that patient retained reasonable expectation of privacy in DNA from clothes cut off him at hospital after incident).

     [40]. See, e.g., Ferguson, Persistent Surveillance, supra note 2, at 13–14 (discussing United States v. Tuggle, 4 F.4th 505 (7th Cir. 2021), which upheld constitutionality of long pole cameras).

     [41]. Id. at 9–10.

     [42]. See generally Matthew Tokson, Government Purchases of Private Data, 59 Wake Forest L. Rev. 269 (2024) (identifying risk that government purchase of private data allows it to circumvent Fourth Amendment protections).

     [43]. See, e.g., Amato v. Dist. Att’y for the Cape & Islands Dist., 80 Mass. App. Ct. 230, 241 (2022) (reinstating plaintiff’s claim).

     [44]. U.S. Const. amend. IV; see also Nat’l Treasury Emps. Union v. Von Raab, 489 U.S. 656, 665 (1989) (“[A] search must be supported, as a general matter, by a warrant issued upon probable cause.”).

     [45]. 462 U.S. 213, 238 (1983) (emphasis added).

     [46]. 568 U.S. 237, 243 (2013) (citation modified). Probable cause means “less than evidence which would justify condemnation.” United States v. Ventresca, 380 U.S. 102, 107 (1965) (quoting Locke v. United States, 11 U.S. 339, 348 (1813)).

     [47]. Law enforcement may not deliberately materially misrepresent the facts in a warrant affidavit. Franks v. Delaware, 438 U.S. 154, 171–72 (1978). But lower courts have held that this does not include a duty to disclose material exculpatory information. See, e.g., Mays v. City of Dayton, 134 F.3d 809, 816 (6th Cir. 1998) (“[A] duty to disclose potentially exculpatory information appropriate in the setting of a trial to protect the due process rights of the accused is less compelling in the context of an application for a warrant.”).

     [48]. 354 U.S. 449 (1957). I am grateful to Stephen Schulhofer for calling this case to my attention.

     [49]. Id. at 450–51.

     [50]. Id. at 456. The technical grounds for reversal was a violation of the then-existing rule of criminal procedure that required prompt arraignment.

     [51]. Tracey Maclin, The Pringle Case’s New Notion of Probable Cause: An Assault on Di Re and the Fourth Amendment, 2003 Cato S. Ct. Rev. 395, 437 (2003).

     [52]. 444 U.S. 85 (1979); see also United States v. Di Re, 332 U.S. 581, 593–94 (1948) (finding no probable cause based on “mere propinquity” to others suspected of criminal activity).

     [53]. Ybarra,444 U.S. at 87–88.

     [54]. Id. at 87.

     [55]. Id. at 88.

     [56]. Id. at 86.

     [57]. 540 U.S. 366, 372 (2003).

     [58]. The court leaned heavily on the confined nature of a car and the possibility that all three were involved in a joint enterprise. Id. at 373–74. But see Maclin, supra note 51, at 417 (“The inference that Pringle and his companions were involved in a drug trafficking conspiracy cannot be reconciled with Di Re’s holding.”).

     [59]. Amanda Peters, Mass Arrests & the Particularized Probable Cause Requirement, 60 B.C. L. Rev. 217, 239 (2019) (internal quotation marks omitted); see also Maclin, supra note 51, at 436–38 (“Although it is not surprising that the Rehnquist Court would perceive the probable cause standard as a tool to facilitate, rather than hinder, police apprehension of multiple persons for purposes of interrogation, Pringle illustrates just how far the Rehnquist Court has separated itself from its predecessors [like the Court in Mallory] on this point.”).

     [60]. Kaley v. United States, 571 U.S. 320, 338 (2014).

     [61]. The scholarship attempting to assign a quantitative value to probable cause, normative or empirical, is too voluminous to count here. See, e.g., Richard Seltzer, Russell F. Canan, Molly Cannon & Heidi Hansberry, Legal Standards by the Numbers: Quantifying Burdens of Proof or a Search for Fool’s Gold?, 100 Judicature 57, 59 (2016) (canvassing literature). A 2016 study of 124 judges found 49.7% to be the mean response when asked to quantify probable cause, with three-quarters of judges placing the range between 21% and 60% (and 51–60% garnering a third of judges’ responses). Id. at 62.

     [62]. Cf. United States v. Rhine, 652 F. Supp. 3d 38, 87 (D.D.C. 2023) (finding that a “32% chance that any given data point is inaccurate” still sufficient for probable cause, noting that probable cause is “less than a preponderance of the evidence” (internal quotation marks and citation omitted)).

     [63]. In re Warrant Application for Use of a Canvassing Cell-Site Simulator, 654 F. Supp. 3d 694, 718 (N.D. Ill. 2023) (emphasis in original).

     [64]. Florida v. Harris, 568 U.S. 237, 243 (2013) (emphasis added).

     [65]. Messerschmidt v. Millender, 565 U.S. 535, 551 (2012) (emphasis added) (internal quotation marks omitted) (quoting Warden v. Hayden, 387 U.S. 294, 307 (1967)).

     [66]. 799 F.3d 1295, 1297–98 (9th Cir. 2015), cert. denied, 577 U.S. 1194 (2016).

     [67]. Id. at 1297.

     [68]. Id. at 1297–98.

     [69]. Id. at 1299.

     [70]. The basis of the order was a statute that required only “reasonable cause ‘for a belief that a felony has been committed’ to support a detention order . . . and specifie[d] no particular quantum of suspicion that the evidence sought ‘may contribute to the identification of the individual who committed such offense.’” Id. at 1300. But judicial precedent had described such orders as “warrants” and had imposed a probable cause requirement on searches that entailed bodily invasions, so the Ninth Circuit analyzed the constitutionality of the orders under that standard. Id.

     [71]. Id. at 1301.

     [72]. Id.

     [73]. Id.

     [74]. Id. (emphasis added) (internal quotations omitted).

     [75]. Id.

