The analysis exactly as it was written, reviewed and stored. The tabs before this one re-cut this same text into sections; the writer refuses to publish a re-cut that does not reproduce it character for character, so nothing here is added, dropped or reworded.
WHAT IS AT STAKE
Policing is where most people actually meet the Constitution. Not in a courtroom and not in a statute, but on a roadside at night, at a front door, in a holding cell, in a municipal court that sits three evenings a week. Four amendments do the work: the Fourth on stops, searches, arrests and force; the Fifth and Fourteenth on interrogation, due process and the money a court may take; the Sixth on counsel; the Eighth on bail, fines and punishment. Each is settled in outline and contested exactly where it matters most.
This page holds two things. The first is that constitutional questions in criminal justice are decided far less often by what a right means than by whether anyone can enforce it. Immunity, standing, causation and the shrinking exclusionary rule do more work in practice than the amendments they surround, and a reader who only learns the rights has learned the smaller half of the subject. The second is StumpWatch's standing position: when a government hands a policing, custodial, adjudicative or supervisory function to a private company, the Constitution goes with it. A private prison, a contract medical provider, a probation company collecting a court's debt, a vendor answering a police query — each is doing something the government could not do free of constitutional limits, and those limits do not evaporate because the payroll changed.
What follows keeps three things apart: what courts have actually held, what is genuinely open right now, and what we believe the law should be. The third is a position and is labelled as one. On several questions here current law runs against us, and we say so plainly rather than blurring the difference. Where this analysis could not establish something — a current participation rate, an appellate posture, an empirical frequency — it says so in the place where the claim would otherwise sit.
One convention has to be stated first, because an earlier draft of this page got it wrong fifteen times over. A Supreme Court opinion is preceded by a headnote written by the Court's Reporter of Decisions, and that headnote carries its own caveat on its face: it "constitutes no part of the opinion of the Court but has been prepared by the Reporter of Decisions for the convenience of the reader." Fifteen quotations on this page, across fourteen cases, were introduced as what a court "held" or as the Court's own words when they were in fact the Reporter's syllabus: Whren, Atwater, Herring, Castle Rock, Hudson, Timbs, Grants Pass, Bearden, Rodriguez, Connick, Culley, Ward, Glover, and Filarsky twice. In fourteen of the fifteen the substance was right and only the attribution was wrong. In the fifteenth, Herring v. United States, the syllabus is materially narrower than the opinion, and the correction runs in this page's favour rather than against it, which is exactly why it is set out below rather than quietly banked. Each of the fifteen is now either replaced with the opinion's own sentence or labelled as the Reporter's summary of the holding. The check that found them classified every page of every primary document by its running head, and it was run over every case quotation on this page, not only the ones already suspected. Where a quotation on this page is the syllabus, it says so.
WHAT THE LAW ESTABLISHES TODAY: THE STOP, THE ARREST, THE FORCE
An officer may stop and briefly detain a person on reasonable suspicion of crime, and may frisk for weapons on reasonable suspicion that the person is armed and dangerous. Terry v. Ohio, 392 U.S. 1 (1968). That is less than probable cause, and it is the doctrinal foundation of ordinary street policing.
A traffic stop requires no more than probable cause that a traffic law was broken, and the officer's real reason for making it is irrelevant. Whren v. United States, 517 U.S. 806 (1996), decided it in two sentences: "Here the District Court found that the officers had probable cause to believe that petitioners had violated the traffic code. That rendered the stop reasonable under the Fourth Amendment, the evidence thereby discovered admissible, and the upholding of the convictions by the Court of Appeals for the District of Columbia Circuit correct." 517 U.S. at 819. The Reporter's syllabus states the holding in the wider, rule-like form this page previously quoted as though it were the Court's: "[t]he temporary detention of a motorist upon probable cause to believe that he has violated the traffic laws does not violate the Fourth Amendment's prohibition against unreasonable seizures, even if a reasonable officer would not have stopped the motorist absent some additional law enforcement objective." Set against a vehicle code long enough that almost any car can be stopped lawfully, that rule reaches very far. It was unanimous and, as federal Fourth Amendment law, it has not been disturbed.
Three limits belong with it, and the draft of this page omitted all three. First, Whren did not say the Constitution is indifferent to racially selective policing; it said the objection belongs elsewhere. The Court's own rationale, at 517 U.S. at 813, is that "[w]e think these cases foreclose any argument that the constitutional reasonableness of traffic stops depends on the actual motivations of the individual officers involved. We of course agree with petitioners that the Constitution prohibits selective enforcement of the law based on considerations such as race. But the constitutional basis for objecting to intentionally discriminatory application of laws is the Equal Protection Clause, not the Fourth Amendment. Subjective intentions play no role in ordinary, probable-cause Fourth Amendment analysis." That is the affirmative case for the rule, not just the rule; a page critical of Whren has to answer both the actual-motivations sentence and the subjective-intentions sentence rather than only its result. That is a real avenue and a demanding one, because it requires proof of discriminatory purpose. Second, probable cause is not the floor in practice: an investigative traffic stop may rest on reasonable suspicion, and Kansas v. Glover, No. 18-556 (U.S. Apr. 6, 2020), upheld a stop made after an officer ran a plate and learned the registered owner's licence was revoked. The condition this page previously quoted as the Court's — "[w]hen the officer lacks information negating an inference that the owner is driving the vehicle" — is the Reporter's syllabus. Third, a stop survives a reasonable mistake of law as well as a reasonable mistake of fact. Heien v. North Carolina, 574 U.S. 54 (2014). And several state supreme courts have rejected Whren's pretext rule under their own state constitutions; this analysis did not survey them and names none.
The stop has a time limit, but not much of one. Rodriguez v. United States, 575 U.S. 348 (2015), held in its own words that "a police stop exceeding the time needed to handle the matter for which the stop was made violates the Constitution's shield against unreasonable seizures," 575 U.S. at 350, because authority for the seizure "ends when tasks tied to the traffic infraction are — or reasonably should have been — completed," id. at 354. The dog-sniff formulation this page previously quoted as the holding — "[a]bsent reasonable suspicion, police extension of a traffic stop in order to conduct a dog sniff violates the Constitution's shield against unreasonable seizures" — is the Reporter's syllabus. The second quotation above is the Court's; the first was not.
An arrest may follow even for an offence carrying no possible jail term. Atwater v. City of Lago Vista, 532 U.S. 318 (2001): "If an officer has probable cause to believe that an individual has committed even a very minor criminal offense in his presence, he may, without violating the Fourth Amendment, arrest the offender." 532 U.S. at 354. The Reporter's syllabus compresses that to "[t]he Fourth Amendment does not forbid a warrantless arrest for a minor criminal offense, such as a misdemeanor seatbelt violation punishable only by a fine," and it was the syllabus, not the Court, that an earlier draft of this page quoted. Gail Atwater was handcuffed and booked over a seatbelt. The Court applied the Fourth Amendment and held the arrest reasonable; it did not hold the Amendment inapplicable, and the manner in which an arrest is carried out remains reviewable.
Force is judged by objective reasonableness from the officer's viewpoint: the reasonableness of a particular use of force "must be judged from the perspective of a reasonable officer on the scene, rather than with the 20/20 vision of hindsight." Graham v. Connor, 490 U.S. 386, 396 (1989). An earlier draft of this page rendered that as "at the scene", which is not what the opinion says. Deadly force against a fleeing suspect requires probable cause that the suspect poses a significant threat of death or serious physical injury. Tennessee v. Garner, 471 U.S. 1 (1985). Garner is an application of the reasonableness test and not a checklist: "Garner did not establish a magical on/off switch that triggers rigid preconditions whenever an officer's actions constitute 'deadly force.'" Scott v. Harris, 550 U.S. 372 (2007).
The most recent change is narrow but real. In Barnes v. Felix, 605 U.S. ___ (2025), decided May 15, 2025, a unanimous Court rejected the Fifth Circuit's moment-of-threat rule, under which a court looked only to the circumstances at the precise time the officer perceived the threat. Justice Kagan wrote that the totality-of-the-circumstances inquiry "has no time limit," that "earlier facts and circumstances may bear on how a reasonable officer would have understood and responded to later ones," and that "[a] court deciding a use-of-force case cannot review the totality of the circumstances if it has put on chronological blinders." The judgment was vacated and remanded. Ashtian Barnes had been pulled over on outstanding toll violations; two seconds passed between the officer stepping onto the doorsill of the moving car and the first shot.
