Why do the most far-reaching government monitoring programs almost never receive a ruling on whether they are legal? That is the threshold question of surveillance law in the United States, and its answer says more about the field than any single holding. A reader who works through this article will be able to explain why collection programs are so rarely reviewed on the merits, name the doctrine that defeats most challenges before any panel reaches the substance, trace the decisions that did reach the merits and what each held, and understand the constitutional shift that has quietly done more to limit monitoring than any case brought against a national security program.

The pattern has a name. The standing wall: American surveillance law is shaped less by what judges have held than by what they have declined to reach, because secrecy keeps plaintiffs from proving injury, and the most consequential constitutional development in the field arrived through an ordinary criminal case about cell phone records rather than through any challenge to a national security program. That claim sounds sweeping until the docket is laid out case by case. The docket is what this article does. Decisions handed down after this article’s reference date of June 1, 2014 are marked as such wherever they appear, so the reader can always tell which rulings belong to the period the article covers and which arrived later.

The litigation described here construes two framework statutes, and each deserves its formal identity stated once. The first is the Foreign Intelligence Surveillance Act of 1978, Public Law 95-511, enacted October 25, 1978 and codified principally at 50 U.S.C. sections 1801 and following, which created a special court and a statutory scheme for foreign intelligence collection inside the United States. The second is Title III of the Omnibus Crime Control and Safe Streets Act of 1968, known as the Wiretap Act, Public Law 90-351, enacted June 19, 1968 and codified principally at 18 U.S.C. sections 2510 and following, which set the rules for criminal wiretapping and electronic eavesdropping by law enforcement. Almost every dispute in this field asks how one or both of these laws, and the Fourth Amendment behind them, apply to techniques their drafters never imagined.

Approaching the subject through litigation history, rather than through the statutes alone, changes what the reader sees. A statute describes what the government may do; the cases describe what happens when someone objects, and the objections reveal the system’s real boundaries. The chapters that follow move from the threshold that defeats most challenges, through the warrant rule and its deliberate reservation, through the secret tribunal’s restructuring of the intelligence-prosecution boundary, to the rare merits decisions, the doctrinal shift that arrived from the criminal docket, and the secrecy privilege that guards the innermost evidence. The through line is the obstacle: each decision either surmounts it, succumbs to it or redraws it.

Secrecy in this field operates on two levels, and both matter. The first is operational: the targets, methods, and scope of collection are classified, so outsiders cannot know whether they have been affected. The second is legal: for years, the specialized court’s interpretations of the governing statutes were themselves classified, which meant that the law authorizing collection was partly secret law, knowable to the executive and the tribunal but not to the public. A plaintiff challenging a program might not know the program exists, might not know the legal theory that authorizes it, and might not be able to prove personal injury even with both pieces of knowledge. Each layer of secrecy reinforces the others, and together they make the threshold doctrines nearly insurmountable.

Three rulings from the period anchor the story. In 1972 the Supreme Court required a warrant for domestic security monitoring while pointedly reserving the foreign intelligence question. In 2002 the intelligence court’s court of review issued its first published opinion and dismantled the wall between intelligence gathering and criminal prosecution. In 2013 the Supreme Court held that challengers to the section 702 framework lacked standing, ending the most direct constitutional attack on that program without reaching its substance. The later rulings, a Second Circuit decision in 2015 that found the bulk metadata program unlawful under the statute, a Supreme Court decision in 2018 that made historical location records a search, a Ninth Circuit decision in 2020 that found the same program violated the statute while declining to suppress, and a unanimous 2022 decision on the state secrets privilege, each arrived after the lock date and are dated where they appear. Together they show a field in which the obstacle is usually not the holding but the hearing.

The standing wall in surveillance law

Why is it so hard to sue over secret surveillance?

A plaintiff must prove personal injury to enter federal court, and a secret program by design leaves its targets unable to prove they were watched. In Clapper v. Amnesty International USA (2013), the justices rejected a challenge to the section 702 framework because the feared interception rested on a speculative chain of possibilities.

The injury requirement is not a contrivance of the surveillance field. Article III of the Constitution limits federal courts to cases and controversies, and the standing doctrine enforces that limit in every kind of litigation, from environmental suits to tax challenges. A plaintiff must show an injury in fact that is concrete and particularized, a causal link between the injury and the challenged conduct, and a likelihood that a favorable decision will redress the injury. The doctrine is described here neutrally because it functions neutrally: it keeps courts from issuing advisory opinions and reserves policy disputes for the political branches. The difficulty is that the same neutral requirement, applied to a program whose targets are never told they are targets, produces a distinctive result. The very secrecy that makes the program effective also makes it nearly impossible for any private plaintiff to satisfy the first element.

The particularity requirement, the demand that the injury affect the plaintiff in a personal and individual way, does independent work in surveillance cases. Even if a challenger could prove that a program exists and operates unlawfully, that proof would not establish standing without showing that the challenger’s own communications were acquired. Generalized grievances about unlawful government conduct, shared by every citizen, do not satisfy Article III. The requirement makes sense as a limit on judicial power, preventing courts from becoming councils of revision for every contested policy, but it means that the more widely a program sweeps, the harder it is for any individual to distinguish their injury from everyone else’s. Bulk collection thus presents a paradox: its very breadth, which makes it most concerning, also makes it hardest to challenge.

Clapper v. Amnesty International USA, 568 U.S. 398 (2013), presented the problem in its purest form. The plaintiffs, a coalition led by Amnesty International USA that included journalists, attorneys, and human rights researchers who regularly communicated with people abroad, challenged section 702 of the Foreign Intelligence Surveillance Act, codified at 50 U.S.C. section 1881a, the framework described in the companion guide to FISA section 702. They did not claim they could prove their own communications had been intercepted. They argued instead that interception was sufficiently likely to count as certainly impending, and that they had already spent money and changed their professional practices to guard against it.

Justice Alito’s majority opinion, joined by four colleagues in a 5-4 decision, rejected both theories. The first theory failed because the feared injury rested on what the opinion called a highly attenuated chain of possibilities: the government would have to decide to target the plaintiffs’ foreign contacts, invoke its section 702 authority rather than another source of authority, succeed in intercepting the communications, and do so in a way that touched the plaintiffs. Each link was possible, but the chain as a whole was too speculative to count as certainly impending injury in fact. The second theory failed because the costs the plaintiffs had incurred to protect their communications could not manufacture standing out of a hypothetical harm. A plaintiff cannot create injury by spending money to guard against a threat the law does not recognize as sufficiently real. Justice Breyer’s dissent, joined by Justices Ginsburg, Sotomayor and Kagan, gave the competing view with equal care: the dissent argued that the plaintiffs’ apprehension was well founded, that interception was not merely possible but probable, and that the measures they had taken were a present and concrete response to a realistic threat rather than an effort to manufacture a lawsuit. The disagreement was not about the standing rules but about how certain the threatened interception had to be before it counted.

The majority opinion contained a passage that would later cause the government significant embarrassment. Quoting the government’s own brief, the opinion repeated the representation that criminal defendants would receive notice of section 702-derived evidence and could challenge the program’s lawfulness in that forum: “If the Government intends to use or disclose information obtained or derived from a [FAA] acquisition in judicial or administrative proceedings, it must provide advance notice of its intent, and the affected person may challenge the lawfulness of the acquisition.” Solicitor General Donald Verrilli had made the same representation to the justices at oral argument on October 29, 2012. The logic was a safety valve. If civil plaintiffs could not prove they had been monitored, then criminal defendants who received formal notice of section 702-derived evidence could litigate the constitutional questions where standing was indisputable.

The safety valve turned out to have been closed for years. In October 2013, several months after the Clapper decision, the Department of Justice filed section 702 notices in two criminal cases, United States v. Muhtorov in the District of Colorado and United States v. Mohamud in the District of Oregon, belatedly acknowledging that evidence in those prosecutions had been obtained or derived from section 702 collection. Reporting then revealed why no such notices had ever been filed before: the Justice Department’s National Security Division had given zero section 702 notices from 2008 to 2013 because its internal reading of which evidence counts as “obtained or derived from” section 702 surveillance was extremely narrow. Under that reading, evidence that flowed from section 702 collection through intermediate steps did not trigger the notice obligation, so defendants never learned of the collection and never had the chance the Clapper majority had described. The department revised its notice policy, and from October 2013 forward it undertook to notify defendants in cases involving section 702-derived evidence. Six defendants received belated notice between October 2013 and April 2014.

The political response was pointed. On November 21, 2013, Senators Mark Udall, Ron Wyden and Martin Heinrich wrote to Solicitor General Verrilli urging him to formally correct the record before the Supreme Court, saying the government had “misrepresented its policy and practice to the Court.” The episode did not change the Clapper holding, which stands on its own reasoning about speculative injury, but it changed how the holding is understood. The decision’s assurance that constitutional challenges could migrate to criminal cases rested on a factual premise about government practice that was wrong for the entire life of the statute up to that point. For a field built on secrecy, the correction saga is almost a parable: even the representations made to the highest court about how the system worked did not match how the system worked.

The procedural posture of Clapper underscores how long the standing question consumed the litigation. The district court for the Southern District of New York granted summary judgment for the government in 2009, reported at 646 F. Supp. 2d 633, holding the plaintiffs lacked standing. The Second Circuit reversed in 2011, reported at 638 F.3d 118, finding standing on the ground that the plaintiffs’ fear of interception was objectively reasonable and their protective expenditures were a present injury. The Supreme Court then reversed 5-4 in 2013, ending the case on the threshold without adjudicating whether section 702 collection comports with the Fourth Amendment. Five years of litigation produced no answer to the constitutional question the plaintiffs had set out to ask. That is the standing wall in operation: not a ruling that the program is lawful, but a ruling that no one before the court was entitled to ask.

The same pattern recurs across the field, and it explains the complication this article must address directly. It is sometimes said that courts have blessed these programs, that years of litigation have produced judicial approval. The record shows the opposite of a blessing. Most challenges were dismissed without reaching the merits, on standing or on secrecy grounds, and the two appellate panels that did reach the merits of the bulk collection program both found that it violated the statute. The absence of adverse constitutional rulings is largely an artifact of justiciability, not a body of precedent upholding the programs. A parallel can be seen in a different field, where deference doctrines have similarly kept substantive questions from decision, a comparison developed in the companion survey of immigration law in the Supreme Court. In both areas, what the courts have declined to decide shapes the law more than what they have decided.

Consider the epistemic trap that secrecy creates. To show injury in fact, a plaintiff must demonstrate a concrete and particularized harm that is actual or certainly impending. But a well-run collection program discloses neither its targets nor its methods, which means the people most likely to have been monitored are the least able to prove it. The paradox is structural: the more effective the secrecy, the less reviewable the program. A plaintiff who sues on the basis of a realistic fear is told the fear is speculative; a plaintiff who waits for proof may wait forever, because the proof is classified. Clapper resolved this paradox in favor of the certainty requirement, holding that the chain of contingencies between the statute’s existence and any particular interception was too attenuated. The decision did not deny that interception was possible, or even likely for some plaintiffs. It held that possibility and likelihood, without more, do not satisfy Article III.

The second standing theory deserved its own analysis because it recurs in later litigation. The plaintiffs argued that the money and effort they had spent to protect their communications constituted a present injury. The majority rejected the theory on the ground that a plaintiff cannot manufacture standing by incurring costs to avoid a hypothetical harm. The logic is straightforward: if protective expenditures created standing, then anyone apprehensive about any government program could buy a lawsuit by spending money to guard against it, and the injury requirement would collapse into a pleading exercise. The dissent’s answer was equally straightforward: when the threatened harm is realistic rather than hypothetical, the costs of avoiding it are a genuine present burden, not a manufactured one. The 5-4 division shows how much turned on characterization. Both sides applied the same doctrine; they disagreed about whether the plaintiffs’ fear belonged in the category of the realistic or the speculative.