     [76]. Id. (“To be sure, the orders here did not seek to obtain evidence that the plaintiffs committed a crime. But contrary to plaintiffs’ intimations, [t]he critical element in a reasonable search is not that . . . the person, to be searched is suspected of crime.” (internal quotations omitted)).

     [77]. Id. (internal quotations omitted).

     [78]. 565 U.S. 535, 552–53 (2012). A frequently cited case in support of this position, and one cited by the Brewer court, is the Supreme Court’s decision in Zurcher v. Stanford Daily, 436 U.S. 547, 565 (1978) (finding that warrants require only “probable cause, specificity with respect to the place to be searched and the things to be seized, and overall reasonableness”). The Court in Zurcher expressly rejected the idea that searches for evidence held by nonsuspects deserve greater protection under the Fourth Amendment:

Because the State’s interest in enforcing the criminal law and recovering evidence is the same whether the third party is culpable or not the premise of the District Court’s holding appears to be that state entitlement to a search warrant depends on the culpability of the owner or possessor of the place to be searched and on the State’s right to arrest him. The cases are to the contrary . . . . The critical element in a reasonable search is not that the owner of the property is suspected of crime but that there is reasonable cause to believe that the specific “things” to be searched for and seized are located on the property to which entry is sought. Id. at 555–56.

     [79]. Messerschmidt v. Millender, 565 U.S. 535, 551 (2012) (internal quotation marks omitted) (quoting Warden v. Hayden, 387 U.S. 294, 307 (1967)). The Bill v. Brewer formulation, “would assist in finding or convicting a particular defendant for a particular offense,” has been cited by a number of district courts in California. See, e.g., United States v. Martin, No. 1:21-cr-00228 NODJ BAM, 2024 U.S. Dist. LEXIS 20680, at 6 (E.D. Cal. Feb. 6, 2024); United States v. Flucas, No. 2:17-CR-0209-KJM, 2018 U.S. Dist. LEXIS 92308, at 2 (E.D. Cal. June 1, 2018).

     [80]. In re G.B., 139 A.3d 885, 888 (D.C. 2016). The court rejected a statutory argument based on the terms of the applicable procedural rule, finding that “evidence of a crime’s commission” requires only “cause to believe that the evidence sought will aid in a particular apprehension of conviction.” Id. at 892–93 (citing Bill v. Brewer and noting that “within its scope [is] ‘evidence that might lead to other evidence’”). The court also rejected a constitutional argument that G.B.’s status as a third party made the warrant invalid, and instead looked to “reasonableness” in place of “probable cause (or even suspicion) that G.B. is guilty of a crime.” Id. at 896–98 (citing Zurcher, 436 U.S. at 556 and other cases).

     [81]. Bill v. Brewer, 799 F.3d 1295, 1302 (9th Cir. 2015) (quoting United States v. Torres, 751 F.2d 875, 883 (7th Cir. 1984)).

     [82]. See, e.g., Slobogin, Suspectless Searches, supra note 2, at 958 (citing Illinois v. Lidster, 540 U.S. 419 (2004), as a model for handling suspectless searches because the brief seizure of motorists to find witnesses to hit and run accident “was constructed in a way that furthered the chances of finding that person while minimizing hassle rates, and involved a relatively minor seizure . . . .”).

     [83]. 470 U.S. 753, 763–66 (1985).

     [84]. Brewer, 799 F.3d at 1302–03.

     [85]. Warden v. Hayden, 387 U.S. 294, 301 (1967) (observing that the Fourth Amendment was specifically adopted to prevent the issuance of the indiscriminate, general warrants and writs of assistance in the Founding era).

     [86]. See, e.g., Jantz, supra note 3, at 19–30 (summarizing standards to help narrow breadth of search).

     [87]. See, e.g., In re G.B., 139 A.2d 885, 888, 897 (D.C. 2016) (rejecting the argument that a warrant can only issue for DNA of the person suspected of the offense); Commonwealth v. Kostka, 31 N.E.3d 1116, 1119 (Mass. 2015) (affirming that the government may compel a DNA sample from a third party when the sample “will probably provide evidence relevant to the question of the defendant’s guilt”).

     [88]. See, e.g., State v. Register, 419 S.E.2d 771, 773 (S.C. 1992) (affirming statutory test for requiring suspect’s girlfriend to give DNA sample to help exclude her as source of stains found in suspect’s car as requiring court to “balance the seriousness of the crime, the importance of the evidence to the investigation, and the unavailability of alternative, less intrusive means of obtaining the evidence, on one hand, against concern for the potential witness’ constitutional right to be free from bodily intrusion on the other.”); In re Morgenthau, 457 A.2d 472, 473–76 (N.J. Super. Ct. App. Div. 1983) (compelling wife and stepdaughter to give samples, as it “is constitutionally permissible to issue search warrants in connection with a prosecutor’s application to obtain material evidence from third parties who are not suspects in a criminal prosecution”).

     [89]. See, e.g., United States v. Chatrie, 590 F. Supp. 3d 901, 914–16 (E.D. Va. 2022) (describing three-step process for geofence search and finding warrant unconstitutional but allowing evidence under good faith exception), aff’d, 107 F.4th 319, 322, 326 (4th Cir. 2024) (finding geofence search not a “search” for Fourth Amendment purposes and thus constitutional, citing third party doctrine), aff’d on reh’g en banc, 136 F.4th 100, 101 (4th Cir. 2025) (per curiam opinion affirming district court), cert. granted, Chatrie v. United States, No. 25-112, 2026 WL 120676 (U.S. Jan. 16, 2026).