Three limits belong with that, and the third is the one a reader is most likely to miss. The Court expressly did not address "whether or how an officer's own 'creation of a dangerous situation' factors into the reasonableness analysis," because the courts below never confronted it and it was not the basis of the petition. Justice Kavanaugh, joined by Justices Thomas, Alito and Barrett — four Justices in all — wrote separately on the dangers of traffic stops, observing that even for routine violations stops are "fraught with danger to police officers" — words he takes from Michigan v. Long, 463 U.S. 1032, 1047 (1983), rather than his own — and adding, in his own, that "[o]fficers cannot let their guard down and assume that any particular traffic stop will be safe." And the case the Court chose to illustrate its own rule cuts against the assumption that a wider frame helps plaintiffs: Plumhoff v. Rickard, 572 U.S. 765 (2014), which the Court said "well illustrates the point," is a case in which what had happened in the preceding minutes is what made the shooting reasonable. Barnes widens what a court may look at. It does not change what the court is looking for, and Graham's deference to the officer's perspective is untouched.
The officer-safety line the concurrence rests on has a doctrinal chain worth setting out for the reader, because a page critical of the direction current doctrine takes owes the government's best rationale for that direction — the Court's own words rather than a paraphrase. Terry itself frames the exception this way: "there must be a narrowly drawn authority to permit a reasonable search for weapons for the protection of the police officer, where he has reason to believe that he is dealing with an armed and dangerous individual, regardless of whether he has probable cause to arrest the individual for a crime." 392 U.S. at 27. Michigan v. Long, which is the underlying source of the "fraught with danger" line Justice Kavanaugh takes through his Barnes concurrence, generalises that rationale to the vehicle setting: "In two cases in which we applied Terry to specific factual situations, we recognized that investigative detentions involving suspects in vehicles are especially fraught with danger to police officers," 463 U.S. 1032, 1047 (1983), citing Pennsylvania v. Mimms, 434 U.S. 106 (1977). That is the case for the doctrine that lets an officer treat a routine stop as potentially armed; the correction here cuts against the page's own argument, and we say so rather than only banking the qualifier.
WHAT THE LAW ESTABLISHES TODAY: THE COURT, THE JAIL, THE CELL
An indigent defendant charged with a felony has a right to appointed counsel. Gideon v. Wainwright, 372 U.S. 335 (1963). For lesser offences the right is tied not to the charge but to the sentence: Argersinger v. Hamlin, 407 U.S. 25 (1972), and Scott v. Illinois, 440 U.S. 367 (1979), condition it on imprisonment actually being imposed, and Alabama v. Shelton, 535 U.S. 654 (2002), extends it to a suspended sentence. That distinction is not academic on this page. It is the mechanism by which uncounselled misdemeanour convictions accumulate in the kind of court described further down.
The prosecution must disclose material exculpatory evidence. Brady v. Maryland, 373 U.S. 83 (1963). A conviction is set aside for bad lawyering only on the two-part showing of Strickland v. Washington, 466 U.S. 668 (1984), requiring both deficient performance and prejudice — a standard written to be deferential, and which is. Prejudice is presumed only in the narrow circumstances of United States v. Cronic, 466 U.S. 648 (1984), and an actual conflict of interest is governed by Cuyler v. Sullivan, 446 U.S. 335 (1980).
Pretrial detention without bail is constitutional where the statute is carefully limited. United States v. Salerno, 481 U.S. 739 (1987). Conditions imposed on a person not yet convicted are measured by whether they amount to punishment. Bell v. Wolfish, 441 U.S. 520 (1979). Force against a pretrial detainee is judged objectively: Kingsley v. Hendrickson, 576 U.S. 389 (2015), held that "a pretrial detainee must show only that the force purposely or knowingly used against him was objectively unreasonable," without proof of the officer's subjective state of mind.
For a convicted prisoner the bar is higher. Deliberate indifference to serious medical needs violates the Eighth Amendment, Estelle v. Gamble, 429 U.S. 97 (1976), but Farmer v. Brennan, 511 U.S. 825 (1994), defined deliberate indifference subjectively: the official must actually have known of and disregarded a substantial risk. The draft of this page glossed that as meaning a prison is not liable for a danger nobody looked at, and Farmer says the opposite in terms. A factfinder "may conclude that a prison official knew of a substantial risk from the very fact that the risk was obvious," 511 U.S. at 842, and an official does not escape liability where the evidence shows "that he merely refused to verify underlying facts that he strongly suspected to be true." The standard is subjective; wilful blindness is not a way through it.
On punishment itself, the Court in 2024 declined to constitutionalise a limit the Ninth Circuit had recognised. City of Grants Pass v. Johnson, 603 U.S. 520 (2024), rejected an Eighth Amendment challenge to camping ordinances. The Reporter's syllabus states the holding as "[t]he enforcement of generally applicable laws regulating camping on public property does not constitute 'cruel and unusual punishment' prohibited by the Eighth Amendment," and that syllabus sentence is what this page previously quoted as the Court's. The reasoning is in the opinion, and it too needs its provenance stated: the Cruel and Unusual Punishments Clause, the Court wrote, quoting the plurality in Powell v. Texas, 392 U.S. 514, 531 to 532 (1968), "has always been considered, and properly so, to be directed at the method or kind of punishment" a government may impose for violating criminal statutes. 603 U.S. at 542. The sentence is the Grants Pass Court adopting Powell's words, not writing its own. Whatever one thinks of the result, the mechanism travels: on that reading the Eighth Amendment polices the punishment and not the decision to criminalise the conduct.
And a rule most people believe is a right is not one for damages purposes. Vega v. Tekoh, 597 U.S. 134 (2022), held that a violation of the Miranda rules does not provide a basis for a claim under 42 U.S.C. 1983, because Miranda imposed "prophylactic rules" rather than defining the Fifth Amendment right itself. An unwarned statement may still be suppressed in the prosecution's case-in-chief — though not necessarily for impeachment, and not necessarily as to physical evidence derived from it. The officer who took it cannot be sued for taking it.
THE REMEDY IS THE WHOLE BALLGAME
Whether a right exists and whether anyone can do anything about it are different questions, and in this field the second decides a great many cases. Five doctrines account for most of it, and each defeats a claim independently of the merits.
Qualified immunity. An official sued for damages is protected unless the right he violated was clearly established at the time. Harlow v. Fitzgerald, 457 U.S. 800 (1982). Pearson v. Callahan, 555 U.S. 223 (2009), then made the order of analysis discretionary, so a court may hold that no prior case put the officer on notice without ever deciding whether the conduct was in fact unconstitutional. That much is holding. The consequence we draw from it is an inference, and we mark it as one: where courts skip the merits, the stock of clearly established law grows more slowly than the conduct it is meant to govern, and the next officer to do the same thing is also without notice. The Supreme Court has continued to apply the doctrine, granting immunity in some cases and denying it in others; what it has not done is reconsider the doctrine's existence.
The Court's own case for qualified immunity, from Justice Powell's opinion in Harlow itself, belongs on the page a reader has to weigh against everything argued above. Powell wrote that "claims frequently run against the innocent as well as the guilty—at a cost not only to the defendant officials, but to society as a whole. These social costs include the expenses of litigation, the diversion of official energy from pressing public issues, and the deterrence of able citizens from acceptance of public office. Finally, there is the danger that fear of being sued will 'dampen the ardor of all but the most resolute, or the most irresponsible [public officials], in the unflinching discharge of their duties.'" 457 U.S. at 814. He continued: "Reliance on the objective reasonableness of an official's conduct, as measured by reference to clearly established law, should avoid excessive disruption of government and permit the resolution of many insubstantial claims on summary judgment." 457 U.S. at 818. That rationale is not answered by pointing to the mismatch it creates between conduct and clearly established law: Harlow accepts that mismatch as the price of keeping the discharge of official duties from being chilled. This cuts against the page—the correction sets out the government's affirmative case where the page had only the criticism of its result—and it belongs in the text the reader sees, not only in a reviewer's notes. What remains is a live disagreement about the price, which is a policy question, not a legal one.