Injury in fact has a taxonomy worth learning, because surveillance injuries sit in its hardest category. The doctrine recognizes actual injuries, harms already suffered, and imminent injuries, harms certain to occur, while rejecting conjectural or hypothetical ones. Most legal injuries fit comfortably: the plaintiff was fired, the property was taken, the regulation applies to her business. Surveillance injuries are probabilistic by nature. The harm is not that the plaintiff was monitored but that she may have been, and the probability cannot be quantified without the classified facts. Clapper placed probabilistic surveillance injury on the conjectural side of the line. The consequence is that an entire category of government action, action whose very nature makes its harms probabilistic, is largely insulated from the doctrine’s protection, not because the harms are unreal but because they are unprovable.

Other areas of law handle probabilistic injury more generously, which throws the surveillance rule into relief. Environmental plaintiffs have been heard on the ground that they reasonably fear losing the use of a place they visit; consumers have been heard on the ground that a data breach exposes them to a realistic risk of identity theft. In those settings courts accept reasonable probability as sufficient, because the alternative is to leave real harms without remedy. The Clapper majority declined to extend that generosity to surveillance challengers, insisting on the stricter certainly-impending formulation. Whether that distinction reflects a principled difference between the fields or simply the gravitational pull of national security on the doctrine is a question the opinions do not answer, and it remains one of the live puzzles for anyone studying the standing wall.

The choice between the Second Circuit’s 2011 approach and the Supreme Court’s 2013 approach continues to matter for every future challenge. The appellate court had found standing on a more forgiving view of the plaintiffs’ showing; the Supreme Court demanded certainly impending injury and found the chain of possibilities too attenuated. The difference between those standards is the difference between a world in which realistic apprehension opens the courthouse door and a world in which only near-certainty does. The Supreme Court chose near-certainty. That choice does not reflect any special hostility to surveillance challengers; the certainly-impending standard governs standing across subject matters. But its interaction with programmatic secrecy is uniquely severe here, because in few other fields is the plaintiff so systematically deprived of the information needed to satisfy the standard. The doctrine is general; the burden is specific.

One question the standing wall raises is whether Congress could simply legislate it away, and the answer shows why the barrier is so durable. Standing enforces Article III’s case-or-controversy limitation, which is constitutional rather than statutory. Congress can create new legal rights whose violation counts as injury, and it has done so in many fields, but it cannot authorize federal courts to hear claims by plaintiffs who have suffered no concrete harm, because the Constitution withholds that power. Any legislative fix would therefore have to work indirectly: by requiring the government to disclose programs, to notify targets, or to create administrative remedies that generate the concrete injuries standing demands. The barrier is constitutional in its foundation, which means it yields to politics only through transparency, not through jurisdictional tinkering.

The opinion’s central formulation, “certainly impending,” does more work than it first appears. By choosing that phrase over looser alternatives like “substantial risk,” the majority set the injury inquiry at the most demanding point on the spectrum of future-harm standards. A substantial-risk test asks whether the feared harm is likely enough to justify present concern; a certainly-impending test asks whether it is close enough to count as present. In most fields the difference between those formulations rarely matters, because the plaintiff can usually describe the mechanism of the threatened harm with specificity. In surveillance litigation the difference is decisive, because the mechanism is classified. The plaintiff who must describe exactly how interception will occur, without knowing the government’s targeting decisions, cannot satisfy a standard that demands near-certainty about future acts of discretionary officials. The formulation is neutral on its face and devastating in this application, which is why the standing wall endures regardless of who sits on the Court.

It is equally important to say what Clapper did not decide, because the decision’s narrowness is often lost. The Court did not hold that section 702 collection is constitutional. It did not hold that no plaintiff could ever challenge the framework. It held that these plaintiffs, on this record, could not show certainly impending injury. A plaintiff who could prove interception, through an official disclosure, a criminal notice or a confirmed acknowledgment, would present a different case. The standing wall is therefore not an absolute bar but a filter that admits only those who can prove what the system is designed to conceal. That the filter admits almost no one is a fact about secrecy, not a holding about the statute’s validity.

There is an irony in how standing is actually satisfied in this field, and it deserves a direct statement. The plaintiffs who reach the merits are almost always those who benefit from unauthorized disclosures or from the government’s own strategic revelations. The bulk metadata plaintiffs succeeded because the program had been disclosed; the Clapper plaintiffs failed because section 702’s operational details had not been. This makes the standing wall, in practice, a function of publicity rather than of law: the same legal theory succeeds or fails depending on what the public happens to know. A doctrine meant to keep courts within their constitutional role ends up making judicial review contingent on leaks, journalism and the executive’s disclosure choices. The rule of law is not well served when the courthouse door opens only after someone outside the system forces it.

The Breyer dissent has had an afterlife that matters for later litigation, even though dissents do not make law. Later challengers invoked its reasoning, arguing that their circumstances supplied the concreteness the Clapper majority found lacking: a disclosed program, a subscriber relationship, a notice in a criminal case. Lower courts sometimes found these distinctions persuasive and sometimes not, but the dissent supplied the vocabulary for the argument, the idea that realistic apprehension plus protective cost can constitute present injury. The 5-4 division thus structured the litigation that followed, with courts applying the certainly-impending standard and challengers arguing their facts satisfied even that standard. A dissent that loses the case can still shape the conversation, and this one has.

The notice correction saga also illustrates a broader point about how secrecy degrades the judiciary’s own information. The Supreme Court decides cases on the record the parties present, and in surveillance litigation the most important parts of that record are classified and in the government’s hands. When the Solicitor General described the notice practice to the justices, the Court had no independent way to verify the description, and no adversary with access to the classified facts stood ready to correct it. The correction came months later, through reporting and a senators’ letter, not through the adversarial process. The episode is a reminder that the standing wall does not only keep plaintiffs out; it keeps the Court itself dependent on the executive’s account of how the system works, and that dependence can mislead even the most careful judges.

Step back and the shape of the standing wall becomes clear. Five years of litigation, from the 2009 district court decision through the 2011 appellate reversal to the 2013 Supreme Court decision, produced no answer to the question the plaintiffs asked: whether section 702 collection comports with the Fourth Amendment. The docket is full and the merits are empty. That is not a malfunction of the system; it is the system operating as designed, with a threshold doctrine doing exactly what threshold doctrines do. But when the threshold doctrine interacts with programmatic secrecy, the result is a body of law in which the most important questions are the least answered. Every later section of this article should be read against that background. The cases that reached the merits are the exceptions, and their exceptional character is itself part of the story.

The warrant rule and its reservation

What did the Keith decision require of domestic security wiretaps?

The justices held unanimously that the government must obtain prior judicial approval before conducting domestic security monitoring, even when national security is invoked. The 1972 decision left open, explicitly and deliberately, whether the same requirement applies to the activities of foreign powers or their agents.

United States v. United States District Court, 407 U.S. 297 (1972), known as the Keith case, arose from a criminal prosecution in which the government acknowledged conducting warrantless electronic surveillance for domestic security purposes and resisted disclosing it to the defense. The district court ordered disclosure so that the lawfulness of the surveillance and its effect on the prosecution could be determined, the court of appeals affirmed, and the government sought review. The government argued that domestic security surveillance fell outside the warrant requirement.

Justice Powell’s opinion for an 8-0 Court, with Justice Rehnquist not participating, rejected the government’s position in terms that still frame the field. “Prior judicial approval is required for the type of domestic security surveillance involved in this case,” the opinion stated at 407 U.S. 324. The reasoning balanced two legitimate interests. The government’s interest in protecting national security is genuine, the Court acknowledged, and domestic security investigations differ from ordinary criminal investigations in ways that might justify different warrant procedures: the emphasis is on prevention rather than prosecution, the investigations run longer, and post-surveillance notice to the target could be inappropriate. But those differences, the Court held, go to the form of the warrant procedure, not to whether judicial approval is required at all. The Fourth Amendment’s warrant clause stands between the executive and the citizen even when the executive acts in the name of security, and the risk of executive overreach is at its height precisely when the government claims an open-ended security justification.

The opinion was careful to leave Congress room to design a suitable procedure. It suggested that a warrant application in the domestic security context might show a different kind of probable cause than in an ordinary criminal case, that the application and order might remain secret, and that notice to the target might be withheld where it would compromise the investigation. What it would not permit was the executive authorizing its own surveillance with no neutral magistrate involved. The requirement of prior judicial approval, the Court explained, is the central mechanism by which the Fourth Amendment restrains unreasonable searches, and domestic security surveillance is not exempt from it.

Then came the reservation that shaped the next half century of legislation. The Court stated plainly that it had not addressed the foreign intelligence question: “We have not addressed, and express no opinion as to, the issues which may be involved with respect to activities of foreign powers or their agents,” at 407 U.S. 321 to 322. The case involved only the domestic aspects of national security, and the Court confined its holding to that domain. The reservation was an invitation as much as a boundary. Congress accepted it six years later by enacting the Foreign Intelligence Surveillance Act of 1978, building the special court and the statutory framework for foreign intelligence collection that the Keith opinion had left unaddressed. Every later dispute about whether that framework satisfies the Fourth Amendment traces back to this deliberate gap: the justices required warrants for domestic security work and said nothing about foreign intelligence work, leaving the political branches to build a system the courts would later be asked to judge.

The reservation created a line-drawing problem that has never been fully solved. Keith required warrants for domestic security surveillance and said nothing about the activities of foreign powers or their agents, which means every investigation must be sorted into one category or the other before the constitutional rule is known. But investigations do not always sort neatly. A domestic organization with foreign contacts, a foreign agent operating inside the United States, an American communicating with a target abroad: each sits near the boundary the opinion drew but did not map. The reservation’s elegance as judicial restraint, deciding only the case at hand, became a practical difficulty for the agencies, which needed to know which rule governed the ambiguous middle. That difficulty was among the forces that pushed Congress toward a comprehensive statute.

Congress answered the invitation with the institution this article’s later sections describe: a special court, secret proceedings, and a statutory framework built specifically for foreign intelligence collection inside the United States. The design reflects Keith’s template, the requirement of prior judicial approval adapted to a new context, rather than the executive self-authorization the opinion rejected. Whether the adaptation satisfies the Fourth Amendment is the question the Court reserved, and it remains the great unanswered question of the field. Keith’s domestic holding is settled law; Keith’s reservation is the field’s founding silence, and everything since has been an attempt to fill it.

Keith also established a tonal precedent for how courts write about surveillance. The opinion treated the executive’s security claims with respect but not deference, insisting that respect for the executive’s expertise does not displace the judiciary’s duty to enforce the warrant requirement. That posture, respectful scrutiny rather than either abdication or hostility, recurs in the later merits decisions. Notable is what Keith did not do: it did not hold that warrantless foreign intelligence surveillance is unconstitutional, it did not prescribe the procedures Congress should adopt, and it did not reach any question of standing or secrecy, because the government itself had disclosed the surveillance in the course of a criminal prosecution. The case reached the merits only because the monitoring surfaced in open court, a foreshadowing of the Clapper problem four decades later.

The warrant requirement the Court enforced has a specific constitutional function worth stating plainly. The Fourth Amendment does not forbid searches; it forbids unreasonable searches, and the warrant clause supplies the mechanism for distinguishing the two. Prior judicial approval interposes a neutral magistrate between the executive’s suspicions and the citizen’s privacy, requiring a showing of probable cause before the intrusion begins. The government’s argument in Keith was that this mechanism, designed for ordinary criminal investigations, had no place in domestic security work, where the executive’s expertise and the stakes of prevention justified self-authorization. The Court’s answer was that the mechanism matters most where the justification is broadest. An open-ended security rationale, left to the executive alone to invoke, is precisely the circumstance in which the risk of overreach is greatest, and the warrant requirement is the Constitution’s chosen check on that risk.

The Fourth Amendment contains two clauses that are often in tension, and Keith is the classic illustration of their interplay. The first clause forbids unreasonable searches and seizures; the second describes the warrant, supported by probable cause and particularly describing the place to be searched and things to be seized. One long-running debate asks whether the warrant clause defines reasonableness, so that a search without a warrant is presumptively unreasonable, or whether reasonableness is the ultimate test and warrants are merely one way to satisfy it. The government’s position in Keith leaned on the second view: domestic security surveillance was reasonable given the stakes, so the absence of a warrant did not matter. The Court’s answer leaned on the first: for this category of intrusion, reasonableness requires prior judicial approval, and no showing of executive good faith substitutes for it.