     [90]. Chatrie v. United States, No. 25-112, 2026 WL 120676 (U.S. Jan. 16, 2026) (granting review of question whether geofence warrant violated the Fourth Amendment). A geofence is a request to a service provider, typically Google, to provide historical location information for devices that connected to its network at a specific time period and geographical place. See Jantz, supra note 3, at 12–13. As explained in a brief amicus curiae filed by Google in Chatrie, geofence warrants are warrants in “reverse,” because rather than warrants that “target an identified suspect based on probable cause,” a geofence warrant “draws a geographic boundary, picks a timeframe, and demands Location History for every Google user whose device happened to report and store a location within that area.” Brief Amicus Curiae of Google LLC in support of neither party, Chatrie v. United States (No. 25-112), 2026 WL 624354, at 4. In part due to privacy concerns, Google changed its practice in 2023 and stored location data on the native device, rather than through its cloud servers. Id. at 10-11. But the concept of a “reverse warrant” still aptly describes any search that trawls data for suspects, whether for keyboard strokes, location history, facial image, or other information. See, e.g., id. at 5–7.

     [91]. Jantz, supra note 3, at 27–28 (summarizing two-step process and discussing different cases that have discussed such a process). The relationship of the police, judiciary, and the private company in crafting these doctrines is itself fascinating. In People v. Seymour, for instance, police sought a “reverse-keyword warrant” for persons who had searched a particular address associated with an arson. 536 P.3d 1260, 1267 (Colo. 2023). The initial warrant was refused by Google (and thus withdrawn by the state) because it contravened the three-step process; a revised warrant was also refused because it was deemed too invasive of privacy. Id. at 1267–69. Finally, a third warrant was requested and granted that disclosed sixty-one searches by eight accounts on the pertinent day. Id. at 1276. A subsequent warrant then revealed the deanonymized information. Id. at 1272.

     [92]. The process is basically the same even though courts tend to label it with differing numbers: (1) probe the data holder for matching information within certain (ideally narrow) parameters; (2) receive back anonymized matching data, which law enforcement then investigates with the goal of narrowing further; and (3) seek to “unmask” users of interest. Courts simply differ on whether they label step three as a unique step. CompareIn re Search of Information that is Stored at the Premises Controlled by Google LLC,579 F. Supp. 3d 62, 87 (D.D.C. 2021) (describing process as two steps), with United States v. Davis, 109 F.4th 1320, 1328 (11th Cir. 2024) (describing three-step process), and United States v. Smith, 110 F.4th 817, 824–25 (5th Cir. 2024) (outlining the three-step process used).

     [93]. See, e.g., In re Search of Information that is Stored At Premises Controlled by Google, No. 2:22-mj-01325, 2023 WL 2236493, at *6 (S.D. Tex. Feb. 14, 2023) (discussing limits at each step); In re Search of Information that is Stored at the Premises Controlled by Google LLC, 579 F. Supp. 3d at 89–90 (“Even if third-party devices appear within the government’s requested geofence, those persons’ location information will be anonymized at step one. . . . [A]ssuming law enforcement seeks to deanonymize any devices at all, additional justification for such request can be required.”); see also In re Warrant Application for Use of a Canvassing Cell-Site Simulator, 654 F. Supp. 3d 694, 716, 721 (N.D. Ill. 2023) (denying warrant as overbroad and standardless, but noting that a multistep process might satisfy particularity and overbreadth concerns).

     [94]. See, e.g., In re Application for Tower Dump Data for a Sex Trafficking Investigation, No. 23 M 87, 2023 WL 1779775, at 3–5 (N.D. Ill. Feb. 6, 2023) (issuing warrant after imposing limitations); In re Application of the United States of America for an Order Relating to Telephones Used by Suppressed, No. 23 M 87, 2015 WL 6871289, at 4 (N.D. Ill. Nov. 9, 2015) (explaining requirements for cell-site simulator warrant); Commonwealth v. Perry, 184 N.E.3d 745, 770 (Mass. 2022) (dictating that warrant requests “must include protocols for the prompt and permanent disposal of any and all data that does not fit within the object of the search following the conclusion of the prosecution.”). California has a statutory provision addressing the sealing and destruction of released electronic data deemed not pertinent to the investigation. Cal. Penal Code § 1546.1(d)(2), (e)(2).

     [95]. See In re Search Warrant Application for Geofence Location Data Stored at Google Concerning an Arson Investigation, 497 F. Supp. 3d 345, 362 (N.D. Ill. 2020) (finding that police have full discretion to request subscriber info for some or all of the location data without a second warrant or any judicial oversight, finding the facts “established probable cause to seize all location and subscriber data within the geofence locations identified. Whether [the government] chooses to obtain all that information, or partial information, is of no matter to the Court’s consideration of the constitutionality of the warrant . . . .”); In re Search of Information Associated with Cellular Telephone Towers Providing Service to [Redacted] that is Stored at Premises Controlled by Verizon Wireless, 616 F. Supp. 3d 1, 10 (D.D.C. 2022) (finding sufficient particularity to disclose subscriber data pursuant to tower dump warrant request).

     [96]. See, e.g., State v. Contreras-Sanchez, 5 N.W.3d 151, 168 (Minn. Ct. App. 2024) (noting that warrant granted discretion to officers to determine relevance). It should be noted that “relevance” is the standard imposed by the pen registers statute, 18 U.S.C. § 3122(b)(2), and “specific and articulable facts showing that there are reasonable grounds to believe that the contents . . . are relevant and material to an ongoing criminal investigation” is the standard for noncontent information in electronic communications, 18 U.S.C. § 2703(d).

     [97]. Contreras-Sanchez, 5 N.W.3d at 171 (“We need not determine whether the warrant’s authorization of the seizure of de-anonymized step-three data met constitutional requirements because that portion of the warrant is severable.”).

     [98]. See, e.g., In re Search of Information that is Stored At Premises Controlled by Google, No. 2:22-mj-01325, 2023 WL 2236493, at *6 (S.D. Tex. Feb. 14, 2023) (“The applicant would then, if appropriate, return to the Court with one or more subsequent warrant requests (‘Step Two’), asking the Court to order Google to unmask devices identified in Step One by providing Google account identifier and basic subscriber information. Each ‘Step Two’ warrant request would itself need to be supported by probable cause in order to obtain court approval.”); United States v. Rhine, 652 F. Supp. 3d 38, 85 (D.D.C. 2023) (upholding geofence warrant sought in connection with January 6 insurrection, even though it returned thousands of phones of interest, noting that the unique features of the location—a tightly controlled government building—and the large number of persons involved in the insurrection led to a “scope of probable cause [that] was uncommonly large”).