Municipal liability. A city is not liable for its officers' violations on a respondeat superior theory; the plaintiff must trace the injury to an official policy or custom. Monell v. Department of Social Services, 436 U.S. 658 (1978). Failure to train qualifies only where it amounts to deliberate indifference, City of Canton v. Harris, 489 U.S. 378 (1989), and Connick v. Thompson, 563 U.S. 51 (2011), put the question and answered it in its own voice: "We granted certiorari to decide whether a district attorney's office may be held liable under [section] 1983 for failure to train based on a single Brady violation. We hold that it cannot." 563 U.S. at 54. The spliced sentence this page previously quoted as the holding — "[a] district attorney's office may not be held liable" for failure to train "based on a single Brady violation" — is the Reporter's syllabus. The Court's own defence of the "policy or custom" gate rests first on statutory text and second on the legislative history of the rejected Sherman Amendment: Monell reads section 1983's "subjects, or causes to be subjected" as language that "cannot be easily read to impose liability vicariously" on a governing body for its employees' torts, 436 U.S. at 692, and treats Congress's rejection of the Sherman Amendment — a proposal that would have imposed a form of municipal liability for constitutional torts — as confirming that a federal law of respondeat superior was refused. The Court then adds an independent third rationale at 436 U.S. 693, flagged in the opinion with the transition "Equally important": that "creation of a federal law of respondeat superior would have raised all the constitutional problems associated with the obligation to keep the peace, an obligation Congress chose not to impose because it thought imposition of such an obligation unconstitutional." That is a discrete federalism / constitutional-avoidance argument, not a facet of the Sherman Amendment history that precedes it. A page arguing for a broader municipal remedy has to answer all three strands — the statutory-text reading at 436 U.S. 692, the Sherman Amendment legislative history, and the keep-the-peace federalism concern at 436 U.S. 693 — rather than only whichever one best fits the criticism.
Federal officers. Against a federal officer there is no statutory constitutional-tort action. The damages action for a constitutional violation is judge-made, and the Court has stopped making it. Egbert v. Boule, 596 U.S. 482 (2022), held that a court "faces only one question: whether there is any rational reason (even one) to think that Congress is better suited to 'weigh the costs and benefits of allowing a damages action to proceed,'" and that if there is even the "potential" for harmful consequences "a court cannot afford a plaintiff a Bivens remedy." In practice that is a rule of no. What does exist is a tort remedy against the United States, not a constitutional one against the officer: the Federal Tort Claims Act's law-enforcement proviso, 28 U.S.C. 2680(h), makes the Act apply to claims "arising, on or after the date of the enactment of this proviso, out of assault, battery, false imprisonment, false arrest, abuse of process, or malicious prosecution" by "investigative or law enforcement officers of the United States Government." That is state tort law applied against the government, decided under the Federal Tort Claims Act's own limits and bars. No constitutional question is asked or answered in it. Congress made that channel deliberately: the 1974 amendment to the law-enforcement proviso was enacted in direct response to Bivens-era federal-officer misconduct — so treating the contraction of Bivens as an unfilled hole ignores that the tort side is where Congress placed the remedy. A reader weighing the erosion of Bivens has to reckon with that choice rather than treat state tort law as a consolation prize.
Injunctions. Adolph Lyons, who had been choked unconscious by Los Angeles police, could not enjoin the chokehold practice because he could not show he would be subjected to it again. City of Los Angeles v. Lyons, 461 U.S. 95 (1983). A practice that harms very many people once and no identifiable person twice is close to unenjoinable at the suit of an individual plaintiff. The route around Lyons is a certified class rather than a single victim, and it works: in Floyd v. City of New York, 959 F. Supp. 2d 540 (S.D.N.Y. 2013), the court found the New York Police Department's stop-and-frisk practices violated the Fourth Amendment and the Equal Protection Clause, and in a companion remedial opinion issued the same day appointed an independent monitor, 959 F. Supp. 2d 668 (S.D.N.Y. 2013).
Convictions. A damages claim that would necessarily imply the invalidity of an outstanding conviction is barred until the conviction is set aside. Heck v. Humphrey, 512 U.S. 477 (1994).
None of these is a fringe doctrine, and together they are why "the police violated my rights" so often ends without a court reaching the question. How often is an empirical question. This analysis does not answer it, the scholarship on qualified immunity's practical effect runs in both directions, and we did not survey it. The claim here is about the structure of the doctrines, not about their frequency.
THE EXCLUSIONARY RULE IS NOW A DETERRENCE CALCULATION
Suppression is not treated as a constitutional right. The Court describes it as a remedy applied only where its deterrent value outweighs its costs, and it has narrowed on that logic for two decades.
Hudson v. Michigan, 547 U.S. 586 (2006), reasoned that "[s]ince the interests that were violated in this case have nothing to do with the seizure of the evidence, the exclusionary rule is inapplicable." 547 U.S. at 594. The general form this page previously quoted as the holding — that violation of the knock-and-announce rule "does not require suppression of evidence found in a search" — is the Reporter's syllabus. Herring v. United States, 555 U.S. 135 (2009), held that "when police mistakes are the result of negligence such as that described here, rather than systemic error or reckless disregard of constitutional requirements, any marginal deterrence does not 'pay its way.'" 555 U.S. at 147 to 148. This page previously quoted the Reporter's syllabus instead, which renders the same holding as isolated negligence "attenuated from the search", and the difference is substantive rather than cosmetic. The syllabus reads as a narrow exception for carelessness remote from the search; the opinion's rule reaches police negligence generally unless it rises to systemic error or reckless disregard. That correction runs in this page's favour — Herring takes more away from the exclusionary rule than the syllabus suggests, and this page had been understating it — which is the reason to state the provenance rather than simply bank the wider rule. Davis v. United States, 564 U.S. 229 (2011), held that "[s]earches conducted in objectively reasonable reliance on binding appellate precedent are not subject to the exclusionary rule" — so when a precedent is later overruled, the people searched under it get nothing. Utah v. Strieff, 579 U.S. 232 (2016), admitted evidence found after an unlawful stop because the discovery of an outstanding arrest warrant attenuated the taint.
The running justification the Court uses for narrowing the rule is one this page owes the reader in the Court's own words rather than around them. In Herring the majority states it plainly: "The principal cost of applying the rule is, of course, letting guilty and possibly dangerous defendants go free—something that 'offends basic concepts of the criminal justice system,'" quoting United States v. Leon, 468 U.S. 897, 908 (1984). 555 U.S. at 141. That framing is the answer the exclusionary-rule majority gives to the pattern described in this section: a majority of the Court thinks the costs of suppression are borne by the public rather than by the officer, and a page that would like to see the rule broadened has to meet that framing on its own terms.
Read together with the previous section, these decisions describe a system in which an unconstitutional search can produce no consequence of any kind: not suppression, because the officer was not culpable enough; not damages, because the right was not clearly established; not an injunction, because no individual plaintiff can show it will happen to him again. We do not claim to know how commonly all three fail at once, and nothing on this page measures it. We claim that each of the three exits is real, that they are independent of one another, and that a right with three independent exits is a weaker thing than its statement suggests.
THERE IS NO GENERAL RIGHT TO BE PROTECTED, AND TWO EXCEPTIONS
Public safety is a political promise, not a general constitutional entitlement, and the two are constantly confused in the language of campaigns.
DeShaney v. Winnebago County Department of Social Services, 489 U.S. 189 (1989), held that a state's failure to protect a child from private violence it knew about did not violate the Due Process Clause. Joshua DeShaney was beaten into permanent brain damage by his father after county social workers had recorded the abuse and left him in the home.
Town of Castle Rock v. Gonzales, 545 U.S. 748 (2005), went further. Jessica Gonzales held a restraining order against her estranged husband. Over several hours she repeatedly told police he had taken their three children in violation of it. The police did not act. He murdered the children. The Court held that she "did not, for purposes of the Due Process Clause, have a property interest in police enforcement of the restraining order against her husband." 545 U.S. at 768. An earlier draft of this page quoted the Reporter's syllabus, which inverts that phrase to "for Due Process Clause purposes". The substance is the same; the attribution was not.
The word doing the work in that summary is "general", and the draft of this page dropped it. DeShaney itself states the custodial exception in terms: "when the State takes a person into its custody and holds him there against his will, the Constitution imposes upon it a corresponding duty to assume some responsibility for his safety and general well-being." 489 U.S. at 199 to 200, citing Estelle v. Gamble and Youngberg v. Romeo. That is why a prisoner has an Eighth Amendment claim for denied medical care at all, and it is the same affirmative obligation the Court relied on in West v. Atkins, discussed below. DeShaney also reasoned that the State "played no part in [the dangers'] creation, nor did it do anything to render him any more vulnerable to them," 489 U.S. at 201, and the courts of appeals have built a state-created-danger theory on that reasoning. This analysis did not survey how they have defined it and states no rule for it.