The government’s deeper argument invoked the separation of powers. Foreign affairs and national defense are executive domains, the argument ran, and the judiciary has neither the expertise nor the constitutional warrant to supervise them. The Court did not reject the premise that the executive leads in security matters; it rejected the inference that leadership means exclusivity. The Constitution assigns the executive the conduct of security policy and assigns the judiciary the enforcement of the Bill of Rights, and those assignments coexist. A warrant requirement does not transfer security policy to judges; it requires the executive to persuade a neutral magistrate before intruding on Americans’ privacy. That division of labor, executive initiative checked by judicial approval, is the constitutional design, and Keith enforced it against the claim that security emergencies dissolve it.

The opinion’s choice of words repays attention. Powell wrote that “prior judicial approval is required,” not simply that a warrant is required, and the phrasing matters. A warrant, in the ordinary criminal sense, carries a full apparatus: probable cause of crime, particular description, neutral magistrate, and eventually notice. Prior judicial approval is the functional core of that apparatus, the neutral check before the intrusion, without all of its ordinary-crime trappings. By framing the requirement at that level of generality, the Court created a principle that could travel: the executive must persuade a judge before intruding, but the form of the persuasion can adapt to the context. That is why the holding survived the migration from domestic security wiretaps to the foreign intelligence framework. The principle is portable because it was stated as a function rather than a form, and Congress’s later design work was an exercise in giving the function new forms.

The opinion’s flexibility about procedure is as important as its firmness about principle. The Court did not order the government to use ordinary criminal warrants for security surveillance. It invited adaptation: the probable cause showing might look different where the purpose is prevention rather than prosecution, applications and orders might remain secret, and notice to the target might be delayed or withheld where it would compromise the investigation. This distinction between the requirement of judicial approval and the form it takes became the template for everything that followed. Congress took the invitation six years later, designing a specialized court and specialized procedures for foreign intelligence collection. The adaptability of the warrant requirement, not just its existence, is Keith’s lasting contribution.

The reservation deserves emphasis because it is the hinge on which the modern field turns. By stating that it expressed no opinion on the activities of foreign powers or their agents, the Court deliberately left the hardest question open. The reservation was candid about the limits of the case before it: the surveillance at issue was domestic, and the Court would not decide what it had not been asked to decide. But the candor created a vacuum, and Congress filled it with the 1978 statute. Every later argument about whether that statute satisfies the Fourth Amendment is, in a sense, an argument about what Keith left unsaid. The decision’s silence on foreign intelligence has been quoted, debated and built upon for decades, which is a reminder that what courts decline to decide can be as consequential as what they hold.

One underappreciated feature of Keith is how it reached the merits at all. The surveillance surfaced because the government sought to prosecute, and prosecution required disclosure, and disclosure enabled challenge. The criminal process functioned as the transparency mechanism that civil litigation so often lacks. The defendants could litigate the lawfulness of the monitoring because the government wanted to use its fruits in open court. Four decades later, the Clapper majority would invoke the same mechanism as the answer to the standing problem, assuring that criminal defendants with notice could litigate what civil plaintiffs could not. Keith is thus both the source of the warrant rule and an early illustration of the forum problem: constitutional review of surveillance tends to happen, if it happens, where the government chooses to appear.

The 8-0 division, with Justice Rehnquist not participating, signals something about the decision’s foundations. Unanimity on the core holding, that domestic security surveillance requires prior judicial approval, suggests the principle was not ideologically contested even amid the era’s security anxieties. The justices disagreed about many things in that period; they did not disagree that the executive cannot be the sole judge of its own surveillance authority over Americans. That consensus has held. No later decision has questioned Keith’s domestic rule. The disputes have all concerned the reserved question, the foreign intelligence domain the Court left open, which is where the field’s energy has concentrated ever since.

Justice Powell’s balancing analysis repays closer study because it established the method courts would use for decades when weighing security against privacy. The opinion identified the government’s interest in protecting domestic security, the individual’s interest in privacy and free expression, and the public’s interest in both security and liberty, then asked whether the warrant procedure could serve the first without sacrificing the second, concluding that prior judicial approval, with its requirements of probable cause and particularity, was compatible with effective security work. The executive’s argument that judges lacked the expertise to evaluate security threats received a pointed answer: the judiciary’s lack of operational expertise was precisely why its detached judgment was valuable, as a check on the institutional biases of those conducting the monitoring.

Finally, Keith established the tone in which courts address executive security claims: respectful but not deferential. The opinion acknowledged the government’s genuine interest in protecting national security and the executive’s superior expertise in assessing threats, then insisted that respect for expertise does not displace the judiciary’s duty to enforce constitutional limits. That posture recurs in the later merits decisions, which treat the government’s factual claims seriously while refusing to abdicate the legal judgment. It is the opposite of both hostility and surrender, and it explains why the field’s landmark opinions so often divide closely: the balance between security and liberty, struck by judges who take both seriously, rarely produces unanimity.

The court built for secrecy

How did a 2002 opinion reshape the boundary between intelligence and prosecution?

The Foreign Intelligence Surveillance Court of Review, in its first published opinion, sustained the PATRIOT Act’s change from a primary purpose test to a significant purpose test and struck down the lower court’s separation rules. The wall between intelligence gathering and criminal prosecution, built by practice and then by court order, came down in a single per curiam decision.

The institutional setting needs explanation, because the courts in this field do not look like ordinary courts. The Foreign Intelligence Surveillance Act created the Foreign Intelligence Surveillance Court, composed of sitting federal district judges designated by the Chief Justice, which hears the government’s applications for surveillance orders in secret, with only the government present. Above it sits the Court of Review, composed of three designated judges, which hears appeals from the lower court’s denials. For the first twenty-four years of the statute’s life, the Court of Review never issued a published opinion, because the government almost never lost below and had no occasion to appeal. The structure of these tribunals, and how cases move between them, is laid out in the companion guide to the 1978 Act.

The first appeal arrived in 2002. The USA PATRIOT Act had amended section 1804(a)(7)(B) of the statute through its section 218, changing the certification the government must make from “the purpose” of the surveillance being foreign intelligence collection to “a significant purpose.” The old formulation had been read to require that foreign intelligence be the primary purpose of the surveillance, and a set of procedures known as the wall had grown up to enforce the separation: intelligence officials and criminal prosecutors were kept apart, information could not freely flow between them, and the FISA court policed the boundary. In May 2002 the FISA court issued an opinion imposing minimization and separation restrictions on the government in connection with FISA applications. The government appealed those restrictions to the Court of Review, the first appeal in the tribunal’s history, presenting a court that had never before spoken publicly with its first occasion to do so.

In re Sealed Case, 310 F.3d 717 (Foreign Intelligence Surveillance Court of Review 2002), decided November 18, 2002, was that first published opinion, issued per curiam. Its core holding was straightforward: the amended significant purpose standard is constitutional, and the FISA court’s May 17, 2002 restrictions “are not required by FISA or the Constitution.” The review court reversed the restrictions and dismantled the wall. The reasoning ran through several layers. The text of the amendment was plain, the court explained: Congress had deliberately replaced the primary purpose test with a significant purpose test, and the statute as amended permits surveillance so long as foreign intelligence collection is a significant purpose, even if criminal prosecution is also a purpose. The review court rejected the argument that this arrangement violated the Fourth Amendment, holding that the statute’s procedures, including judicial approval based on probable cause that the target is a foreign power or its agent, satisfy the reasonableness requirement. It also rejected the notion, advanced by some amici, that the wall had been constitutionally required all along; the separation had been a prudential practice, not a constitutional command, and Congress was free to alter it.

The opinion is notable for what it assumed as much as for what it held. The review court proceeded on the premise, unexamined because unchallenged in that proceeding, that the FISA framework as a whole comports with the Fourth Amendment. No adversarial party had ever tested that premise in the Court of Review, and none did in Sealed Case either; the dispute was between the government and the FISA court’s own restrictions, with amici filling but not fully occupying the adversarial role. The decision thus stands as the field’s most important merits ruling on the statute’s structure, and also as an illustration of the field’s structural limits: the most consequential opinion about the law was issued by a court that hears only one side, in a case where the constitutional premise went uncontested. The wall’s demolition reshaped practice immediately, permitting the coordination between intelligence and law enforcement that the amendment envisioned, and it set the terms on which every later challenge to the statute’s architecture would have to argue.

The change from “the purpose” to “a significant purpose” sounds technical, and in a sense it is: a single adjective replaced in a certification requirement. Its practical meaning was anything but technical. Under the old reading, the government had to certify that foreign intelligence collection was the primary purpose of the surveillance, which meant that criminal prosecution had to be subordinate, and the wall’s procedures enforced that subordination by restricting contact between intelligence officers and prosecutors. Under the amended standard, foreign intelligence collection need only be a significant purpose, which permits the government to pursue prosecution as a co-equal or even dominant objective so long as the intelligence purpose remains significant. The amendment thus authorized what the wall had forbidden: the free flow of information between the intelligence and law enforcement worlds, and the use of FISA-derived material in criminal cases without the old separation.

The wall itself deserves a clear description, because its demolition is the decision’s headline. The wall was not a single rule but an accumulation of practices: guidelines restricting communications between intelligence agents and criminal prosecutors, procedures requiring high-level approval before information crossed the divide, and eventually court-imposed minimization and separation requirements. It had been built on the premise that the primary purpose test demanded real separation, and the FISA court’s May 17, 2002 opinion had given that premise judicial form. The review court’s response was that the premise had never been constitutionally required. Congress had now replaced the primary purpose test by statute, and the court held the replacement constitutional. The wall, deprived of both its statutory and constitutional foundations, fell.

The textual mechanics of the amendment are worth a final look, because they show how little statutory language it takes to move a field. Congress changed one word, replacing “the” with “a significant,” and the review court treated the change as deliberate and dispositive. The old formulation asked the government to certify a singular, dominant purpose; the new one asks it to certify a purpose that clears a threshold of significance. The difference between a definite article and an indefinite one, between “the purpose” and “a significant purpose,” became the difference between a regime of separation and a regime of coordination. The episode is a lesson in how courts read amendments: when Congress alters language that courts have glossed for decades, the judiciary takes the alteration as a signal that the old gloss is dead, and the new text gets a fresh, literal reading.

What makes Sealed Case enduring as precedent is also what makes it limited. The opinion is the most thorough judicial examination of the statute’s architecture, and it remains the leading authority for the proposition that the significant purpose standard comports with the Fourth Amendment. But it was issued by a tribunal that hears only the government, in a proceeding where the constitutional premise went unchallenged by any true adversary. The participants who argued against the government filed as friends of the court, not as parties; they could not develop a factual record or press a claim the way a defendant facing prosecution can. The decision’s reasoning is careful and extensive, and it has never been overturned, but its one-sided procedural posture is a permanent caveat. It is the field’s most important structural ruling and, simultaneously, an illustration of why structural rulings in this field carry an asterisk.

One more feature of the opinion repays attention: its treatment of the FISA court’s own prior gloss. The review court explained that the primary purpose test had been, in significant part, the lower court’s own interpretation rather than a statutory command, and that Congress was free to override that interpretation by amending the statute. This is an ordinary point about the relationship between courts and legislatures, but it lands with special force in a secret court system. When the only tribunal construing a statute sits in secret and hears one side, its interpretations can harden into unexamined orthodoxy. Sealed Case is the rare instance in which a higher tribunal publicly corrected the lower court’s reading, and the correction ran in the government’s favor. The episode shows both the value of appellate review within the system and how rarely that review occurs.

The per curiam form of the opinion is worth a moment’s attention. Per curiam, by the court, signals an institutional rather than individual voice, and the review court chose it for its first published statement. The choice fits the occasion: a new tribunal speaking for the first time on the architecture of the system it oversees, with no prior jurisprudence of its own to extend or distinguish. The opinion reads as the court introducing itself to the legal world, laying out its understanding of the statute in comprehensive terms rather than deciding a narrow dispute. That comprehensiveness is why the decision remains the leading authority on the significant purpose standard; there has been no second published opinion to refine or reconsider it.