              People v. Seymour wrestled with just this question of both the court’s role and the applicable standard for the deanonymization phase. First the court observed that warrants typically require “individualized probable cause,” not just “merely in relation to the property or premises” but “in relation to each individual’s constitutionally protected interests.” Seymour, 536 P.3d 1260, 1277–78 (Colo. 2023) (“When evaluating probable cause, however, the quantum of evidence is not the only variable at issue. This court has held that the Fourth Amendment requires ‘any invasion of a person’s Fourth Amendment interests [to] be justified . . . by “specific and articulable facts” directed to the person whose interests are to be invaded.’ Such ‘individualized probable cause’ is necessary for each intrusion of an individual’s constitutionally protected privacy interest, whether that interest lies in the individual’s person or in records held by a third party.” (internal citations omitted)). But the court took pains to describe the reverse keyboard search as a “far cry” from a physical search and ultimately determined that “we need not resolve today whether a search of such data requires probable cause individualized to a single Google account holder.” Id. at 1278 (noting application of good faith exception).

     [99]. United States v. James, 3 F.4th 1102, 1105 (8th Cir. 2021) (upholding multiple warrants in their opinion by finding a “fair probability” that the tower dump would reveal a suspect and that the geographical and temporal constraints were sufficient, but not addressing process required for the final deanonymization step).

   [100]. In United States v. Rhine, the court concluded that although “there is still a roughly 32% chance that any given data point is inaccurate, error radius notwithstanding . . . there is still a ‘substantial basis’ for the magistrate judge to have identified a ‘fair probability’ that all of these 1,498 devices were linked to suspects or witnesses,” and upheld the warrant. 652 F. Supp. 3d 38, 87 (D.D.C. 2023).

   [101]. In re Search of Information Associated with Cellular Telephone Towers Providing Service to [Redacted] that is Stored at Premises Controlled by Verizon Wireless, 616 F. Supp. 3d 1, 12 (D.D.C. 2022).

   [102]. United States v. Smith, No. 23-6032, 2023 WL 1930747, at *4 (N.D. Miss. 2023), aff’d, 110 F.4th 817, 820 (5th Cir. 2024).

   [103]. Id. at *2.

   [104]. United States v. Smith, 110 F.4th 817, 827–28 (5th Cir. 2024) (noting that warrant request promised “further legal process” before proceeding to Steps 2 and 3, as did issued warrant, but that “prior to submitting Step 2, neither [police investigator] applied for another warrant” and “[a]gain, without seeking any new warrants, [the police investigators] sent off their Step 3 request . . . .”).

   [105]. Id. at 828.

   [106]. Id. at 840 (affirming opinion, notwithstanding holding geofence warrants unconstitutional, by applying good faith exception).

   [107]. 913 S.E.2d 700, 704 (Ga. 2025).

   [108]. Id. at 703.

   [109]. See, e.g., In re Search of Information that is Stored at the Premises Controlled by Google LLC, 579 F. Supp. 3d 62, 85 (D.D.C. 2021) (“It may be the case that the requested geofence, when considering its margin of error, will capture the location information for other customers inside the [Redacted] center or motorists merely driving by the [Redacted] center on the abutting road or an employee in the adjoining business during the requested time segments. The possibility of capturing those persons’ location information is not fatal to the warrant’s constitutionality. The particularity inquiry turns on what [is] realistic or possible in this investigation.” (alterations in original) (internal quotation marks omitted)); see also Jantz, supra note 3, at 24–25 (summarizing case law in which law enforcement may only obtain deanonymized information when there is multiple or cross-referenced “hits,” where the target device appears in multiple locations of interest).

   [110]. In re Warrant Application for Use of a Canvassing Cell-Site Simulator, 654 F. Supp. 3d 694, 709 (N.D. Ill. 2023).

   [111]. Id. at 717 (“The government has not explained how it can distinguish between the suspect and others who live in the surrounding residences, so the proposed ‘multiple times at a common location’ limitation on further investigation does not seem to be a real limitation at all.”).

   [112]. Id. at 718 (discussing the need for different areas where suspect is seen or reasonably believed to be “sufficiently far apart” so that the CCSS coverage areas are less likely to overlap).

   [113]. In re Search of Information that is Stored at the Premises Controlled by Google LLC, 579 F. Supp. 3d at 82.

   [114]. See, e.g., United States v. James, No. 18-cr-216, 2018 WL 6566000, at *5 (D. Minn. 2018) (denying suppression on grounds that “tower dump” warrant asking whether any cell number was present at each of six different robberies was “carefully tailored”).

   [115]. Fan, supra note 17, at 927 (emphasis added).

   [116]. Id. at 931.

   [117]. See Erin Murphy, License, Registration, Cheek Swab: DNA Testing and the Divided Court, 127 Harv. L. Rev. 161, 183 (2012) (“Reconciling the two clauses of the Fourth Amendment—commonly referred to as the Warrant Clause and the Reasonableness Clause—has long vexed scholars and practitioners alike.”). Compare Telford Taylor, Two Studies in Constitutional Interpretation 19–20 (1969) (citing historical examples supporting reasonableness), and Akhil Reed Amar, Fourth Amendment First Principles, 107 Harv. L. Rev. 757, 801 (1994) (urging separate reading of clauses), with Anthony G. Amsterdam, Perspectives on the Fourth Amendment, 58 Minn. L. Rev. 349, 358 (1974) (defending warrants as default). See generally Barry Friedman, Unwarranted: Policing Without Permission (2017) (criticizing both courts and legislatures for abdicating responsibility to safeguard Fourth Amendment rights and encouraging popular pressure to enact regulatory restraints on policing).

   [118]. See, e.g., United States v. Watson, 423 U.S. 411, 423 (1976) (authorizing warrantless felony arrests in public with probable cause).