We do not say the doctrine is incoherent. A general constitutional duty to police well would be unadministrable, and the Court's reluctance is not obviously wrong. But it should be understood plainly by anyone reading a promise about safety: outside custody and outside a danger the state itself created, a government may constitutionally choose not to protect you. What the Constitution mainly regulates is what the government does to you when it acts.
THE FINE IS THE POINT
The oldest and clearest constitutional rule in this field is that the decision-maker may not have a financial stake in the outcome. Tumey v. Ohio, 273 U.S. 510 (1927), turned on a mayor-judge's personal fee interest; Ward v. Village of Monroeville, 409 U.S. 57 (1972), is the institutional case. In Ward the Court held that the "possible temptation" Tumey identified "may also exist when the mayor's executive responsibilities for village finances may make him partisan to maintain the high level of contribution from the mayor's court." 409 U.S. at 60. The fuller statement this page previously quoted as the Court's — that the petitioner "was denied a trial before a disinterested and impartial judicial officer as guaranteed by the Due Process Clause of the Fourteenth Amendment where he was compelled to stand trial for traffic offenses before the mayor, who was responsible for village finances and whose court through fines, forfeitures, costs, and fees provided a substantial portion of village funds" — is the Reporter's syllabus, at 409 U.S. 57.
And a person may not be jailed for being too poor to pay. Bearden v. Georgia, 461 U.S. 660 (1983): "In other words, if the State determines a fine or restitution to be the appropriate and adequate penalty for the crime, it may not thereafter imprison a person solely because he lacked the resources to pay it." 461 U.S. at 667 to 668. The capitalised "If a State" this page previously quoted is the Reporter's syllabus. The opinion states the rule as a restatement of Williams v. Illinois and Tate v. Short rather than as a free-standing pronouncement, and the "in other words" is the Court's own signal that it is summarising those two cases. Bearden's actual rule is narrower than the slogan and the narrowing matters: a court "must inquire into the reasons for the failure to pay", and if the person "willfully refused to pay or failed to make sufficient bona fide efforts legally to acquire the resources to pay," imprisonment remains available. What Bearden forbids is jailing someone for poverty without asking. It does not forbid jailing someone who could pay and would not.
The Excessive Fines Clause now binds the states. Timbs v. Indiana, 586 U.S. 146 (2019), held: "The Excessive Fines Clause is therefore incorporated by the Due Process Clause of the Fourteenth Amendment." 586 U.S. at 150. The phrasing this page previously attributed to the Court — an "incorporated protection applicable to the States under the Fourteenth Amendment's Due Process Clause" — is the Reporter's syllabus. What counts as excessive it did not decide.
Those rules exist. The Justice Department's own investigation of Ferguson, Missouri, published March 4, 2015, is the most detailed federal record of a jurisdiction operating against them. It found that "Ferguson's law enforcement practices are shaped by the City's focus on revenue rather than by public safety needs"; that "[t]he City budgets for sizeable increases in municipal fines and fees each year, exhorts police and court staff to deliver those revenue increases, and closely monitors whether those increases are achieved"; that in March 2010 the City Finance Director wrote to the police chief that "unless ticket writing ramps up significantly before the end of the year, it will be hard to significantly raise collections next year"; and that officer evaluations and promotions depended "to an inordinate degree on 'productivity,' meaning the number of citations issued." The Department also found that the municipal court "primarily uses its judicial authority as the means to compel the payment of fines and fees that advance the City's financial interests," and that in 2013 alone it issued over 9,000 warrants, largely on minor matters such as parking infractions, traffic tickets and housing code violations.
Ferguson is one documented instance of a structure that exists wherever a police department and a court are funded out of the penalties they impose. This analysis does not establish how many other jurisdictions are built that way; no federal record measures it. What it establishes is that the constitutional rules against it are almost a century old, that a federal investigation found a city operating against them in detail, and that those rules are enforced only when someone with standing, a lawyer and a surviving claim reaches a court.
CIVIL FORFEITURE: THE PROPERTY IS THE DEFENDANT
Civil forfeiture proceeds against the thing rather than the person, so the protections that attach to a criminal prosecution do not attach to it. Timbs settled that the Excessive Fines Clause reaches state forfeitures. It did not settle how much process is owed before the state may keep a car for months.
Culley v. Marshall, 601 U.S. 377 (2024), answered that narrowly and against the owner: "In civil forfeiture cases, the Due Process Clause requires a timely forfeiture hearing, but does not require a separate preliminary hearing." 601 U.S. at 393. The Reporter's syllabus adds the limiting words "involving personal property", and it was the syllabus, not the opinion, that this page previously quoted as the holding. Halima Culley had loaned her car to her son and Lena Sutton had loaned hers to a friend; both borrowers were arrested on drug charges and both cars were seized under Alabama law. The owners asked for a prompt hearing on whether the state could hold the car while the forfeiture case ran. The Court held the Constitution does not require one.
The consequence does not depend on anyone acting in bad faith. The state may take a vehicle from someone who is not accused of anything, and the owner's only constitutionally guaranteed remedy is a hearing that arrives when the case arrives. For an owner who needs the car to keep a job, the process is the punishment.
THE RECORD DOES NOT EXIST, AND THAT IS A CHOICE
A constitutional argument about policing depends on facts about policing, and the United States does not systematically collect them. This is the least discussed and most consequential fact on this page, and it is also the one this analysis has to be most careful about, because the state of the record changes.
There is no complete national count of how often police use force, and the collection is voluntary. When the FBI obtained approval for its National Use-of-Force Data Collection in September 2018, the Office of Management and Budget set participation thresholds: agencies representing at least 60 percent of all sworn officers before the FBI could publish limited data, 80 percent before it could publish without conditions, and the collection was to end if 60 percent was not reached by the end of 2022. The Government Accountability Office reported that as of November 2021, after two years of collection, the FBI "had achieved 44 percent and 55 percent participation for calendar years 2019 and 2020, respectively," and that FBI officials said the thresholds were a known risk "because law enforcement agencies may be reluctant to share sensitive data on use of force incidents." GAO, Law Enforcement: DOJ Can Improve Publication of Use of Force Data and Oversight of Excessive Force Allegations, GAO-22-104456 (Dec. 7, 2021).
Those figures are a 2021 snapshot, and this page has to correct itself twice about what came after them. An earlier draft said no current figure could be obtained because "the site refuses automated requests". That is true of www.fbi.gov, which answers an automated request with a Cloudflare challenge and HTTP 403, and it is not true of cde.ucr.cjis.gov, the Crime Data Explorer host named in the preceding sentence, which answers normally. The draft then asserted, with no citation at all, that the FBI "has continued to issue use-of-force data updates" — an assertion it could not have drawn from a source it declared unreachable. Both defects are fixed by opening the record.
The record runs against this page's argument and is published for that reason. The FBI announced on June 10, 2024 that it had released the National Use-of-Force Data Collection Update, March 2024 on the Crime Data Explorer, and stated that the June 2024 release "reflects data from 72 percent of the law enforcement population participating in the collection"; it also recorded that from January 2024 participation is computed over a rolling twelve-month span rather than reset to zero each year. Its August 28, 2025 release states: "The release of use-of-force incident data is contingent upon achieving an 80% coverage of the law enforcement population by participating agencies. Information released in August 2025 reflects data reported by agencies covering 78% of the law enforcement population; therefore, the number of incidents are not available." Both pages were read from Internet Archive captures of the FBI's own press releases, because www.fbi.gov will not serve them to a script. The collection has therefore moved into a range where the FBI now reports event-type distributions on a rolling basis, and the national record is markedly less incomplete than the 2021 snapshot alone implies — though the pre-2024 calendar-cumulative figures and the post-2024 rolling twelve-month figures are not on the same measurement scale, so a direct 44-to-78 comparison overstates the improvement.
What has not changed is the design, and the design is still the point. At 78 percent the FBI reports the distribution of event types and cannot publish the number of incidents, because publication is gated at 80 percent. Participation is voluntary, what may be published is set by how many agencies volunteer, and the completeness of the national record is consequently decided by the agencies being measured. This analysis states no participation figure later than August 2025.