One limit on the opinion’s scope is worth stating. The review court struck down the separation requirements as unnecessary to the statute or the Constitution; it did not purport to dismantle the rest of the statutory framework, including the court’s ongoing supervision of applications and the procedures governing them. The decision changed who could share information and for what purposes, permitting the coordination the amendment envisioned. It did not end judicial supervision of the collection itself. The wall fell; the court remained.

The decision’s influence extends beyond its holding because of the questions it framed for later litigants. By treating the wall as a prudential practice rather than a constitutional command, the opinion told future challengers where not to aim: arguments that the Constitution requires strict separation between intelligence and law enforcement would have to overcome Sealed Case’s contrary conclusion. By grounding the amendment’s constitutionality in the statute’s judicial-approval procedures, it told future defenders where to aim: the probable-cause and court-order requirements are the load-bearing elements, and any erosion of those protections would reopen the constitutional question. The opinion is thus both a shield and a map, protecting the amended framework while marking the routes by which it might someday be attacked.

The amendment was part of a broad legislative reworking of the statute’s purpose requirement, and the review court’s opinion must be read as the judiciary’s first word on that reworking. Before Sealed Case, the significant purpose standard existed only as statutory text and executive practice; no court had said whether it satisfied the Constitution. The opinion supplied that missing judicial approval, and its reasoning has structured every subsequent debate about the boundary between intelligence and prosecution. Had the review court struck the amendment down, the entire post-2001 coordination regime would have been stillborn. Instead the opinion ratified it, and the ratification, issued by the only tribunal empowered to speak, carried a finality that no executive legal memorandum could match.

The review court’s dispositive sentence, that the lower court’s restrictions “are not required by FISA or the Constitution,” is a double holding disguised as one. The first half is statutory: nothing in the Act’s text demands the separation the FISA court had imposed. The second half is constitutional: nothing in the Fourth Amendment demands it either. The doubling matters because each half does independent work. Had the court held only that the statute did not require the wall, Congress could have reimposed it by statute at any time, and the separation would have remained a live legislative option. Had it held only that the Constitution did not require the wall, the statute as then written might still have been read to require it, preserving the old practice by interpretation. By rejecting both foundations at once, the opinion closed the question in both dimensions, which is why the wall’s demolition was immediate and total rather than gradual and contested.

When the merits were reached

What did two appellate courts find when they finally reached the merits?

Both panels found that the bulk telephony metadata program violated the statute that supposedly authorized it. In May 2015, the year after this article’s lock date, the Second Circuit held the business records provision had never authorized it; the Ninth Circuit agreed years later, adding that the collection may have violated the Fourth Amendment while declining suppression.

The program at issue was the bulk collection of telephony metadata under section 215 of the USA PATRIOT Act, the business records provision codified at 50 U.S.C. section 1861, which requires the government to show that the things sought are relevant to an authorized investigation. After the program became publicly known, the American Civil Liberties Union sued, and the dispute reached the Second Circuit.

American Civil Liberties Union v. Clapper, 785 F.3d 787 (Second Circuit 2015), decided May 7, 2015, is a study in judicial restraint as much as in statutory interpretation. The panel held that section 215 had never authorized the bulk program. The word “relevant” in 50 U.S.C. section 1861(b)(2)(A), the court explained, cannot stretch to cover the indiscriminate collection of every American’s call records on the theory that all of them might someday prove relevant to some future investigation. Relevance requires a connection to an authorized investigation, and a program that collects everything in advance of any investigation has no such connection. Having resolved the case on statutory grounds, the court declined to reach the plaintiffs’ First and Fourth Amendment claims. The doctrine of constitutional avoidance counsels courts to decide cases on the narrowest available ground, and the panel followed that counsel: because the program violated the statute, there was no need to decide whether it also violated the Constitution. The standing problem that had doomed the earlier Clapper challenge did not recur here, because the program’s existence was by then a matter of public record and the plaintiffs’ status as telephone subscribers gave them a concrete stake.

The Second Circuit’s statutory reading repays close attention because it shows how much work a single word can do. The panel compared the business records provision to familiar legal standards, such as grand jury subpoenas and civil discovery, and found that none of those analogues permitted the limitless collection the executive defended. Relevance, the court explained, requires some limiting principle connecting the records sought to the investigation at hand; a theory under which everything is relevant because anything might someday matter reads the limitation out of the statute. In ordinary subpoena practice, relevance connects the thing sought to a specific, existing inquiry: the documents must bear on the investigation at hand, not on investigations that might someday exist. The government’s bulk-collection theory severed that connection, treating relevance as a property of the database rather than of the inquiry. The panel’s rejection restored the ordinary meaning: relevance is relational, and a demand with no pending or authorized investigation to relate to cannot satisfy it. The opinion also noted the anomaly that the executive’s interpretation had been developed in secret, without the adversarial testing that normally refines statutory meaning.

The word “relevant” does quiet work across many areas of law, which is why the panel’s reading resonated beyond surveillance. In ordinary subpoena practice, relevance connects the thing sought to a specific, existing inquiry: the documents must bear on the investigation at hand, not on investigations that might someday exist. The government’s bulk-collection theory severed that connection, treating relevance as a property of the database rather than of the inquiry. The panel’s rejection restored the ordinary meaning: relevance is relational, and a demand with no pending or authorized investigation to relate to cannot satisfy it. The lesson generalizes. Statutory words do not shed their meaning when the government invokes security; if anything, the stakes make fidelity to the text more important, because the text is the only public constraint on secret action.

Constitutional avoidance did the rest. Having found the program unauthorized by statute, the court declined to decide whether it also violated the First or Fourth Amendment. The doctrine is a familiar exercise of judicial restraint: courts should not decide constitutional questions unnecessarily, both to respect the political branches and to avoid premature pronouncements. In this field, however, avoidance has a cumulative effect worth naming. When courts repeatedly resolve surveillance disputes on statutory grounds, the constitutional questions accumulate without answers. The program is stopped or narrowed, which is a real victory for the challengers, but no precedent emerges about what the Constitution permits. The next program, built under a new statute, starts the constitutional inquiry from scratch. Avoidance is prudent in each case and impoverishing across cases.

The constitutional claims the court declined to reach are worth naming, if only to show what avoidance left undecided. The First Amendment claim alleged that bulk collection chilled associational freedoms, deterring people from contacting organizations, journalists and religious groups for fear their associations would be logged. The Fourth Amendment claim alleged that indiscriminate collection of every call record was an unreasonable search regardless of the third party doctrine. Both claims raised profound questions, and the court’s silence on them means neither has an answer. A future program, authorized by a future statute, could present the same constitutional questions anew, with no precedent to guide the court except the avoidance itself. This is the cost of deciding on the narrowest ground: the ground holds for the case at hand, and the Constitution waits.

Congress responded to the ruling with legislation. The statute that replaced the bulk program, ending collection under section 215 and substituting a targeted system for obtaining call detail records, is described in the companion guide to the USA FREEDOM Act of 2015. The sequence matters for understanding how litigation and legislation interact in this field. The court did not order the program shut down on a particular date or design its replacement; it held that the statute as written did not authorize what the executive was doing, and the political branches then wrote a new statute. Litigation tested the old law’s limits, and legislation redrew them.

Six years after this article’s reference date, the Ninth Circuit confronted the same program in a criminal appeal and reached the same statutory conclusion with a different remedial result. United States v. Moalin, 973 F.3d 977 (Ninth Circuit 2020), decided by a panel led by Judge Berzon, held that the bulk metadata collection exceeded the scope of Congress’s authorization in 50 U.S.C. section 1861 and therefore violated that section of the statute. The panel agreed with the Second Circuit’s reading: bulk collection of every call record was not what the business records provision authorized. But the panel went further in one respect and stopped shorter in another. It went further by stating that the collection “may have violated” the Fourth Amendment, while expressly declining to decide the question. It stopped shorter by refusing to suppress the evidence. Suppression of the metadata, the panel explained, is not clearly contemplated by section 1861, and there is no statutory basis for it; moreover, the wiretap evidence actually used at trial was not tainted by the unlawful metadata collection under fruit-of-the-poisonous-tree analysis, so suppression was not warranted under 50 U.S.C. section 1806(e). The convictions were affirmed.

Moalin deserves careful description because it is easy to misread. It was not a good faith exception case; the panel did not hold that agents acted in reasonable reliance on a defective order and therefore kept the evidence on that ground. It was a case about the limits of the statutory suppression remedy: the violation was real, but the statute did not provide suppression as the remedy for this kind of violation, and the evidence that secured the convictions had an independent source untainted by the unlawful collection. The decision thus illustrates a recurring feature of surveillance litigation. Winning on the merits, showing that the government broke the law, does not automatically produce relief for the challenger. The remedial doctrines have their own thresholds, and a program can be held unlawful without any individual case being undone. For the reader keeping score, the two appellate decisions that reached the merits of bulk collection both found statutory violations, and neither produced a constitutional holding against a program.

The remedial gap Moalin exposes is general, not peculiar to that case. The exclusionary rule exists to deter unlawful searches by denying the government the evidentiary benefit of its wrongdoing. But deterrence works only when the wrongdoing produced the evidence; when the trial proof has an independent source, suppression would punish without deterring, and courts decline to order it. Add to this the statutory limits on suppression remedies, and the result is a field in which unlawfulness and relief routinely diverge. A program can be held to violate the statute while every conviction it touched remains intact. For challengers, this means that winning the legal question is only half the battle; the other half is finding a remedy the law provides. For the public, it means that judicial findings of illegality do not automatically translate into accountability for past collection.

The Second Circuit’s standing analysis in ACLU v. Clapper deserves attention because it shows what separates a justiciable surveillance challenge from Clapper’s failed one. The plaintiffs were telephone subscribers whose records the program collected; the program’s existence was publicly known; the injury was not a feared future interception but an ongoing collection the government had acknowledged in general terms. The chain-of-possibilities problem did not arise, because there was no chain: the collection was happening, the plaintiffs were among those collected from, and the only question was whether the statute authorized it. The contrast is the whole lesson of the standing wall in miniature. Give plaintiffs public knowledge and a concrete stake, and courts reach the merits; withhold either, and the threshold defeats them.

The agreement of the two circuits matters independently. The Second Circuit in 2015 and the Ninth Circuit in 2020, examining the program from different procedural postures, a civil challenge and a criminal appeal, converged on the same statutory reading: section 215 never authorized bulk collection. Cross-circuit agreement of this kind is the closest the federal system comes to settling a question short of Supreme Court review, and no court has disagreed. The statutory question is therefore as settled as it can be without the highest tribunal’s imprimatur. The constitutional question, by contrast, has never been answered by any court, which returns the reader to the field’s defining asymmetry: the unlawful is established, the unconstitutional remains unknown.

The convergence of the two circuits is worth pausing over because the panels reached the same reading from opposite procedural directions. The Second Circuit decided a civil challenge brought by subscribers whose records were being collected; the Ninth Circuit decided a criminal appeal brought by defendants whose prosecution had touched the program. Different parties, different postures, different remedies sought, and the same statutory conclusion: the business records provision never authorized bulk collection. When two panels with different incentives and different questions agree on what a statute means, the agreement carries unusual weight, because it cannot be dismissed as the product of one court’s sympathies. The statutory reading is as settled as interpretation gets without the Supreme Court’s imprimatur, and the settlement was achieved without any court needing to decide what the Constitution requires.

The doctrinal shift

How did an ordinary criminal case do more than any national security challenge?

Four years after this article’s reference date, the Supreme Court held that acquiring historical cell site location records is a Fourth Amendment search requiring a warrant supported by probable cause. The 2018 decision narrowed the third party doctrine that had, since 1979, supplied the constitutional foundation for treating metadata as unprotected, and it did so in an ordinary criminal case.