   [119]. See, e.g., Camara v. Mun. Ct., 387 U.S. 523, 538 (1967) (requiring a warrant, but only generalized probable cause, for building inspection). See generally Eve Brensike Primus, Disentangling Administrative Searches, 111 Colum. L. Rev. 254 (2011) (discussing the need for at least “generalized government interests” to conduct administrative searches, including those conducted using dragnets).

   [120]. See, e.g., Terry v. Ohio, 392 U.S. 1, 30 (1968).

   [121]. Compare, e.g., City of Indianapolis v. Edmond, 531 U.S. 32, 37–38 (2000) (striking down drug interdiction checkpoint), with Michigan v. Sitz, 496 U.S. 444, 455 (1990) (holding sobriety checkpoint aimed at public safety constitutional), and Illinois v. Lidster, 540 U.S. 419, 423–24 (2004) (upholding checkpoint aimed at seeking witnesses in hit-and-run accident). See also Primus, supra note 119, at 275–76 (describing special needs test). Of course, what constitutes “ordinary criminal wrongdoing” can be a malleable concept. See, e.g., Barry Friedman, The Constitutionality of Indiscriminate Data Surveillance, 174 U. Pa. L. Rev. 293, 322–24 (2026) (criticizing Edmond and Sitz distinctions as indefensible). In Maryland v. King, the Supreme Court upheld the systematic DNA sampling of arrestees as constitutional in part because arrested persons have diminished privacy, but also in part because it described the primary purpose of testing as identification verification, rather than crime solving. 569 U.S. 435, 451–53, 462 (2013).

   [122]. Terry, 392 U.S. 1, 21 (1968) (citing Camara, 387 U.S. 523).

   [123]. Maryland v. King, 569 U.S. 435, 446 (2013) (“It can be agreed that using a buccal swab on the inner tissues of a person’s cheek in order to obtain DNA samples is a search.”).

   [124]. Carpenter v. United States, 138 S. Ct. 2206, 2219 (2018).

   [125]. Primus, supra note 119, at 254.

   [126]. Id. at 257–58 (internal quotation marks and citations omitted).

   [127]. Id. at 259–60.

   [128]. Davis v. Mississippi, 394 U.S. 721, 722–23 (1969).

   [129]. Id. at 724–25.

   [130]. Id. at 728.

   [131]. Id. at 727 (citing Camara v. Mun. Ct., 387 U.S. 523 (1967)). That choice is not altogether surprising. Terry requires reasonable, articulable suspicion of individual wrongdoing, which was unquestionably lacking in Davis’s case. In contrast, the special needs cases authorize suspicionless, programmatic searches. Primus, supra note 119, at 276.

   [132]. Id. (referring to the “unique nature of the fingerprinting process”).

   [133]. Id. (emphasis added).

   [134]. Peter P. Swire & Erin E. Murphy, How to Address Standardless Discretion After Jones 1 (Ohio St. Univ. Moritz Coll. of L., Working Paper No. 177, 2012), http://ssrn.com/abstract=2122941 [https://perma.cc/HD7T-JEUB]; seealso Erin Murphy, Back to the Future: The Curious Case of United States v. Jones, 10 Ohio St. J. Crim. L. 325, 340 (2012) (“Rather than defer entirely to the political branches, then, a better approach is one that uses judicial review to focus legislatures on the promulgation of procedural safeguards against abuse, and the assurance of actual police compliance with announced rules.”). But see Orin S. Kerr, The Effects of Legislation on Fourth Amendment Protection, 115 Mich. L. Rev. 1117, 1139 (2017) (arguing against the influence of legislation on Fourth Amendment doctrine).

   [135]. Swire & Murphy, supra note 134, at 2.

   [136]. Slobogin, Virtual Searches, supra note 2, at 41; Christopher Slobogin, Privacy at Risk: The New Government Surveillance and the Fourth Amendment 28–30 (2007); seealso Jeffrey Bellin, Crime-Severity Distinctions and the Fourth Amendment: Reassessing Reasonableness in a Changing World, 97 Iowa L. Rev. 1, 26–37 (2011) (arguing for consideration of crime-severity in Fourth Amendment questions).

   [137]. Friedman, supra note 121, at 318–19.

   [138]. See, e.g., Orin S. Kerr, Executing Warrants for Digital Evidence: The Case for Use Restrictions on Nonresponsive Data, 48 Tex. L. Rev. 1, 24 (2015) (arguing that the “use of nonresponsive files violates the Fourth Amendment in a two-stage computer search because it renders the ongoing seizure of the nonresponsive files constitutionally unreasonable”).

   [139]. See Carpenter v. United States, 138 S. Ct. 2206, 2213 (2018) (affirming that searches typically require a warrant supported by probable cause).

   [140]. Id. at 2221–23.

   [141]. Id. at 2221.

   [142]. Id. at 2221 (internal quotation marks omitted). But see Slobogin, Virtual Searches, supra note 2, at 37 (“Jones and Carpenter have set the stage for a revolution in search and seizure doctrine both in terms of the Fourth Amendment’s threshold and with respect to its reasonableness inquiry.”).

   [143]. Carpenter, 138 S. Ct. at 2221. The Court’s hedge of “typically” seemed to also leave a crack in the door for a search to be deemed reasonable on grounds less than demanded by a traditional warrant and probable cause.

   [144]. Id.

   [145]. Id. at 316.

   [146]. See, e.g., Camara v. Mun. Court, 387 U.S. 523, 538 (1967) (requiring warrant, but only generalized probable cause, for building inspection). Compare Indianapolis v. Edmond, 531 U.S. 32, 38 (2000) (finding police checkpoint unconstitutional as it was for the purpose of detecting “evidence of ordinary criminal wrongdoing”), with Illinois v. Lidster, 540 U.S. 419, 421 (2004) (finding police checkpoint to find witnesses to hit and run constitutional as it was simply “information seeking”). See generally Primus, supra note 119.