There is no reliable count of deaths in custody either, and that one is required by statute. The Death in Custody Reporting Act of 2013 conditions certain federal grants on reporting deaths in custody and directs the Attorney General to study the data. 34 U.S.C. 60105(a), (f)(1). The condition is weaker than the word suggests: a State that fails to report "shall, at the discretion of the Attorney General, be subject to not more than a 10-percent reduction" of its Byrne/JAG allocation. 34 U.S.C. 60105(c)(2). The penalty is capped at a tenth, and whether to impose any part of it is one official's choice. Testifying to the Senate Permanent Subcommittee on Investigations on September 20, 2022, GAO reported that it compared fiscal year 2021 state submissions against publicly available records and "identified nearly 1,000 deaths that potentially should have been reported to DOJ but were not"; that "70 percent of the records provided by states were missing at least one element required by DCRA, such as a description of the circumstances surrounding the individual's death or the age of the individual"; and that DOJ had not studied the data at all, "in part, because the data were incomplete." GAO added that "four states that accepted JAG awards did not report any deaths in custody in their state — even though reporting this information is a requirement of receiving the grant funding and deaths occurred in their state during this time period." GAO-22-106033. GAO described the deaths it found as potentially reportable rather than confirmed, and said its own count may be narrower than the true universe because not every state published the underlying records. This analysis did not verify whether reporting improved after fiscal year 2021.
Even the crime numbers are two numbers. The Bureau of Justice Statistics reports, in the NIBRS Estimation and Extract File Creation Methodology Report (NCJ 310516, December 2025), that "in 2023, about 13,730 agencies (representing 82% of the U.S. population) submitted at least 3 months of data through NIBRS", so national figures are weighted estimates built to compensate for the agencies that did not report. The National Crime Victimization Survey, which asks people rather than police, found that "[a]pproximately 48% of violent victimizations were reported to police in 2024", and its table 4 puts the equivalent share of property victimizations at 30.5 percent. Criminal Victimization, 2024 (NCJ 310547, September 2025), at 7. An earlier draft of this page cited both figures to this site's own evidence page for the issue rather than to the documents; the figures were right and the sourcing was not. A claim that crime is up or down is a claim about one of two measurements, taken over different populations with different denominators.
The point is not that anyone is hiding something. It is that the constitutional argument, the policy argument and the campaign claim all rest on a record the federal government has repeatedly declined to complete, and that a rights framework which depends on proving a pattern cannot function without one. That cuts both ways, and we say so: a record too incomplete to prove a pattern is also too incomplete to disprove one, and too incomplete to support confident claims in our own direction.
TECHNOLOGY SOLD AS PUBLIC SAFETY, MEASURED
Two of the most widely funded policing technologies have been evaluated by government bodies, and those evaluations are on this issue's evidence page. This section reports what the evaluations say. It does not quote a vendor's or a purchasing government's claim of effectiveness against them, because this pass did not obtain one; a reader should treat the heading as our characterisation and the figures as the record.
On body-worn cameras, the National Institute of Justice, in an article published January 7, 2022, reports that "[a] comprehensive review of 70 studies of body-worn cameras use found that the larger body of research on body-worn cameras showed no consistent or no statistically significant effects", and that "[t]his meta-analysis was rated by CrimeSolutions and resulted in a No Effects rating for the impact of body-worn cameras on use of force, assaults on officers, officer-initiated calls for service, arrests, traffic stops and tickets, and field interviews (i.e., stop and frisk)." Those are two sentences and an earlier draft of this page fused them into one, which made the review of 70 studies the subject of a verb that belongs to the meta-analysis. A randomised controlled trial run by the District of Columbia's own Lab @ DC with the Metropolitan Police Department, in a working paper dated October 20, 2017, covering 2,224 officers under a pre-registered analysis plan, reported that the authors "estimated very small average treatment effects on all measured outcomes, none of which rose to statistical significance." NIJ also notes that individual programmes have rated differently, including an Effective rating for a Boston evaluation, so the evidence base is mixed rather than uniformly null, and we do not overstate it.
On gunshot detection, the Chicago Office of Inspector General, in a report dated August 24, 2021, analysed city data on alerts between January 1, 2020 and May 31, 2021. Of 50,176 confirmed and dispatched alerts, 41,830 recorded a disposition, and "[a] total of 4,556 of those 41,830 dispositions indicate that evidence of a gun-related criminal offense was found, representing 9.1% of CPD responses to ShotSpotter alerts." The denominator is worth stating plainly, because the quoted sentence contains two of them: 4,556 is 10.9 percent of the 41,830 alerts that recorded a disposition, and 9.1 percent of all 50,176 dispatched alerts. Documented investigatory stops shared an event number with an alert in 2.1 percent of the 50,176 dispatched alerts. The OIG concluded that such responses "rarely produce documented evidence of a gun-related crime, investigatory stop, or recovery of a firearm," and that it found evidence the technology changed how some officers perceive and interact with people in high-alert areas.
The constitutional relevance is direct, and it is the reason these figures sit in a constitutional analysis rather than a policy brief. An alert that sends officers to a location expecting an armed person is an input to reasonable suspicion under Terry, and the officer's perception on arrival is the input Graham then defers to. A system whose alerts produce evidence of a gun crime in roughly one dispatched response in eleven, and which the city's own inspector general found changes how officers approach people in the areas it covers, is shaping the predicate for stops and the reasonableness of force at the same time.
One government-adjacent body has weighed in on the broader ecosystem within which these technologies sit, and its findings run in both directions. The National Academies of Sciences, Engineering, and Medicine's 2017 consensus-study Highlights, Proactive Policing: Effects on Crime and Communities (DOI 10.17226/24928, November 2017), reports a genuine crime-reduction effect for the tightly focused end of the field: "The available research suggests that hot spot policing interventions produce short-term crime-reduction effects without simply displacing crime into surrounding areas." It reports no comparable effect for the indiscriminate end: "generalized, aggressive enforcement tactics such as stop-question-frisk used indiscriminately across a city, or broken-windows policing programs that rely on a 'zero tolerance' generalized approach to misdemeanor arrests, do not show evidence of effectiveness." And on the two outcomes closest to this page's argument, the finding cuts against certainty in either direction: "the effects of proactive policing on other important outcomes—such as on the legality of police behavior and on racially biased behavior—are unclear because of gaps in research." The cut is against the page in both directions the page would like to run: hot-spots and problem-oriented work have short-term effects the page's technology critique does not concede, and the empirical case against proactive policing on constitutional and equal-protection grounds the page relies on is what the study calls unclear. We state both, because a page that only quotes the half that helps is not the page this standard is for.
WHAT IS GENUINELY CONTESTED
Four questions are live, and each is open for a different reason. The Supreme Court has left the first and third undecided; the second is a question about what a 2025 decision will do; the fourth has real authority running both ways and a section of its own.
First, whether Kingsley's objective standard extends beyond force to a pretrial detainee's medical-care and conditions-of-confinement claims. Kingsley held that a detainee challenging force need not prove the officer's subjective state of mind. Farmer requires exactly that of a convicted prisoner. Kingsley concerned force, and the Supreme Court has not said which standard governs a detainee's medical-care or conditions claim. This analysis did not survey how the courts of appeals have answered that and names no alignment. What it can say is that the question sits between two Supreme Court holdings pointing in different directions, and that the person it affects is someone the state is holding without having convicted him of anything.
Second, what Barnes v. Felix will actually change. A unanimous Court removed a rule that let courts ignore everything before the final seconds. Whether widening the frame changes outcomes is unknown: Graham's deference to the officer's on-scene perspective is intact, qualified immunity is intact, the Court expressly left open whether an officer's own creation of the danger counts against him, and the Court's own illustration of the wider frame — Plumhoff — is a case in which the wider frame justified the shooting. A reader should treat Barnes as a live question about consequences rather than a settled improvement.
Third, wealth-based pretrial detention. Salerno upheld detention on dangerousness findings under a carefully limited federal statute. Bearden bars imprisoning a person after conviction solely for inability to pay, subject to the wilfulness inquiry set out above. Whether Bearden's reasoning reaches the front end — conditioning release on money a defendant does not have — has never been decided by the Supreme Court. This analysis did not survey the lower-court decisions on the question and takes no position on how they have divided; that is a gap in this page, not an absence of litigation. Salerno's own affirmative language belongs on the page rather than paraphrased away, because a page arguing to shrink pretrial detention on constitutional grounds has to meet the Court's language: "the Government's regulatory interest in community safety can, in appropriate circumstances, outweigh an individual's liberty interest," 481 U.S. at 748, and "In our society liberty is the norm, and detention prior to trial or without trial is the carefully limited exception. We hold that the provisions for pretrial detention in the Bail Reform Act of 1984 fall within that carefully limited exception," 481 U.S. at 755. That is the framing this page's argument would extend into money bail, and the reader is entitled to see it stated in the Court's own words before reading our position on where the law should go.