The third party doctrine began with Smith v. Maryland, 442 U.S. 735 (1979). The Supreme Court held that a telephone user has no reasonable expectation of privacy in the numbers voluntarily dialed and conveyed to the phone company, so obtaining them is not a Fourth Amendment search. Information a person voluntarily turns over to a business, the doctrine holds, carries no reasonable expectation of privacy against the government obtaining it from that business. The logic extended naturally to bank records, and for decades it supplied the constitutional premise for treating communications metadata, the addressing information about who contacted whom and when, as categorically unprotected. If the content of a call required a warrant but the fact of the call did not, then large-scale collection of metadata rested on firm constitutional ground, whatever the statutes said. The doctrine was the quiet engine of the entire field: it explained why the government could argue that sweeping collection programs did not even implicate the Fourth Amendment.

Carpenter v. United States, 585 U.S. 296 (2018), confronted the doctrine with a technology the Smith Court never imagined. In an ordinary criminal prosecution, the government obtained historical cell site location information from the defendant’s wireless carriers under a court order authorized by the Stored Communications Act, issued on a showing below probable cause. The records at issue captured the defendant’s physical movements over an extended period. Chief Justice Roberts, writing for a 5-4 majority, held that acquiring these records is a search within the meaning of the Fourth Amendment and requires a warrant supported by probable cause. “We decline to extend Smith and Miller to cover these novel circumstances,” the opinion stated, and it explained why: an individual “maintains a legitimate expectation of privacy in the record of his physical movements as captured through CSLI.” The third party doctrine’s premise of voluntary conveyance, the majority reasoned, does not fit location records generated automatically whenever a phone connects to the network, a process the user neither meaningfully controls nor can realistically avoid while participating in modern life. A Stored Communications Act order, supported by less than probable cause, is not a permissible mechanism for obtaining what is now recognized as a search.

The dissenters gave the competing view with the care it deserves. The dissenters defended the third party doctrine’s application to records held by the carrier and warned that the majority’s effort to distinguish location data from other voluntarily conveyed information would prove difficult to administer. On their account, the majority had replaced a clear rule with a standard whose boundaries would have to be worked out case by case, leaving law enforcement uncertain about which records require warrants. The disagreement, like the one in Clapper, was genuine and closely divided, and it turned on how much weight to give technological change in applying old doctrines.

Carpenter’s significance for surveillance law is indirect but profound, and it must be stated with precision to avoid overclaiming. The decision did not hold any intelligence program unconstitutional. It did not address section 215, section 702, or any other national security authority. It expressly preserved the national security question: “our opinion does not consider other collection techniques involving foreign affairs or national security.” What it did was narrow the constitutional premise on which the broadest claims for metadata collection rested. If historical location records carry a reasonable expectation of privacy despite being held by a phone company, then the categorical claim that all metadata is unprotected can no longer be sustained, and any future defense of bulk collection programs will have to reckon with Carpenter’s reasoning rather than rely on Smith alone. That is why the namable claim of this article locates the field’s most consequential constitutional development in an ordinary criminal case: the doctrine that did the most work to limit collection was reshaped where the standing wall did not apply, in a prosecution where the defendant plainly had standing to challenge the search of his own records. The carve-out matters equally. By reserving foreign affairs and national security techniques, the Court left the hardest question for another day, and the reservation ensures that Carpenter’s full implications for intelligence collection remain untested in the decisions traced here.

The third party doctrine’s longevity is part of its power. Smith was decided in 1979, when a telephone was a device on a wall and dialing a number was a deliberate act. The doctrine’s premise, that conveying information to a business extinguishes the privacy interest in it, fit the world it was made for. It fit less well with each passing decade, as the volume of information automatically shared with service providers exploded: not just dialed numbers but location data, browsing histories and financial transactions, all generated as the byproduct of ordinary life. Lower courts applied Smith faithfully to new technologies, because faithful application is what lower courts do, and the doctrine’s categorical quality made it a reliable foundation for the government’s broadest claims. By the time Carpenter arrived, the question was whether a doctrine built for an earlier era of telephony could govern the digital exhaust of modern existence.

The majority’s answer turned on the voluntariness premise. Smith’s logic assumed a meaningful choice: the caller chose to convey the number to the company to complete the call. Carpenter’s location records, by contrast, are generated automatically whenever the phone connects to the network, without any affirmative act by the user beyond carrying a powered phone. The opinion reasoned that this automatic generation, combined with the comprehensive and revealing nature of the resulting record, takes the information outside the doctrine’s rationale. An individual maintains a legitimate expectation of privacy in the record of his physical movements as captured through CSLI, the Court held, because that record reveals an intimate window into a person’s life. The doctrine was not overruled; it was confined to circumstances resembling the ones that produced it.

The dissenters’ administrability concern deserves a fair hearing, because it identifies a real cost of the majority’s approach. A categorical rule, that voluntarily shared information is unprotected, is easy to apply: courts need only ask whether the information was shared. The majority’s standard, which asks whether the information is deeply revealing and automatically generated, requires case-by-case judgment about new technologies as they emerge. The dissenters warned that law enforcement would face uncertainty about which records require warrants and which do not, and that courts would be drawn into technology assessments beyond their competence. The majority’s answer, implicit in the opinion, is that the Fourth Amendment has always required such judgments, and that fidelity to its protections matters more than the convenience of a bright line. The 5-4 division reflects a genuine disagreement about how the Constitution should confront technological change.

The national security carve-out is the opinion’s most consequential sentence for this field, and its brevity is deliberate. By stating that the opinion does not consider other collection techniques involving foreign affairs or national security, the Court fenced its holding off from the intelligence domain. The practical effect is that Carpenter’s reasoning, powerful as it is, has no direct application to any program discussed in this article. A future litigant will argue that the logic extends: if historical location records carry a reasonable expectation of privacy in a criminal case, the same records collected for intelligence purposes implicate the same interest. The government will answer that the carve-out, combined with the special needs of national security, distinguishes the contexts. No court in this history has resolved that dispute, because no challenge to an intelligence program has presented it. The carve-out thus functions as another form of the standing wall: a doctrinal boundary that keeps the most important application of a new rule undecided.

The carve-out’s relationship to Keith’s reservation is worth a direct comparison, because the two silences rhyme. In 1972 the Court required warrants for domestic security surveillance and said nothing about foreign intelligence, leaving the hardest question for the political branches. In 2018 the Court required warrants for historical location records and said nothing about foreign affairs or national security techniques, leaving the hardest application for another day. Both reservations were acts of judicial restraint, and both had the same structural effect: the constitutional rule develops fully in the domain the Court decides, while the adjacent domain where the largest programs operate remains governed by older law. Whether Carpenter’s carve-out will prove as generative as Keith’s reservation, producing decades of legislation and litigation around an undecided question, depends on whether any future case forces the Court to cross the boundary it drew.

There is a larger lesson in where Carpenter came from. The decision moved constitutional doctrine more than any national security case has, precisely because it was not a national security case. The defendant had undeniable standing to challenge the search of his own records; no secrecy doctrine barred the courthouse door; no privilege withheld the evidence; the adversarial process functioned normally. Doctrine develops where cases can be litigated, and in surveillance law the litigable cases are overwhelmingly ordinary criminal prosecutions. The field’s constitutional law is therefore being written, case by case, in the criminal docket, while the intelligence programs that raise the hardest questions remain largely beyond judicial examination. That asymmetry is the standing wall’s deepest consequence: not just that particular challenges fail, but that the law itself develops around the programs rather than through them.

What remains of Smith after Carpenter is a narrower doctrine with an uncertain future. The third party doctrine still governs the records it was built for: information a person meaningfully and voluntarily conveys to a business in the course of a discrete transaction. What it no longer governs, after Carpenter, is the comprehensive, automatically generated record of a person’s movements and, by the opinion’s logic, potentially other categories of revealing data created without meaningful choice. The boundary between those categories will be litigated for years, as new collection techniques present new variations on the voluntariness question. Each case will ask whether the information resembles the dialed numbers in Smith or the location records in Carpenter, and the answer will determine whether a warrant is required.

The distinction between the Stored Communications Act order and a warrant is the practical heart of Carpenter. The SCA order was issued on a showing below probable cause, a standard Congress designed for records held by third parties on the assumption, derived from Smith, that such records carry no full Fourth Amendment protection. The warrant requires probable cause, the higher showing the Fourth Amendment demands for a search. By holding that acquiring historical location records is a search, the Court necessarily held that the lower showing is insufficient: what the Constitution calls a search cannot be authorized on less than probable cause. The opinion thus did two things at once. It reclassified the records, moving them from the unprotected category into the protected one, and it reclassified the procedure, demoting the SCA order from a sufficient mechanism to an insufficient one for this kind of data.

The reclassification has a forward-looking edge. Congress wrote the Stored Communications Act on the premise that Smith settled the constitutional question for business records, and the statute’s tiered standards reflect that premise. Carpenter unsettled the premise without rewriting the statute, which means the statutory tiers now rest on a constitutional foundation that has shifted beneath them. Future litigants will argue that other categories of records, held by other providers under the same lower standards, deserve the Carpenter treatment. The government will argue that Carpenter is confined to location data’s unique revelatory power. The statute stands unchanged while its constitutional presuppositions erode, a familiar pattern in this field, where legislation and doctrine move at different speeds.

For the practitioner or the student, Carpenter offers a usable method for analyzing any new collection technique. Ask first whether the information is generated automatically or conveyed by meaningful choice; ask second whether the resulting record is comprehensive and revealing or narrow and transactional; ask third whether the government’s access resembles the targeted acquisition in Smith or the prolonged tracking in Carpenter. The answers will not always be clear, and close cases will divide courts as Carpenter divided the justices. But the questions discipline the analysis, forcing attention to what the technology actually does rather than to the labels the parties attach to it. In a field where programs are described in euphemism and challenged in abstraction, that discipline is valuable.

The opinion’s key sentence, “We decline to extend Smith and Miller to cover these novel circumstances,” is a study in judicial technique. The Court did not overrule the third party doctrine; it declined to extend it, which is a narrower and more durable move. Overruling would have forced the justices to say what replaces the doctrine, a question on which the 5-4 majority might not have agreed. Declining to extend leaves Smith standing for the records it was built for while fencing new technologies outside its reach, and it puts future courts to work case by case rather than handing them a finished replacement. The technique is incrementalism as strategy: move the law where it must move, and leave the rest for the next case. That is why Carpenter narrowed the doctrine without killing it, and why the doctrine’s future will be written one technology at a time.

Secrecy as a defense

What happens when the state secrets privilege meets the statute’s own procedure?

Eight years after this article’s reference date, the Supreme Court held unanimously that FISA’s procedures for handling surveillance evidence do not displace the state secrets privilege. The 2022 decision resolved the interaction between a judge-made secrecy doctrine and the statute’s own secrecy procedures in favor of the doctrine, and then declined to say what the statute’s procedures mean.

The state secrets privilege is a common law evidentiary privilege, recognized by the courts rather than created by Congress, that permits the government to withhold information whose disclosure would harm national security. When the government invokes it, courts typically dismiss the affected claims or exclude the affected evidence, because litigation cannot proceed without the withheld material. The privilege is distinct from standing: standing asks whether the plaintiff is entitled to be in court at all, while the privilege asks whether the case can proceed once the plaintiff is there. In surveillance litigation the two often work in sequence, with secrecy defeating at the threshold those claims that standing does not.

Federal Bureau of Investigation v. Fazaga, 595 U.S. 344 (2022), arose from litigation challenging FBI surveillance activities. The plaintiffs alleged the surveillance was unlawful, and the government invoked the state secrets privilege over much of the relevant material. The plaintiffs pointed to section 1806(f) of FISA, which provides that when the government gives notice of its intent to use FISA-derived evidence, or when an aggrieved person moves to suppress, the court shall review the relevant materials in camera and ex parte to determine the lawfulness of the surveillance. The Ninth Circuit had held that this statutory procedure displaced the privilege: Congress, by creating its own mechanism for judicial review of surveillance materials, had replaced the judge-made privilege in this domain.

Justice Alito’s unanimous opinion, with no dissent and no concurrence, reversed. FISA section 1806(f) does not displace the state secrets privilege, the Court held, because the statute’s text nowhere mentions the privilege, and displacing an established common law privilege requires clear congressional language. Nor is the statutory procedure operationally incompatible with the privilege: a court can conduct in camera, ex parte review of materials under section 1806(f) while still honoring a valid privilege assertion over other materials. Having resolved the displacement question, the Court declined to go further. It did not interpret what section 1806(f)’s procedures require in detail, and it did not decide whether the particular evidence at issue was privileged. It remanded for the lower courts to address those questions.