   [147]. Lidster, 540 U.S. at 421 (upholding checkpoint ostensibly to find witnesses to hit and run as only incidentally about finding perpetrator, but primarily “information seeking”); see also Michael Gentithes, Suspicionless Witness Stops: The New Racial Profiling, 55 Harv. C.R.-C.L. L. Rev. 491, 508–14 (2020) (discussing cases). This principle has been criticized as too thinly policing the line between action truly intended as information seeking as opposed to perpetrator seeking, for stigmatizing those stopped, and for heightening the risk of abusive or biased policing. Id. at 518–22.

   [148]. Maryland v. King, 569 U.S. 435, 446 (2013).

   [149]. See Murphy, supra note 117, at 177–79.

   [150]. Compare, e.g., United States v. Smith, 110 F.4th 817, 840 (5th Cir. 2024) (holding geofence warrant an unconstitutional “general warrant”), withIn re Search Warrant Application for Geofence Location Data Stored at Google Concerning Arson Investigation,497 F. Supp. 3d 345, 360 (N.D. Ill. 2020) (upholding warrant to obtain the geofence data based on a showing of probable cause), and In re Search of Information that is Stored at the Premises Controlled by Google LLC, 579 F. Supp. 3d 62, 85 (D.D.C. 2021).

   [151]. Friedman, Lawless Surveillance, supra note 2, at 1169 (citing cases); seealso United States v. R. Enters., Inc., 498 U.S. 292, 301 (1991) (finding relevance standard in grand jury subpoena context means that motion to quash should be denied unless “there is no reasonable possibility that the category of materials the Government seeks will produce information relevant to the general subject of the grand jury’s investigation”). The two major statutory standards, applying to pen registers and electronic communications, both also impose a relevance standard. See 18 U.S.C. § 3122(b)(2); 18 U.S.C. § 2703(d) (requiring “specific and articulable facts showing that there are reasonable grounds to believe that the contents . . . are relevant and material to an ongoing criminal investigation” as the standard for noncontent information in electronic communications); see also Fed. R. Crim. P. 17 (addressing subpoenas).

   [152]. See, e.g., Friedman, Lawless Surveillance, supra note 2, at 1169–71 (summarizing cases).

   [153]. 44 F.3d 1127 (2d Cir. 1995).

   [154]. Id. at 1137.

   [155]. Id. at 1138.

   [156]. Id.

   [157]. Although exemplars are typically not considered “searches” under the Fourth Amendment, United States v. Dionisio, 410 U.S. 1, 6–8 (1973),an analogy might be drawn to subpoenas to multiple people to appear before a grand jury to give voice or handwriting exemplars.

   [158]. Id. at 9–10 (emphasis added).

   [159]. 138 S. Ct. 2206 (2018).

   [160]. See, e.g., Shawn Marie Boyne, Stingray Technology, the Exclusionary Rule, and the Future of Privacy: A Cautionary Tale, 119 W. Va. L. Rev. 915, 918 (2017) (“Currently, law enforcement agencies in at least 23 states, as well as 13 federal agencies, use StingRay devices to track cellphone locations and access cellphone data. Despite their intrusive capabilities, to use these devices, law enforcement officers in most states need only to obtain a low-level court order called a PEN register, also known as a ‘trap and trace,’ to obtain permission for their use.”); Harvey Gee, Last Call for the Third-Party Doctrine in the Digital Age After Carpenter, 26 B.U. J. Sci. & Tech. L. 286, 292 (2020) (“[T]he government’s use of the third-party doctrine to subpoena records from wireless carriers dramatically accelerated as the sales of cell phones and smartphones increased exponentially in the digital age.”).

   [161]. Carpenter, 138 S. Ct. at 2222.

   [162]. Id. Lower courts, in turn, have limited the application of this holding by narrowly construing it to the search specifically decided by Carpenter. See Alan Z. Rozenshtein, Fourth Amendment Reasonableness After Carpenter, 128 Yale L.J.F. 943, 950–51 (2019).

   [163]. Bambauer, supra note 14, at 575.

   [164]. Id. at 602.

   [165]. See, e.g., Friedman, Lawless Surveillance, supra note 2, at 1194 (“The best way to assure non-arbitrariness in the context of digital surveillance is universality.”); id. at 1194–95 (citing Delaware v. Prouse for proposition that stopping motorists at will violates the Constitution, but stopping all motorists, or a random subset of motorists, does not because “it avoids arbitrary (or discriminatory) selection”); Ferguson, supra note 11, at 184 (citing organizing efforts to stop data collection); Bambauer, supra note 14, at 626 (suggesting that “filtered dragnets” “should include everyone possible whose data is accessible” in order to ward against arbitrariness or bias); I. Bennett Capers, Race, Policing, and Technology,95 N.C. L. Rev. 1241, 1283–91 (2017)(advocating for “more policing of everyone” in the form of widespread surveillance as a way to “redistribute” and “equali[ze]” privacy); D.H. Kaye & Michael E. Smith, supra note 14, at 439–440 (“[W]e believe that a properly designed and administered national database might well be the best solution to the coverage question. It would entail but a limited intrusion on individual freedom and privacy while advancing both public safety and racial even-handedness in the criminal justice system and serving as a firewall against far greater intrusions on privacy . . . .”).

   [166]. Of course, there are arguments against reliance on political process. Some scholars question the wisdom of an approach that looks to legislative action to determine the scope of constitutional protections, as well the extent to which such action in fact reflects true political preferences. See, e.g., Kerr, supra note 134, at 1139–40 (arguing against allowing legislation to influence Fourth Amendment doctrine, citing problems of distortion, federalism, and necessity).

   [167]. As one court explained, “there are already documented accounts of innocent bystanders being swept into geofence warrants based solely on their proximity to a crime.” United States v. Smith, 110 F.4th 817, 825 (5th Cir. 2024) (giving examples).