Fourth, whether the state-action cases reach the private carceral and enforcement industry. That has its own section below, and it is genuinely contested rather than settled in either direction.
Qualified immunity's future does not belong on this list, and the draft of this page put it here. In the federal courts the question is not open in any practical sense: Pearson remains good law and the Supreme Court has not reconsidered the doctrine's existence. Where it is open is in legislatures, and legislatures are a different bucket. What two states have done about it, and what their statutes do not do, is set out under where we believe the law should go.
SOLDIERS IN AMERICAN CITIES
The most consequential live question in this field is not about a stop or a search. It is whether the federal executive may put soldiers into American cities to do police work, and since 2025 it has been litigated hard. This section is a record of documents this analysis read. It is not a current account of where troops are, and it says where it stops.
The statutory frame matters more than most constitutional argument here. The Posse Comitatus Act makes it a crime, "except in cases and under circumstances expressly authorized by the Constitution or Act of Congress," wilfully to use "any part of the Army, the Navy, the Marine Corps, the Air Force, or the Space Force as a posse comitatus or otherwise to execute the laws," 18 U.S.C. 1385. The list of services is current text: Congress added the Navy, the Marine Corps and the Space Force in December 2021, and a reader working from the older "Army and Air Force" version will wrongly conclude that Marines sit outside the Act. Under 10 U.S.C. 12406 the President may federalise the National Guard in three circumstances, the third being that he "is unable with the regular forces to execute the laws of the United States." A separate authority, 32 U.S.C. 502(f), lets Guard members perform "operations or missions" requested by the President or Secretary of Defense while remaining in their state chain of command. The Insurrection Act, 10 U.S.C. 251 to 253, is a third and much broader authority that was not invoked in the litigation described here.
What the Supreme Court did. In Trump v. Illinois, 607 U.S. ___ (Dec. 23, 2025) (No. 25A443), the Court denied the Government's application to stay a district court order barring deployment of federalised Guard units in Illinois. It concluded that "regular forces" in the statute "likely refers to the regular forces of the United States military," so that before federalising the Guard on that ground the President "likely must have statutory or constitutional authority to execute the laws with the regular military and must be 'unable' with those forces to perform that function." Because the Posse Comitatus Act bars the military from executing the laws except where expressly authorised, the Court reasoned, the circumstances in which the third ground is available "are exceptional." And: "At this preliminary stage, the Government has failed to identify a source of authority that would allow the military to execute the laws in Illinois." The Court also stated that it did not address the reviewability of findings made by the President under that statute or any other.
Three limits belong with that, and the draft of this page carried only one. It is an order on a stay application, not a merits judgment, and most of its operative sentences say "likely" — though not all of them do. It was not unanimous: Justice Kavanaugh concurred in the judgment only, voting to deny the application "on narrow grounds" and declining to reach the statutory questions the Court decided; Justice Alito, joined by Justice Thomas, dissented, objecting among other things that the Court decided the case on a reading of "regular forces" that neither party had advanced below; and Justice Gorsuch dissented separately. The reasoning quoted above is the Court's, and it commands five votes, not nine.
What happened in the District of Columbia, in sequence, because the sequence is what a reader is most likely to be told wrongly. Over 2,000 Guard members had been deployed in the District since August 11, 2025, including units from nine States. On November 20, 2025, the district court granted a preliminary injunction and a stay under the Administrative Procedure Act, finding that the Defense Department defendants had exceeded their authority in deploying the D.C. Guard for non-military, crime-deterrence missions without a request from the city's civil authorities and lacked statutory authority under 32 U.S.C. 502 to support the out-of-state units, and stayed its own order for 21 days. District of Columbia v. Trump, No. 25-cv-3005 (JMC) (D.D.C. Nov. 20, 2025). On December 4, 2025, the D.C. Circuit administratively stayed that order. On December 17, 2025, the same panel — Circuit Judges Millett, Katsas and Rao — dissolved the administrative stay and granted a stay pending appeal, holding that "the Defendants appear on this early record likely to prevail on the merits of their argument that the President possesses a unique power within the District — the seat of the federal government — to mobilize the Guard" under 32 U.S.C. 502(f), "[b]ecause the District of Columbia is a federal district created by Congress, rather than a constitutionally sovereign entity like the fifty States." The district court's order therefore never took effect, and the panel recorded that the Guard was "set to remain in the District until at least February 28, 2026, unless the district court's ordered relief takes effect."
That stay decision carries its own limits, stated on its face: "our assessment of the merits is rooted in the preliminary and hurried posture of a stay motion," it "does not bind the merits panel," and the panel expressly did not reach whether the Guard units were engaged in law enforcement in violation of the Posse Comitatus Act or whether the federal government was improperly exercising command and control over the state Guards.
Where this analysis stops. It verified the December 23, 2025 Supreme Court order, the December 4 and December 17, 2025 D.C. Circuit orders, and the district court's disposition as those orders describe it, against the documents themselves. It did not establish what has happened since December 2025 — not the outcome of the merits appeal, not the current terms of any deployment, and not which cities have troops in them as this is published. Those are checkable facts and we decline to state them from anything but the record.
The constitutional point survives the procedural detail. The Posse Comitatus Act is a criminal statute, not a constitutional provision, and Congress may narrow or repeal it. The constitutional limits on domestic military force are correspondingly thin, largely untested, and mostly structural rather than individual. A soldier conducting a stop is still bound by the Fourth Amendment, because the Fourth Amendment binds the government and not a particular uniform. But almost every doctrine in the two sections above — qualified immunity, Lyons, the deterrence-only exclusionary rule — applies with at least equal force to a federal deployment, and Bivens applies with less. The person stopped by a soldier in an American city has, on current doctrine, fewer damages remedies than the person stopped by a city police officer, not more.
THE STRUCTURAL REMEDIES ARE DISCRETIONARY BY DESIGN
When a police department has a pattern of unconstitutional conduct, there are two federal structural remedies and one person may invoke either.
The first is 34 U.S.C. 12601. Subsection (a) makes it unlawful "for any governmental authority, or any agent thereof, or any person acting on behalf of a governmental authority, to engage in a pattern or practice of conduct by law enforcement officers or by officials or employees of any governmental agency with responsibility for the administration of juvenile justice or the incarceration of juveniles that deprives persons of rights, privileges, or immunities secured or protected by the Constitution or laws of the United States." Subsection (b) then hands the enforcement power to one person: "Whenever the Attorney General has reasonable cause to believe that a violation ... has occurred, the Attorney General, for or in the name of the United States, may in a civil action obtain appropriate equitable and declaratory relief to eliminate the pattern or practice." There is no private right of action in the section.
The second is the Civil Rights of Institutionalized Persons Act, 42 U.S.C. 1997 and following, which covers ground 12601 largely does not: conditions inside institutions, including jails, prisons and pretrial detention facilities. It is also the Attorney General's alone. Section 1997a(a) authorises suit where the Attorney General has reasonable cause to believe that "any State or political subdivision of a State, official, employee, or agent thereof, or other person acting on behalf of a State or political subdivision of a State" is subjecting institutionalised persons to "egregious or flagrant conditions" pursuant to a pattern or practice, and section 1997a(c) requires the Attorney General to sign the complaint personally.
On April 28, 2025, Executive Order 14288 directed the Attorney General, within 60 days, to "review all ongoing Federal consent decrees, out-of-court agreements, and post-judgment orders to which a State or local law enforcement agency is a party and modify, rescind, or move to conclude such measures that unduly impede the performance of law enforcement functions." 90 Fed. Reg. 18765 (May 2, 2025).
On May 21, 2025, the Civil Rights Division announced it was "taking all necessary steps to dismiss the Louisville and Minneapolis lawsuits with prejudice, to close the underlying investigations into the Louisville and Minneapolis police departments, and to retract the Biden administration's findings of constitutional violations," and that it was also closing investigations into and retracting findings as to police departments in Phoenix, Trenton, Memphis, Mount Vernon and Oklahoma City, and the Louisiana State Police. Department of Justice, Office of Public Affairs, Press Release 25-534.
That is a lawful exercise of the discretion both statutes confer, and we do not say otherwise. The structural fact is worth stating without spin: Congress wrote two remedies for systemic constitutional violations and made a single political officer the only gate on both. A remedy that switches on and off with an administration is functioning as an enforcement priority rather than as a remedy for a constitutional violation.