Fazaga is the secrecy counterpart to Clapper’s standing barrier, and the pairing completes the picture of why merits rulings are rare. Clapper closed the courthouse door to plaintiffs who could not prove they were monitored; Fazaga confirmed that even plaintiffs who get inside may find the evidence they need withheld under a privilege the statute does not displace. The unanimity of the decision is itself significant: unlike the 5-4 divisions in Clapper and Carpenter, the justices spoke with one voice on the privilege question, suggesting that the protection for genuine state secrets commands broader agreement than the standing and Fourth Amendment questions do. For the field, the practical effect is that section 1806(f), the statute’s own mechanism for testing surveillance lawfulness in court, operates alongside rather than instead of the privilege, and challenges that depend on privileged material will continue to fail without any ruling on whether the underlying collection was lawful.

The privilege and the standing doctrine operate as a sequence, and the sequence matters. Standing asks whether the plaintiff belongs in court; the privilege asks whether the lawsuit can proceed once the plaintiff is there. A challenge to a secret program must therefore survive two filters: first the plaintiff must prove injury without access to the classified facts that would prove it, then the plaintiff must prove the claim without access to the classified evidence that would establish it. Clapper shows the first filter; Fazaga shows the second. Together they explain why merits decisions are the rarest events in this field. It is not that judges are unwilling to decide; it is that the doctrines of justiciability and evidence, each defensible on its own terms, combine to make decision nearly impossible.

The unanimity of Fazaga stands in deliberate contrast to the 5-4 divisions in Clapper and Carpenter, and the contrast is informative. The justices divide closely over how much uncertainty standing tolerates and over how the Fourth Amendment applies to new technology, but they spoke with one voice on the privilege question. The common ground appears to be the institutional judgment that courts are poorly positioned to second-guess executive determinations about what disclosures would harm national security, and that an established evidentiary privilege should not be displaced without Congress saying so clearly. Unanimity does not mean the decision was easy; it means the justices converged on the principle that the privilege survives unless Congress clearly abrogates it. For litigants, the message is that the privilege is the most durable of the field’s barriers, the one least likely to yield to creative lawyering.

The Court’s refusal to interpret section 1806(f) or to decide whether the evidence was privileged is as significant as what it held. Having resolved the displacement question, the justices remanded the remaining issues, leaving for another day the meaning of the statute’s in camera, ex parte procedures and their application to the materials at issue. This is judicial minimalism of the same kind seen in the constitutional avoidance of ACLU v. Clapper: decide what is necessary, reserve the rest. But the reservation has a cost. Section 1806(f) is the statute’s own mechanism for judicial review of surveillance lawfulness, the procedure Congress designed to let courts test collection without public disclosure. Until its requirements are authoritatively construed, litigants and lower courts must guess at how it operates alongside the privilege the Supreme Court has now confirmed survives. The mechanism exists; its contours do not.

The privilege Fazaga preserved has its own rationale, distinct from the statute. The common-law state secrets privilege rests on the judgment that some information is so sensitive that its disclosure in litigation would harm the nation, and that the judiciary should not compel disclosure even at the cost of dismissing claims. The rationale is institutional: courts lack the executive’s capacity to assess the full range of risks that disclosure creates, and the adversary process cannot function when one side cannot see the evidence. Critics answer that the privilege is invoked too broadly and shields misconduct; defenders answer that the alternative, disclosure or dismissal, is worse. Fazaga did not resolve that debate. It held only that Congress had not displaced the privilege in the FISA context, leaving the debate, and the privilege, where it found them.

What the remand left open was defined by the opinion itself. The lower courts were left to decide two questions the Supreme Court reserved: what section 1806(f)’s in camera, ex parte procedures actually require, and whether the state secrets privilege properly covers the specific materials at issue. The first question is one of statutory interpretation that the Supreme Court declined to answer: how searching must the judge’s review be, what showing the government must make, and what relief follows if the surveillance was unlawful. The second is intensely fact-specific, turning on the nature of the materials and the plausibility of the claimed harm from disclosure. Fazaga settled the framework; the substance remained to be built by the courts below.

The opinion’s two rationales for preserving the privilege deserve separate attention because they do different work. The clear-statement rule is about institutional respect: an established common-law privilege should not be abrogated by inference, and Congress must speak plainly if it means to displace what the courts have long recognized. The operational-compatibility holding is about practical coexistence: the statute’s in camera, ex parte review and the privilege can function side by side, the first letting a judge test lawfulness privately, the second letting the government withhold what testing would expose. Together they produce a belt-and-suspenders protection for secrecy. Even if a litigant persuades a court that the statute’s procedures should do more work, the privilege remains as the backstop; and even if the privilege claim fails on particular materials, the statutory procedure still keeps the review private. The challenger must defeat both layers, and defeating either one still leaves the other standing.

Step back once more and the full architecture of unreviewability comes into view. A program is created in secret under a statute construed in secret by a court that hears one side. Civil challengers cannot prove they were targeted, so standing defeats them. If they somehow establish standing, the privilege withholds the evidence, so the claim fails without a ruling on lawfulness. Criminal defendants who receive notice can litigate, but notice was not given for years, and even successful challenges to statutory authority do not necessarily produce suppression. At each stage a neutral doctrine, standing, privilege, avoidance, remedial limits, does legitimate work, and the cumulative effect is a field in which the lawfulness of the most consequential programs is the least examined question. That is what it means for a statute to be effectively unreviewable: not that review is forbidden, but that the ordinary mechanisms of review cannot reach it.

The in camera, ex parte procedure at the heart of section 1806(f) deserves a plain description, because its mechanics explain both its promise and its limits. In camera means in the judge’s chambers, outside public view; ex parte means with only one side present, here the government. When a defendant moves to suppress FISA-derived evidence, the court privately reviews the applications, orders and related materials to determine whether the surveillance was lawfully authorized and conducted, without disclosing classified details to the defense. The procedure is Congress’s attempt to square the circle: adversarial testing of lawfulness without public exposure of sources and methods. Its adequacy depends entirely on the judge’s rigor, since no defense counsel sees the materials being tested.

The forum problem

Surveillance law is decided in three very different forums, and the forum often determines the outcome before the legal question is reached. The first forum is the FISA court itself, which hears the government’s applications in secret and, on rare occasions, its appeals. The second is the civil lawsuit, in which a plaintiff challenges a program directly. The third is the criminal prosecution, in which a defendant challenges the lawfulness of collection used against him. Each forum has its own procedural character, and the field’s history can be read as the story of which forum gets to decide.

The secret forum has the advantage of expertise and the disadvantage of one-sidedness. The FISA court’s judges see the applications, the minimization procedures and the compliance record; no outsider sees any of it. When the system works, the judges test the government’s submissions rigorously. When it errs, no adversary corrects the error, and the mistake can persist across years of applications. Sealed Case is the celebrated exception, the one published appellate correction, but its rarity proves the point: a tribunal that hears one side will almost always agree with that side, not from bias but from the absence of counterargument. The adversarial system assumes that truth emerges from contest; the secret forum suspends that assumption by design.

The civil forum has the advantage of openness and the disadvantage of the threshold. Civil plaintiffs litigate in public, with briefing, argument and published opinions, which is why the great landmarks of American constitutional law are civil cases. But civil surveillance plaintiffs must first prove they were monitored, and secrecy makes that proof nearly impossible. Clapper is the demonstration: a fully briefed, fully argued challenge to a major program, resolved without any ruling on the program’s legality because the plaintiffs could not cross the standing threshold. The civil forum is where doctrine would develop most naturally, and it is the forum most reliably closed.

The criminal forum has the advantage of concreteness and the disadvantage of dependence on the government’s choices. A criminal defendant who receives notice of FISA-derived evidence has undeniable standing, access to counsel, and the full adversarial machinery; the court reviews the materials in camera and the defense presses every weakness. Keith reached the merits this way, and Carpenter moved constitutional doctrine this way. But the forum opens only when the government chooses to prosecute and chooses to give notice, and for five years under section 702 it gave no notice at all. A system of judicial review that depends on the executive’s willingness to disclose is review at the executive’s sufferance. The notice correction of October 2013 repaired the mechanism, but the years of silence show how fragile it was.

The forum problem explains the field’s central asymmetry. The constitutional questions that matter most, about intelligence programs conducted in secret, arise in the forums least able to answer them: the secret tribunal that hears one side, and the civil courts that cannot get past the threshold. The questions get answered, if at all, in the criminal forum, which handles ordinary prosecutions rather than programs. Doctrine therefore develops sideways, through cases about ordinary crime, while the programs themselves remain in the shadows. Anyone who wants to understand surveillance law must understand this: the law is not only what the opinions hold, but where the opinions come from, and the where shapes the what.

The three forums also differ in what they produce for the public. The secret forum produces classified orders and, very occasionally, a published opinion like Sealed Case. The civil forum produces public opinions, but only about the threshold, as Clapper demonstrates. The criminal forum produces public opinions about the merits, but only about ordinary prosecutions, as Carpenter demonstrates. No forum reliably produces public opinions about the merits of intelligence programs. That is the information deficit at the heart of the field: the opinions the public most needs are the opinions no forum is positioned to write. Democratic oversight depends on knowing what the law permits, and on this subject the public is asked to trust without knowing.

What review would require

If the standing wall, the privilege and the remedial gap keep programs from being judged, what would it take to make them reviewable? The question has no single answer, but the options clarify the obstacle. The most direct route is official disclosure: when the government publicly acknowledges a program, as it eventually did with bulk metadata collection, plaintiffs gain the concrete stake that standing demands. Disclosure converts speculative fear into demonstrable fact, and the courthouse door opens. But disclosure is the executive’s choice, and the executive discloses what it chooses. A system of review that waits on voluntary transparency is not a system of review at all; it is a system of permission.

A second route runs through the criminal forum: notice to defendants, followed by adversarial litigation of lawfulness. The Clapper majority envisioned this route, and the October 2013 policy correction finally made it real. But the route is narrow by design. It serves only defendants the government chooses to prosecute, only when the evidence is used at trial, and only when notice is given. Most collection never leads to prosecution; most prosecutions never involve the challenged program; and the notice obligation, as the 2008 to 2013 gap showed, is only as reliable as the executive’s interpretation of it. The criminal forum is the best available venue for testing surveillance lawfulness, and it is available only at the government’s discretion.

A third route would be damages actions against officials, which give plaintiffs a concrete injury, money lost, and thus solve the standing problem. But damages actions face their own barriers: qualified immunity shields officials who did not violate clearly established law, and the privilege can still withhold the evidence needed to prove the claim. Congress could create a statutory damages remedy with procedural protections, and the idea has been discussed for years, but no such remedy exists for the programs examined here. The gap between the injury that standing requires and the remedy the law provides remains unfilled.

The honest conclusion is that effective review would require the political branches to want it. Courts cannot manufacture cases; they decide the disputes brought to them. If the executive keeps programs secret and Congress does not compel disclosure or create remedies, the judiciary’s role stays marginal no matter how willing judges are to decide. The standing wall is therefore not only a doctrinal problem but a structural one, reflecting a constitutional allocation in which the courts are the least dangerous branch precisely because they cannot act unprompted. Surveillance law will be reviewable when the political branches make it reviewable, and the history traced here suggests that happens, if at all, only after disclosure forces the issue.

Why the Constitution stays silent

Across the decisions traced here, a striking pattern emerges: courts repeatedly resolve surveillance disputes without deciding what the Constitution permits. The Second Circuit struck down the bulk program on statutory grounds and declined the constitutional claims. The Ninth Circuit found a statutory violation, said the collection may have violated the Fourth Amendment, and decided nothing. Carpenter decided a Fourth Amendment question but fenced intelligence programs outside its holding. Fazaga decided a privilege question and reserved the statute’s meaning. Keith decided the domestic warrant question and reserved the foreign intelligence question. Only Sealed Case reached a constitutional holding about the statute’s architecture, and it did so in a one-sided proceeding where the premise went unchallenged. The constitutional law of surveillance is thus mostly a law of reservations, avoidances and carve-outs.