   [168]. See, e.g., Bambauer, supra note 14, at 589–98 (identifying benefits of filtered dragnets as, inter alia, increased accuracy, an increased probability that a case is solved, decreased discretion in suspect selection, and a reduced likelihood of intense physical confrontations or intrusions by police onto a person’s body, possessions, or home); Kaye & Smith, supra note 14, at 458–59 (arguing that a “population-wide DNA database could serve as at least a partial, much-needed antidote for the racial distortions that plague the criminal justice system” and that “settling for a DNA identification database restricted to convicts, or to convicts and arrestees, is sure to aggravate racial polarization in society, undermine the legitimacy of law and law enforcement, and further compromise public safety”); Capers, supra note 165, at 1271–83 (advocating for universal surveillance as a way of addressing underenforcement, overenforcement, and bias).

   [169]. See, e.g., Murphy, supra note 39, at 260–62 (contesting the notion that universal DNA databases are immune from fears of selectivity or abuse). The problem with indiscriminate methods is that they are never truly indiscriminate because they operate against the backdrop of enforcement discretion that empowers law enforcement to determine which crimes to enforce and when. See id. at 256–62; Bambauer, supra note 14, at 614–22 (addressing concerns about ways in which dragnets may still exploit system pathologies to perpetuate inequities).

   [170]. Assume for the purposes of the hypothetical that the robberies were perpetrated by a person acting alone.

   [171]. In other words, when an indiscriminate search isolates a single suspect (or, of course, ordinary investigation pinpoints a single suspect), further intrusion on that person is easily governed by existing doctrine. In contrast, the closed universe suspect pools at issue here lie somewhere between the presumed innocent persons subject to indiscriminate searches and the reasonably suspected person singularly identified by evidence. Each person could be said to be presumptively suspicious—a little bit more down the road of “probable cause” than every other person in the world but always holding the distinct probability (by virtue of and depending on the size of the pool) of being totally innocent.

   [172]. Barry Friedman has described this as the “[c]hallenges of an [a]ll or [n]othing regime.” Friedman, supra note 121, at 52.

   [173]. Monteiro v. Cormier, No. 1:21-cv-00046-MSM-LDA, 2023 WL 6314658 (D.R.I. Sept. 28, 2023).

   [174]. Id. at *3.

   [175]. Id. Because the sample was consumed, it could not be retested. Id. at *3 n.10.

   [176]. Id. at 3. The detective had previously been “reprimanded . . . for inadequate investigations.” Id. at 4.

   [177]. Id. at *12.

   [178]. Id. at *5. For instance, the affidavit alleged Monteiro was itinerant and secretive when in fact he maintained a public Facebook account, had been employed at the same place for fifteen years, and demonstrably did not live in the building where the victim was last seen at the time of her disappearance. Id.

   [179]. Id. at *7. The civil suit settled. See Brian Crandall, Man Once Accused in Pawtucket Cold Case Killing Reaches $1 Million Settlement, NBC 10 News (Rhode Island) (Nov. 26, 2024), https://turnto10.com/news/local/man-once-accused-in-pawtucket-cold-case-killing-reaches-1-million-settlement-police-city-sue-lawsuit-murder-dropped-dna-evidence-november-26-2024 [https://perma.cc/RY9G-U9LC]. A similar case, involving genetic genealogy, arose in Washington. See Mike Carter, WA Man Released as Cold-Case Murder Suspect Sues Detective, Seattle Times (Mar. 4, 2024), https://www.seattletimes.com/seattle-news/law-justice/wa-man-released-as-cold-case-murder-suspect-sues-detective/ [https://perma.cc/65U9-WVHF].

   [180]. See, e.g., Keith A. Findley & Michael S. Scott, The Multiple Dimensions of Tunnel Vision in Criminal Cases, 2006 Wis. L. Rev. 291 (explaining psychological phenomenon of tunnel vision as a cause of wrongful convictions).

   [181]. Dianne L. Martin, Lessons About Justice from the “Laboratory” of Wrongful Convictions: Tunnel Vision, the Construction of Guilt and Informer Evidence, 70 UMKC L. Rev. 847, 848 (2002).

   [182]. D. Kim Rossmo & Joycelyn M. Pollock, Confirmation Bias and Other Systemic Causes of Wrongful Convictions: A Sentinel Events Perspective, 11 Ne. U. L. Rev. 790, 814 (2019).

   [183]. See, e.g., id. at 801–02 (reporting results of study that found confirmation bias and tunnel vision to be the two most frequent causes of wrongful conviction).

   [184]. See, e.g., Ferguson, supra note 11, at 58–60, 125–31 (cataloging problems with databases and algorithmic technologies); Murphy, supra note 39, at 18–84 (discussing fallibility of DNA testing); Vladan M. Jovanovic & Brian T. Cummings, Analysis of Mobile Phone Geolocation Methods Used in US Courts, 11 IEEE Access 28037, 28050 (2022) (comparing different methods of geolocation, calling some “useful and reliable” and others “inherently unreliable”).

   [185]. See, e.g., Murphy, supra note 39, at 175–78 (recounting story of DNA dragnet in which law enforcement threatened to broadcast the name of a man if he refused to give a DNA sample, and then did release the name widely as the primary suspect in a serial rape-murder that he did not in fact commit); id. at 179 (giving other examples of persons who lost jobs or were wrongfully publicly accused because of dragnet DNA searches).

   [186]. See, e.g., id. at 173 (describing political blowback after parents learned law enforcement was using DNA from newborn blood spot program); Natalie Ram, America’s Hidden National DNA Database, 100 Tex. L. Rev. 1253, 1273–76 (2022) (describing law enforcement access to newborn blood sampling program).

   [187]. See generally Friedman, Lawless Surveillance, supra note 2, at 1187–88 (discussing the lack of transparency in the deployment of new surveillance methods).

   [188]. Compare, e.g., United States v. Chatrie, 107 F.4th 319, 339 (4th Cir. 2024), aff’d on reh’g en banc, 136 F.4th 100 (4th Cir. 2025) (per curiam) (holding geofencing searches not a “search” and thus constitutional), cert. granted, Chatrie v. United States, No. 25-112, 2026 WL 120676 (U.S. Jan. 16, 2026), with United States v. Smith, 110 F.4th 817, 840 (5th Cir. 2024) (holding geofencing searches unconstitutional). See generallysupra Part II.