An earlier draft of this page put that point far too high, and the correction matters. Federal pattern-or-practice authority is not the only structural remedy that exists, and in the very case the draft used it was not the operative one. Minneapolis is bound by a court-enforceable settlement agreement obtained by the Minnesota Department of Human Rights under the Minnesota Human Rights Act, approved by a Hennepin County District Court judge on July 13, 2023 — a state remedy, unaffected by what the federal government did in May 2025. Chicago is bound by a consent decree obtained not by the United States but by the Illinois Attorney General, entered in Illinois v. Chicago, No. 17-cv-6260 (N.D. Ill.), and approved on January 31, 2019. And a certified class can obtain structural injunctive relief in federal court notwithstanding Lyons, as the New York stop-and-frisk litigation did.
So the accurate claim is narrower and still serious. The federal structural remedies are discretionary and they moved with the administration. What did not move is state civil-rights enforcement and private class litigation. Which means the availability of a structural remedy for unconstitutional policing depends on whether the department sits in a state whose attorney general has the authority and the will to act, or has drawn a class action capable of surviving Monell, standing and Heck. Where neither is true, the federal remedy is the only one, and it is one officer's choice.
WHAT THE STATUTES ACTUALLY SAY ABOUT AGENTS AND CONTRACTORS
This section sets out statutory text and holdings. Our own position is in the two sections after it, and the line between them is deliberate: an earlier draft of this page put the argument here, under established law, and headed it as though Congress had enacted our position. Congress enacted something narrower, and the difference is the kind of thing this page exists not to blur.
What 12601 does. The defendant clause is broad: the prohibition runs against "any governmental authority, or any agent thereof, or any person acting on behalf of a governmental authority." Congress wrote the agent, and the person acting on the government's behalf, into the prohibition itself, in the Violent Crime Control and Law Enforcement Act of 1994. The conduct clause is narrower, and the draft of this page omitted it: what is prohibited is a pattern or practice "of conduct by law enforcement officers or by officials or employees of any governmental agency with responsibility for the administration of juvenile justice or the incarceration of juveniles." The juvenile branch is expressly confined to a governmental agency. Whether the employees of a private contractor are "law enforcement officers" within the first branch has not been construed by any decision this analysis located — unsurprisingly, since only the Attorney General may sue and there is very little litigated law under the section at all. Adult corrections is not this section's subject; it is CRIPA's.
What CRIPA does. Its definition of a covered institution is the plainest agent language in either statute: an institution is a facility "which is owned, operated, or managed by, or provides services on behalf of any State or political subdivision of a State," 42 U.S.C. 1997(1)(A), and section 1997a(a) reaches an "agent thereof, or other person acting on behalf of a State or political subdivision of a State." A privately owned and operated facility is excluded only where the State's licensing of it, or its receipt of payments under title XVI or XVIII or a State plan approved under title XIX of the Social Security Act, or the two together, "constitutes the sole nexus" between the facility and the State. 42 U.S.C. 1997(2). An earlier draft of this page glossed that exclusion as meaning "contracting or funding alone is not enough", and that gloss conceded ground the statute does not take. The word contract, in any form, appears nowhere in section 1997(2); neither does funding at large, the three named Social Security payment streams being the whole of it. Contracting runs the other way in this statute: providing services on the State's behalf is precisely what brings a facility in, under section 1997(1)(A). The accurate statement is that licensing or those benefit payments alone are not enough, and that providing the service on the State's behalf is — which is the same idea as the state-action cases discussed below, written into a statute, and written more favourably to us than this page had been reporting.
What West holds. Section 1983 reaches anyone acting under colour of state law, and the Supreme Court has held that a private physician under part-time contract to treat prisoners acts under colour of state law when he treats an inmate. West v. Atkins, 487 U.S. 42 (1988). The reasoning is the part that matters: "It is the physician's function within the state system, not the precise terms of his employment, that determines whether his actions can fairly be attributed to the State," and "[c]ontracting out prison medical care does not relieve the State of its constitutional duty to provide adequate medical treatment to those in its custody, and it does not deprive the State's prisoners of the means to vindicate their Eighth Amendment rights." 487 U.S. at 56. The Court added that "[t]he State bore an affirmative obligation to provide adequate medical care to West; the State delegated that function to respondent Atkins; and respondent voluntarily assumed that obligation by contract."
And what West does not hold. The affirmative obligation in that sentence is the custodial duty DeShaney describes, and it exists because the State had incarcerated West. West is a holding that a private party performing a function the State is constitutionally obliged to perform, for a person the State is holding, acts under colour of state law. It is not a holding that any private party performing any government function is bound by the Constitution. A reviewer can accept West in full and deny attribution to a data vendor, a forensic laboratory or a diversion programme, and nothing in the opinion answers him.
Two more authorities are often stretched here, and we state their real width. Griffin v. Maryland, 378 U.S. 130 (1964), treated a private amusement park's security guard as a state actor — but he was an actually deputised sheriff exercising police power, which is narrower than function-by-contract. And the federal criminal civil-rights statute, 18 U.S.C. 242, reaches whoever acts "under color of any law, statute, ordinance, regulation, or custom"; as applied to a private person that element is satisfied by acting jointly with state officials, United States v. Price, 383 U.S. 787 (1966), not by holding a contract. Lugar v. Edmondson Oil Co., 457 U.S. 922 (1982), supplies the general frame: the question is whether the challenged conduct is fairly attributable to the state.
So the honest summary is this. That a private party performing a function the State is constitutionally obliged to perform acts under colour of state law is a holding. That Congress has repeatedly written agents and persons acting on a government's behalf into the text of its civil-rights enforcement statutes is enacted law. That any private party carrying out any government function is bound by the Constitution is neither. It is our position, and it is argued as one below.
THE PRIVATISATION GAP: NEITHER THE IMMUNITY NOR THE LIABILITY
Where the state contracts out custody, the doctrine has produced a result that is difficult to defend on any theory of the Constitution.
Richardson v. McKnight, 521 U.S. 399 (1997), held that prison guards employed by a private firm are not entitled to qualified immunity from a section 1983 suit by prisoners. That is a real gain for accountability, and it cuts our way. Two things about it cut the other way and we report both. The Court reasoned that "correctional functions have never been exclusively public," which is a direct obstacle to arguing that running a prison is a traditional exclusive public function within the state-action cases. And the Court confined its own holding: it answered the immunity question "narrowly, in the context in which it arose", a context "in which a private firm, systematically organized to assume a major lengthy administrative task (managing an institution) with limited direct supervision by the government, undertakes that task for profit and potentially in competition with other firms", and expressly not one involving "a private individual briefly associated with a government body, serving as an adjunct to government in an essential governmental activity, or acting under close official supervision."
Fifteen years later the Court took up that reservation and answered it against plaintiffs. Filarsky v. Delia, 566 U.S. 377 (2012), held unanimously that "immunity under [section] 1983 should not vary depending on whether an individual working for the government does so as a full-time employee, or on some other basis," 566 U.S. at 389, and distinguished Richardson as a case about prison guards employed by a private company in a privately run facility. Both of the sentences this page previously quoted for those propositions — that "[a] private individual temporarily retained by the government to carry out its work is entitled to seek qualified immunity from suit", and that Richardson "involved the unusual circumstances of prison guards employed by a private company who worked in a privately run prison facility" — are the Reporter's syllabus, at 566 U.S. 377 and 379 respectively. So the rule is not that private actors lose the immunity. It is that the large private prison operator loses it and the retained individual does not — a line drawn by employment arrangement rather than by what the person was doing.
Against the federal government, privatisation runs the other way. Correctional Services Corp. v. Malesko, 534 U.S. 61 (2001), refused to extend a Bivens damages action to a private corporation operating a federal halfway house; part of its ground was the narrower one that a Bivens action does not lie against an entity at all, FDIC v. Meyer, 510 U.S. 471 (1994), rather than anything about privatisation as such. Minneci v. Pollard, 565 U.S. 118 (2012), refused to extend Bivens to the employees of a privately run federal prison where the conduct was "of a kind that typically falls within the scope of traditional state tort law," on the ground that state tort law supplied an adequate alternative remedy. Egbert has since narrowed Bivens close to its own facts for everyone.