The avoidance is principled in each instance. Constitutional avoidance is a longstanding doctrine of restraint: courts should not decide constitutional questions when a narrower ground resolves the case, both to respect the separation of powers and to avoid premature rulings on profound questions. Judicial minimalism counsels the same. No one of these decisions can be faulted for declining to go further than necessary. But principles that are sound case by case can produce a troubling pattern case after case. When every court avoids the constitutional question, the constitutional question is never answered, and programs operate for years under a cloud of unresolved legality. The restraint of each judge aggregates into the silence of the judiciary.

The distinction between statutory and constitutional rulings matters for what follows. A statutory holding, that a program exceeded what Congress authorized, can be cured by Congress authorizing it. The bulk program litigation illustrates the point: the courts held the program violated section 215 as written, and Congress wrote a new statute. A constitutional holding, by contrast, binds Congress; no statute can authorize what the Fourth Amendment forbids. By resolving cases on statutory grounds, courts leave the political branches free to reauthorize the same conduct under new language, which is exactly what happened. This is not a criticism of the courts, which were right to decide no more than necessary. It is an observation about the structural consequence: statutory victories are reversible by legislation, and only constitutional rulings entrench limits.

The complication answered

A persistent claim in public discussion holds that the judiciary has reviewed these collection programs and approved them, lending them constitutional legitimacy. The record examined here does not support that claim. Most challenges to the programs were dismissed on standing or other threshold grounds without any court reaching the merits, which means no tribunal ever evaluated whether the programs complied with the Constitution. Dismissal on justiciability grounds is not approval; it is a refusal to decide, and treating it as endorsement misreads what the opinions actually held.

The decisions that did reach the merits point in the opposite direction. The Second Circuit in 2015 and the Ninth Circuit in 2020 each held that the bulk telephony metadata program violated the governing statute, the first on direct review and the second in a criminal appeal. Those are not blessings; they are findings of illegality, limited to statutory grounds by the courts’ own restraint but findings of illegality nonetheless. The absence of adverse constitutional rulings is largely an artifact of justiciability: courts decide constitutional questions only when they must, and in this field they have almost never been forced to, because standing and secrecy so often end cases first. Readers who encounter confident assertions about what the courts have supposedly approved will find those assertions examined and sorted in the companion page on surveillance law myths.

The logic by which non-decisions become misread as approvals deserves a final unpacking. Courts in the American system decide only the disputes properly before them, and a dismissal on threshold grounds says nothing about the merits by design. But those virtues have a cost in the surveillance context: each dismissal adds to a public record in which programs appear to have survived judicial scrutiny, when in fact scrutiny never occurred. The distinction between surviving review and evading it is the difference this article has tried to keep visible.

The confusion is fed by a second source as well: the specialized court that authorizes collection in the first place. The Foreign Intelligence Surveillance Court reviews executive applications and issues orders permitting monitoring, and its approval rate has historically been extremely high. Observers sometimes cite those approvals as evidence that courts have blessed the programs, but the comparison misunderstands the tribunal’s function. The specialized court conducts ex parte review of applications under deferential standards, without an adversary to test the government’s claims. Its orders are authorizations, not adjudications of contested legality, and equating them with merits review confuses permission with vindication. The standing wall ensures that the authorizing tribunal is often the only court that ever examines a program, which makes the confusion easy and the correction important.

There is a final reason worth naming: the asymmetry of precedent. When a court upholds a program on the merits, the government cites the decision as approval. When a court strikes a program down on statutory grounds, the decision settles only the statute. When a court dismisses on standing or secrecy, the decision settles nothing about lawfulness but is sometimes misread as approval. The result is a one-way ratchet in public understanding, in which the absence of adverse rulings is mistaken for the presence of favorable ones. The obstacle ledger at the end of this article is designed as an antidote: for each case, what was decided and on what ground, so that silence is not confused with sanction.

The standing wall, then, is the organizing fact of surveillance litigation. American surveillance law is shaped less by what courts have held than by what they have declined to reach, because secrecy prevents plaintiffs from proving injury, and the most consequential constitutional development in the field arrived through an ordinary criminal case about cell phone records rather than through any challenge to a national security program. For readers who want to work through these materials systematically, the VaultBook legislation study notebook offers a structured way to track holdings, doctrines and open questions across the series.

The rule-of-law implications are worth stating without exaggeration. A legal system in which the executive’s most intrusive powers are the least subject to judicial examination is not lawless; the statutes exist, the courts exist, and the doctrines that limit review are themselves law. But there is a difference between a power that is regulated and a power that is reviewable, and surveillance law tests that difference. Regulation without review depends on the executive’s good faith compliance and on the political branches’ oversight, both of which are real but neither of which substitutes for an independent tribunal saying what the law permits. The article’s claim is not that every program is unlawful; it is that the public would not know if one were, because the mechanisms for finding out so rarely function.

Readers encountering confident public claims about this field should apply the ledger test: was the claim resolved on standing, on secrecy or on the merits? Assertions that courts have approved a program usually dissolve under that question into dismissals that decided nothing about lawfulness. Assertions that a program was held unconstitutional usually turn out to be statutory holdings with the constitutional question avoided or reserved. Precision about what was decided, and what was declined, is the reader’s best defense against both alarmism and complacency, and it is the habit this article has tried to instill.

What the pattern means

The question posed at the outset can now be answered in the reader’s own words. Programs are rarely reviewed on the merits because standing demands proof of injury that secrecy withholds, and because the privilege withholds the evidence even when standing is satisfied. The doctrine that defeats most challenges is the standing wall, with the state secrets privilege as its inner companion. The decisions that reached the merits held, in Keith, that domestic security surveillance requires a warrant; in Sealed Case, that the significant purpose standard is constitutional and the wall was not required; in ACLU v. Clapper and Moalin, that the bulk program violated the statute; and in Fazaga, unanimously, that the statute does not displace the privilege. The constitutional shift that did the most work arrived in Carpenter, through an ordinary criminal case, narrowing the doctrine that had left metadata unprotected while reserving the national security question. That is the field in one paragraph, and the ledger below in one table.

The reader who has followed the argument this far possesses something rare in public discussion of this subject: a precise vocabulary for what courts have and have not done. When the next disclosure arrives, and the next program is debated, the questions to ask are the ones this article has asked throughout. Was the challenge resolved on standing, on secrecy or on the merits? Did the court construe the statute or the Constitution? Did the holding reach the program’s lawfulness, or did it stop at the threshold? The answers will usually show that less was decided than the headlines suggest, and that the standing wall still stands where it has always stood, at the courthouse door.

The series thesis thread runs through every section above, and it can be stated as a pair. Litigation is the stage where a statute’s limits are tested: Keith tested the warrant requirement, Sealed Case tested the purpose standard, the bulk cases tested relevance, Carpenter tested the third party doctrine, Fazaga tested the privilege’s boundary. And a statute can be effectively unreviewable: the tests happen only where the threshold doctrines permit, which is rarely where the programs operate. The pair holds the field’s tension. The law is discovered in court, and the court is often closed. Any honest account of surveillance law must keep both truths in view at once.

The ledger below is more than a summary; it is a tool for reading the news. Each new disclosure, each new filing, each new opinion can be entered in the same columns: what was decided, on what ground, and whether the ground was standing, secrecy or merits. The habit the ledger teaches is the habit the field requires, distinguishing what was held from what was avoided, and approval from silence. Readers who keep the ledger will find that most confident claims about surveillance law fail its test, and that the failures are instructive. The law of this field is written in the negative space around the decisions, in what the courts declined to reach, and the ledger makes that negative space visible.

The obstacle ledger

Case	Year	Barrier or breakthrough	Holding in one sentence	Resolved on standing, secrecy or merits
Clapper v. Amnesty International USA	2013	Barrier	The justices held 5-4 that challengers to the section 702 framework lacked Article III standing because their feared interception rested on a speculative chain of possibilities.	Standing
United States v. United States District Court (Keith)	1972	Breakthrough	The justices held 8-0 that domestic security monitoring requires prior judicial approval while expressly reserving the foreign intelligence question.	Merits
In re Sealed Case	2002	Breakthrough	The review court sustained the significant purpose amendment and struck down the FISA court’s wall restrictions as not required by FISA or the Constitution.	Merits
ACLU v. Clapper	2015	Breakthrough	The Second Circuit held that section 215 had never authorized bulk telephony metadata collection, deciding the case on statutory grounds alone.	Merits
Carpenter v. United States	2018	Breakthrough	The justices held 5-4 that acquiring historical cell site location records is a search requiring a warrant supported by probable cause.	Merits
United States v. Moalin	2020	Breakthrough	The Ninth Circuit held the bulk metadata program violated section 1861 but affirmed the convictions because suppression was not warranted.	Merits
FBI v. Fazaga	2022	Barrier	The justices held unanimously that FISA section 1806(f) does not displace the state secrets privilege.	Secrecy

Frequently Asked Questions

Q: Why is it hard to challenge surveillance law in court?

Challenges founder on two threshold doctrines before any judge reaches the substance. First, standing requires a plaintiff to prove personal injury, and a secret program by design leaves its targets unable to prove they were monitored. In Clapper v. Amnesty International USA (2013), the Supreme Court held 5-4 that challengers to the section 702 framework lacked Article III standing because their feared interception rested on a speculative chain of possibilities rather than certainly impending harm. Second, the state secrets privilege lets the government withhold evidence whose disclosure would harm national security, which can end a case even when the plaintiff gets past standing. The result is structural: secrecy defeats proof of injury at the door and proof of the claim inside, so most programs are never judged on the merits.

Q: What did Clapper v. Amnesty International decide about surveillance law?

Clapper v. Amnesty International USA, 568 U.S. 398 (2013), decided 5-4 with Justice Alito writing for the majority, held that the plaintiffs lacked Article III standing to challenge FISA section 702, codified at 50 U.S.C. section 1881a. The plaintiffs’ first theory, that their communications with foreign contacts would likely be intercepted, failed because the feared interception depended on a highly attenuated chain of possibilities too speculative to count as certainly impending injury in fact. Their second theory, that money spent protecting their communications created injury, failed because a plaintiff cannot manufacture standing through expenditures based on hypothetical harm. Justice Breyer dissented, joined by Justices Ginsburg, Sotomayor and Kagan, arguing the threatened interception was sufficiently realistic. The decision ended the case without adjudicating section 702’s constitutionality.

Q: What did Carpenter v. United States hold about surveillance law?

Four years after this article’s reference date, in Carpenter v. United States, 585 U.S. 296 (2018), the Supreme Court held 5-4 that acquiring historical cell site location information is a Fourth Amendment search requiring a warrant supported by probable cause. Chief Justice Roberts wrote that an individual maintains a legitimate expectation of privacy in the record of his physical movements as captured through CSLI, and that a Stored Communications Act order, which requires less than probable cause, is not a permissible mechanism for obtaining it. The opinion expressly declined to extend the third party doctrine of Smith v. Maryland to these novel circumstances. It also expressly preserved the national security question, stating that the opinion does not consider other collection techniques involving foreign affairs or national security, so it decided no intelligence program’s legality.

Q: What is the third party doctrine in surveillance law?

The third party doctrine holds that a person has no reasonable expectation of privacy in information voluntarily conveyed to a third party, such as a business. It comes from Smith v. Maryland, 442 U.S. 735 (1979), where the Supreme Court held that telephone users have no protected privacy interest in the numbers they dial because they voluntarily turn those numbers over to the phone company. For decades the doctrine supplied the constitutional foundation for treating communications metadata as unprotected: if the addressing information about calls was voluntarily shared with the carrier, the government could obtain it without implicating the Fourth Amendment. Four years after this article’s reference date, Carpenter v. United States (2018) narrowed the doctrine, holding it does not extend to historical cell site location records, while expressly leaving national security collection techniques outside the opinion’s scope.

Q: Did any court rule bulk collection illegal under surveillance law?