   [189]. Cf. Principles of the Law, Policing, General Principles of Searches, Seizures, and Information Gathering § 2.01 (A.L.I. 2017) (observing that “[a]lthough most policing activities are relatively easy to categorize as either suspicion-based or suspicionless, some arguably could fall on either side of the line,” but advocating in either case for “legislative authorization or an agency policy” before engaging in certain technological searches and propounding guidance); Kerr, supra note 134, at 1159 (“The possibility of statutory privacy offers several substantial advantages over the Fourth Amendment alone. Legislatures are not bound by text, history, or precedent.”).

   [190]. Steven M. Bellovin, Renée M. Hutchins, Tony Jebara & Sebastian Zimmeck, When Enough is Enough: Location Tracking, Mosaic Theory, and Machine Learning, 8 N.Y.U. J.L. & Liberty 556, 560 (2014); see also Felix T. Wu, Defining Privacy and Utility in Data Sets, 84 U. Colo. L. Rev. 1117, 1141–42 (2013) (attributing the origins of the concept to the work of Latanya Sweeney). I am grateful to Peter Swire for pointing me to this literature.

   [191]. Bellovin et al., supra note 190, at 598 (quoting Latanya Sweeney, k-Anonymity: A Model for Protecting Privacy, 10(5) Int’l J. on Uncertainty, Fuzziness and Knowledge-Based Sys. 557, 557 (2002)).

   [192]. The concept of k-anonymity is not meant to be a literal guide to determining the boundary between indiscriminate and closed universe searches, as there are “known limitations” to the concept and it imperfectly fits the concept enumerated here. Id. at 1142.

   [193]. Cf., e.g., Daphna Renan, The Fourth Amendment as Administrative Governance, 68 Stan. L. Rev. 1039, 1042, 1051–67 (2016) (criticizing the “transactional” nature of the search inquiry in constitutional law as applied to programmatic surveillance and proposing an administrative governance model).

   [194]. Compare State v. Glover, 89 A.3d 1077, 1082 (Me. 2014) (“[I]n most circumstances, there is a constitutional right to refuse to submit to warrantless DNA sampling.”), with Joh, supra note 39, at 865 (citing cases upholding collection of abandoned DNA).

   [195]. Cf. Carpenter v. United States, 138 S. Ct. 2206, 2218 (2018) (noting that technology gave “police access to a category of information otherwise unknowable” since location information is “continually logged for all of the 400 million devices in the United States—not just those belonging to persons who might happen to come under investigation”).

   [196]. This idea of sequential authorization was pressed by Justice Jackson at the recent oral argument in Chatrie. Transcript of Oral Argument at 70, No. 25-112 (Apr. 27, 2026), https://www.supremecourt.gov/oral_arguments/argument_transcripts/2025/25-112_7k47.pdf [https://perma.cc/76AE-TXWU] (questioning why counsel was not “arguing something about needing more probable cause as each one of these steps goes along”); id. at 74–75 (suggesting “real problem” is lack of information about how the government went from nineteen accounts to nine accounts); id. at 136–38 (describing an “iterative process” where “you need more to get to each step”). This is not to say that each step must be the least intrusive means of ascertaining the perpetrator’s identity. That is because many “minor” intrusions may be more collectively intrusive than a single “major” intrusion. For instance, it may be that a simple DNA test would conclusively identify the likely perpetrator. In such a case, that may be preferable to a series of more incremental intrusions (such as location tracking or image scanning).

   [197]. Md. Code Ann., Crim. Proc. §§ 17-101–17-105

   [198]. Id. § 17-102(b).

   [199]. Id. § 17-102(f).

   [200]. Id.

   [201]. Murphy, supra note 11, at 834.

   [202]. See, e.g., Friedman & Citron, supra note 2, at 1423–37 (urging greater legislative action regulating data searches with a backstop of robust judicial review); Gee, supra note 160, at 322–23 (citing efforts to regulate various technologies).

   [203]. See, e.g., Andrew Guthrie Ferguson, Facial Recognition and the Fourth Amendment, 105 Minn. L. Rev. 1105, 1197 (2021) (proposing a legislative framework for facial recognition surveillance). Orin Kerr has likewise proposed limits on the use of material garnered from a lawful search that ultimately proves nonresponsive, Kerr, supra note 138, at 18 (2015), and on the use of the federal preservation statute to freeze accounts, Orin S. Kerr, The Fourth Amendment Limits on Internet Content Preservation, 65 St. Louis U. L.J. 753, 756–57, 799–802 (2021) (arguing that a broad interpretation of the preservation statute is unconstitutional and that freezing accounts is permissible only for the time it takes to draft a warrant).

   [204]. See, e.g., Jantz, supra note 3, at 25–26 (“Some courts have found that a requirement that the government delete any data associated with uninvolved devices is a crucial minimum standard in an appropriately tailored geo-location warrant.”); Bill v. Brewer, 799 F.3d 1295, 1302 (9th Cir. 2015) (dismissing concerns about misuse of compelled samples by noting that the government “expressly guarantee[d]” the samples “will be used for comparison to evidence in this report only and will not be used for any research type testing, including race, ethnicity or health, provided to any outside organization for those purposes, entered into the employee database, or entered into CODIS” (internal quotation marks omitted)).

   [205]. See,e.g.,Swire & Murphy, supra note 134, at 1–2 (citing case law and scholarship that tethers constitutionality to statutory authorization).

   [206]. But see Kerr, supra note 189, at 1139–49 (critiquing deference to legislatures and instead encouraging courts to continue to act independently in interpreting the Fourth Amendment).

   [207]. See, e.g., Ronald Dworkin, Taking Rights Seriously xi (1977) (“[I]ndividuals can have rights against the state that are prior to the rights created by explicit legislation.”).