Yes, twice, on statutory grounds. In May 2015, the year after this article’s lock date, the Second Circuit held in ACLU v. Clapper, 785 F.3d 787, that FISA section 215, 50 U.S.C. section 1861, had never authorized the NSA’s bulk telephony metadata program, because indiscriminate collection of every call record was not “relevant” to an authorized investigation under section 1861(b)(2)(A). The court decided the case on statutory grounds alone and declined to reach the constitutional claims. Six years after this article’s reference date, the Ninth Circuit agreed in United States v. Moalin, 973 F.3d 977 (2020), holding the bulk metadata collection exceeded Congress’s authorization and violated section 1861, while affirming the convictions because suppression was not warranted. Neither court held any program unconstitutional.

Q: What is the state secrets privilege in surveillance law cases?

The state secrets privilege is a judge-made evidentiary privilege that lets the government withhold information whose disclosure would harm national security. When invoked, courts typically exclude the affected evidence or dismiss the affected claims, because litigation cannot proceed without the withheld material. Eight years after this article’s reference date, the Supreme Court addressed its interaction with FISA in FBI v. Fazaga, 595 U.S. 344 (2022), holding unanimously that FISA section 1806(f), which provides for in camera, ex parte judicial review of surveillance materials, does not displace the privilege, since the statute never mentions it and displacement requires clear language. The privilege is distinct from standing: standing asks whether the plaintiff may be in court, while the privilege asks whether the case can proceed once there.

Q: Has anyone won a surveillance law case against the government?

Plaintiffs have won significant rulings, though victories rarely produce individual relief. In May 2015, the year after this article’s lock date, the ACLU won a Second Circuit holding that the bulk telephony metadata program had never been authorized by section 215, decided on statutory grounds. Six years after this article’s reference date, the Ninth Circuit held in United States v. Moalin that the same program violated the statute, but affirmed the convictions because suppression was not warranted and the trial evidence was untainted. Criminal defendants who receive notice of FISA-derived evidence can and do litigate its lawfulness at trial under section 1806 procedures. But civil challengers seeking to strike down programs have mostly been turned away on standing or secrecy grounds without a ruling on the merits, which is why the field’s wins cluster around statutory interpretation rather than constitutional invalidation.

Q: What did the Keith case decide about surveillance law?

United States v. United States District Court, 407 U.S. 297 (1972), decided 8-0 with Justice Powell writing and Justice Rehnquist not participating, held that prior judicial approval is required for domestic security surveillance, at 407 U.S. 324. The government had argued that warrantless wiretaps authorized by the Attorney General were permissible when national security was invoked; the Court rejected that position, holding that the Fourth Amendment’s warrant requirement applies even to domestic security investigations, though the procedures could be adapted to that context. The opinion expressly reserved the foreign intelligence question, stating at 407 U.S. 321 to 322 that the Court expressed no opinion on the issues involved with respect to activities of foreign powers or their agents. Congress later filled that reserved space by enacting the Foreign Intelligence Surveillance Act of 1978.

Q: What does the “speculative chain of possibilities” phrase mean in Clapper v. Amnesty International?

The phrase describes why the Clapper plaintiffs’ feared injury was too uncertain to support standing. Justice Alito’s majority opinion reasoned that for the plaintiffs’ communications to be intercepted under section 702, a chain of contingencies would all have to occur: the government would have to target their foreign contacts, choose section 702 authority rather than another source, succeed in intercepting the communications, and do so in a way that swept in the plaintiffs. Each link was possible, but the chain as a whole was too speculative to constitute certainly impending injury in fact, the standard Article III requires. The phrase matters because it captures the structural mismatch: a secret program’s targets can describe exactly how they might be monitored, yet still fail the certainty test, because the program’s secrecy keeps every link in the chain unprovable.

Q: What did In re Sealed Case dismantle in surveillance law?

In re Sealed Case, 310 F.3d 717 (Foreign Intelligence Surveillance Court of Review 2002), dismantled the wall separating intelligence gathering from criminal prosecution. The wall was a set of practices and court-imposed procedures that kept intelligence officials and criminal prosecutors apart, rooted in the old requirement that foreign intelligence be the primary purpose of FISA surveillance. The USA PATRIOT Act’s section 218 had changed the certification to a significant purpose standard, and the review court, in its first published opinion, sustained that amendment as constitutional and reversed the FISA court’s May 17, 2002 restrictions, holding they were not required by FISA or the Constitution. The decision permitted the coordination between intelligence and law enforcement that the amendment envisioned and remains the leading judicial treatment of the statute’s architecture.

Q: Why did the Second Circuit refuse to decide the constitutional claims in ACLU v. Clapper?

In May 2015, the year after this article’s lock date, the Second Circuit decided ACLU v. Clapper, 785 F.3d 787, on statutory grounds alone and declined to reach the First and Fourth Amendment claims under the doctrine of constitutional avoidance. That doctrine counsels courts to resolve cases on the narrowest available ground and to avoid constitutional questions when a non-constitutional basis decides the case. Because the panel held that section 215 had never authorized the bulk telephony metadata program, there was no need to decide whether the program also violated the Constitution. The restraint is characteristic of the field: even when courts reach the merits, they often stop at statutory interpretation, which means constitutional questions about collection programs remain undecided long after the programs themselves have been judged unlawful.

Q: What did United States v. Moalin hold about suppression of surveillance evidence?

Six years after this article’s reference date, the Ninth Circuit held in United States v. Moalin, 973 F.3d 977 (2020), that the bulk metadata collection violated FISA section 1861 but declined to suppress any evidence. Suppression of the metadata, the panel explained, is not clearly contemplated by section 1861, and there is no statutory basis for ordering it. The panel further held that the FISA wiretap evidence actually used at trial was not tainted by the unlawful metadata collection under fruit-of-the-poisonous-tree analysis, so suppression was not warranted under 50 U.S.C. section 1806(e), and it affirmed the convictions. The panel said the collection may have violated the Fourth Amendment but did not decide the question. The ruling was not a good faith exception case; it turned on the limits of the statutory suppression remedy and the independent source of the trial evidence.

Q: What is the FISA Court of Review and when did it first publish an opinion?

The Foreign Intelligence Surveillance Court of Review is the appellate tribunal above the Foreign Intelligence Surveillance Court, composed of three federal judges designated by the Chief Justice. It hears appeals when the government challenges a FISA court’s denial or restriction, proceedings that occur in secret with only the government present. For the first twenty-four years of the statute’s life it issued no published opinion, because the government almost never lost below. Its first published opinion was In re Sealed Case, 310 F.3d 717, decided November 18, 2002, a per curiam decision that sustained the PATRIOT Act’s significant purpose amendment and struck down the FISA court’s wall restrictions. The court’s published docket remains tiny, which means the most important judicial statements about the statute’s structure come from a tribunal that hears only one side.

Q: Why do statutory violations in surveillance cases rarely lead to suppressed evidence?

Suppression depends on the remedy the statute provides and on whether the tainted evidence led to the proof used at trial. In United States v. Moalin, decided six years after this article’s reference date, the Ninth Circuit held the bulk metadata program violated section 1861 yet refused suppression, because suppression is not clearly contemplated by that section and no statutory basis for it exists, and because the wiretap evidence at trial was not tainted by the unlawful collection under fruit-of-the-poisonous-tree analysis. More generally, the exclusionary rule deters unlawful searches by removing their evidentiary payoff, but it only helps a defendant when the challenged collection produced or tainted the evidence against him. A program can therefore be held unlawful in the abstract while every individual conviction stands, which is why merits victories in this field seldom translate into relief for particular challengers.

Q: What is the difference between standing and the merits in a surveillance challenge?

Standing is the threshold question of whether the plaintiff is entitled to be in court at all; the merits are the question of whether the challenged program is lawful. Article III requires a plaintiff to show a concrete and particularized injury in fact, causation and redressability before a federal court may decide anything else. In Clapper v. Amnesty International USA (2013), the Supreme Court held 5-4 that the plaintiffs lacked standing to challenge section 702, so the case ended without any ruling on whether the program comports with the Fourth Amendment. A merits decision, by contrast, answers the lawfulness question, as the Second Circuit did in May 2015, the year after this article’s lock date, when it held the bulk metadata program violated section 215. The distinction explains the field’s central frustration: most surveillance cases are resolved at the threshold, producing no precedent about the programs themselves.

Q: How did Smith v. Maryland shape surveillance law before Carpenter?

Smith v. Maryland, 442 U.S. 735 (1979), established the third party doctrine for communications metadata: because telephone users voluntarily convey dialed numbers to the phone company, they retain no reasonable expectation of privacy in those numbers, so the government may obtain them without a Fourth Amendment search occurring. For nearly four decades this holding supplied the constitutional foundation for treating metadata as categorically unprotected, which in turn supported the broadest legal claims for bulk collection programs. If addressing information carried no privacy interest, large-scale metadata collection did not even implicate the Fourth Amendment, whatever the statutes provided. Four years after this article’s reference date, Carpenter v. United States (2018) narrowed the doctrine by holding it does not extend to historical cell site location records, while expressly reserving national security techniques.

Q: What role does section 1806(f) of FISA play in surveillance litigation?

Section 1806(f) of FISA provides the in-court mechanism for testing the lawfulness of FISA surveillance. When the government gives notice of its intent to use FISA-derived evidence, or when an aggrieved person moves to suppress, the court reviews the relevant applications, orders and related materials in camera and ex parte, meaning privately and with only the government present, to determine whether the surveillance was lawfully authorized and conducted. Eight years after this article’s reference date, the Supreme Court held unanimously in FBI v. Fazaga, 595 U.S. 344 (2022), that this procedure does not displace the state secrets privilege, because the statute never mentions the privilege and displacement requires clear language. The Court declined to interpret the provision’s detailed requirements or to decide whether particular evidence was privileged, remanding those questions, so the provision’s full scope remains to be worked out.

Q: Can a criminal defendant challenge the lawfulness of FISA surveillance at trial?

Yes. When the government intends to use evidence obtained or derived from FISA surveillance in a criminal prosecution, it must give the defendant advance notice, and the defendant may then move to suppress on the ground that the surveillance was unlawful. The court reviews the FISA applications and orders in camera and ex parte under section 1806(f) to determine lawfulness. This criminal forum was the safety valve the Clapper majority described in 2013: civil plaintiffs who cannot prove they were monitored cannot establish standing, but a defendant who receives formal notice of FISA-derived evidence plainly can litigate the question. The mechanism only works, however, when notice is actually given. From 2008 to 2013 the Justice Department gave zero section 702 notices because of its narrow reading of what counts as derived evidence, a practice it corrected beginning in October 2013.

Q: What changed after the government corrected its notice policy on section 702 evidence?

Beginning in October 2013, the Justice Department filed section 702 notices in criminal cases, starting with United States v. Muhtorov in Colorado and United States v. Mohamud in Oregon, belatedly acknowledging section 702-derived evidence it had previously not disclosed. The shift followed the revelation that the National Security Division had given zero such notices from 2008 to 2013 under an extremely narrow reading of which evidence counts as obtained or derived from section 702 collection. The department revised its policy so that defendants would be notified going forward, and six defendants received belated notice between October 2013 and April 2014. On November 21, 2013, Senators Udall, Wyden and Heinrich wrote to the Solicitor General saying the government had misrepresented its policy and practice to the Supreme Court in Clapper. The correction opened the criminal forum the Clapper majority had envisioned, though the Clapper holding itself stands unchanged.

Q: Why does Carpenter’s national security carve-out matter for surveillance law?

Four years after this article’s reference date, Carpenter v. United States, 585 U.S. 296 (2018), expressly stated that the opinion does not consider other collection techniques involving foreign affairs or national security. The carve-out matters because it confines the decision’s holding to ordinary criminal investigations, leaving the constitutional status of intelligence collection programs undecided. Carpenter narrowed the third party doctrine that had underwritten the broadest claims for metadata collection, which weakens any future argument that bulk programs raise no Fourth Amendment issue at all, but the reservation means no court has applied Carpenter’s reasoning to a national security program. The result captures the article’s central claim: the most consequential constitutional development in the field arrived through an ordinary criminal case about cell phone records, and its implications for intelligence collection remain untested because challenges to those programs so rarely reach the merits.