The composite is this, and the last step of it is an inference we mark as one. A federal prisoner injured by a federal employee has a Bivens claim in theory, and after Egbert rarely in practice. A federal prisoner injured by a private employee doing the identical job in the identical building is directed by Minneci to state tort law, where the Eighth Amendment standard does not apply, no constitutional question is asked or answered, and nothing about the practice is adjudicated for the next prisoner. Minneci's holding is expressed as a category — conduct typically covered by state tort law — rather than as an absolute bar, so the flat statement that such a prisoner has no federal constitutional damages claim at all is our reading of what Minneci plus Egbert leaves, not a quotation from either.
What the gap is not is a holding that the conduct is constitutional. Nothing in Malesko or Minneci says a private jailer may treat a prisoner in ways the Constitution forbids. They say only that the federal courts will not create the damages action. The two are often collapsed, and the collapse is what makes the gap look like a considered judgment about rights rather than an accident of remedial doctrine.
WHETHER THE CONTRACTOR CASES REACH THE CARCERAL STATE
Standing against the agent principle is a line of cases holding that a contract and a subsidy do not make a private firm the state. We describe it accurately rather than around it, and then say where we think it does not fit. The description is the law. What follows the description is our argument, and it is marked.
Rendell-Baker v. Kohn, 457 U.S. 830 (1982), rejected state action for a private school drawing nearly all its income from public funds, comparing it to firms that build roads, bridges, dams, ships or submarines for the government. Blum v. Yaretsky, 457 U.S. 991, 1004 (1982), asks whether the State is responsible for "the specific conduct of which the plaintiff complains." Jackson v. Metropolitan Edison Co., 419 U.S. 345 (1974), and Flagg Bros. v. Brooks, 436 U.S. 149 (1978), are the core of the exclusive-public-function test and are demanding: a heavily regulated monopoly utility terminating service was not a state actor, and a warehouseman selling stored goods under a state statute was not either. Manhattan Community Access Corp. v. Halleck, 587 U.S. 802, 814 (2019), holds that "the fact that the government licenses, contracts with, or grants a monopoly to a private entity does not convert the private entity into a state actor," and describes the traditional-and-exclusive-public-function route as one very few functions satisfy.
That is a demanding body of law and we do not pretend otherwise. It is also narrower than the use commonly made of it. These cases hold that contracting or funding, standing alone, does not establish attribution. They do not hold, and do not purport to hold, that a firm exercising a coercive government power at the government's direction stands outside the Constitution. Those are different propositions, and treating them as the same is the most common error in this area.
From here on this is our argument. The mismatch is in the analogy itself. Rendell-Baker's comparison is to firms that build a thing and hand it over: the builder delivers a road, the government then owns it, operates it, and answers for how it is used, and the builder's role ends at delivery. That is procurement. It is not what happens here. The private prison does not deliver a prison and leave; it holds the prisoners. The contract medical provider does not deliver medicine; it decides who is treated. The private probation company does not deliver a service to the court; it collects the court's debt from the defendant under threat of the court's power. The forensic laboratory does not deliver equipment; it produces the evidence. The surveillance vendor does not deliver a database; it answers the query. In each case the private party is not supplying an input to government power. It is exercising the power.
Blum asks the right question and we accept it: is the State responsible for the specific conduct complained of? Our answer is that where a public authority defines the programme, funds it, refers the person into it by legal compulsion, and enforces its outcomes with its own coercive machinery, the specific conduct is the government's, and the private ownership of the payroll is an accounting fact rather than a constitutional one. A rule that says otherwise invites a government to buy what it is not permitted to build.
No court has adopted that reasoning across the board on these facts. This page says so. Richardson's own observation that correctional functions "have never been exclusively public" is a serious obstacle to the exclusive-function route in particular, and it is not answered here. We think the position is right anyway, and we will keep saying so while the question is litigated.
STUMPWATCH'S POSITION: THE GOVERNMENT MAY NOT HIRE OUT WHAT IT MAY NOT DO
This is our position. It is not a description of what courts have held, and we do not present it as one. It does not change if a court rules against it, because it is a claim about what the Constitution is for rather than a prediction about litigation.
A civilian company or individual acting on behalf of the government is a government agent, directly or by proxy. Constitutional protections against the government extend to its authorised agents by whatever method is used to obscure the government's control. The test we apply, on this issue and every other:
Would this be a constitutional violation if a government official did it directly? If it would, then it is the same violation when an authorised agent of the government does it instead.
Applied here the test does most of its work without controversy. A government may not hold a person in conditions that violate the Eighth Amendment, and it may not do so through a company. A government may not jail a person for inability to pay without asking why he has not paid, and it may not have a private probation company produce the same result by proxy. A government may not obtain a record it would need a warrant to compel, and it may not buy the same record from a vendor to avoid the requirement. A government may not use unreasonable force, and neither may a firm it has deputised. Nobody would accept the contrary answer if the subject were a warrantless house search, a seized printing press, or a coerced confession. It should not become acceptable because the subject is a jail contract or an alert from a sensor.
The purchase example is the one a reader will push on hardest, so we state its status precisely. Carpenter v. United States, 585 U.S. 296 (2018), expressly declined to disturb the third-party doctrine, and this analysis located no decision holding that a government's purchase of commercially available location data is a search. On that question current doctrine is not with us and there is no holding for us to point to. We hold the position anyway, and we would rather say that plainly than leave the strongest objection unnamed.
The anti-circumvention principle is real doctrine, not invention. Skinner v. Railway Labor Executives' Association, 489 U.S. 602 (1989), is the clearest illustration, and it has to be cited to the right half of the regulation. Subpart D authorised railroads to test breath and urine without requiring them to, and the Fourth Amendment applied to those tests anyway, because the government had authorised them, removed the legal barriers to them, and had an interest in the results; no compulsion was needed for that half. The mandatory blood and urine testing was Subpart C, and there the Court grounded the Amendment in the compulsion itself. Lugar asks whether the conduct is fairly attributable to the state. West answers it for prison medical care. Griffin answers it for a deputised private guard.
What is honestly unsettled is whether courts will find attribution on particular modern facts: private prisons and transport, privately operated probation and diversion programmes, private forensic laboratories, commercial location and biometric platforms sold to police. Never the principle, and never our position. A reviewer who shows that current doctrine runs against us has not refuted the position; only a reviewer who defeats the functional-equivalence reasoning itself would.
STATEMENTS MEASURED AGAINST THE RECORD
This section compares public statements against documents that can be checked. It is not a claim that anyone lied. Where a document says something different from the way it has been described, we set out both and let the difference stand. Where we could not obtain an underlying document, we say so and draw no conclusion from its absence.
AN ORDER TITLED "PROTECT INNOCENT CITIZENS," READ IN FULL
Executive Order 14288 is titled "Strengthening and Unleashing America's Law Enforcement To Pursue Criminals and Protect Innocent Citizens," and its stated purpose is that "[s]afe communities rely on the backbone and heroism of a tough and well-equipped police force."
It has five operative sections and they are not all of a piece, so we identify them individually rather than by a tally. Section 2 directs the Attorney General to "create a mechanism to provide legal resources and indemnification to law enforcement officers who unjustly incur expenses and liabilities for actions taken during the performance of their official duties to enforce the law," including through "private-sector pro bono assistance." Section 3(a) directs a broad set of measures to maximise federal resources for state and local police, from best practices and training to pay, benefits, legal protections, enhanced sentences for crimes against officers, prison investment and crime data. Section 3(b) is the 60-day consent-decree review quoted above. Section 4 directs the Attorney General and the Secretary of Defense to "increase the provision of excess military and national security assets in local jurisdictions to assist State and local law enforcement," and directs the Secretary of Defense to determine how "military and national security assets, training, non-lethal capabilities, and personnel can most effectively be utilized to prevent crime." Section 5 directs prosecution priority against state and local officials who obstruct law enforcement or who "unlawfully engage in discrimination or civil-rights violations under the guise of 'diversity, equity, and inclusion' initiatives that restrict law enforcement activity or endanger citizens." Section 6 routes implementation through the Homeland Security Task Forces created by Executive Order 14159.
What is visible on the face of that list is that Sections 2, 3(b) and 4 concern, respectively, the legal exposure of officers, the removal of judicial supervision, and the transfer of military assets and personnel to civilian policing. An earlier draft of this page described Section 3 as a whole as the removal of judicial supervision, which is not accurate: only 3(b) is. We record the item that cuts the other way, because omitting it would be the same failure in the opposite direction: Section 3(a)(vii) directs agencies to "increase the investment in and collection, distribution, and uniformity of crime data across jurisdictions," which addresses the measurement gap described above. Whether it was implemented, this analysis did not establish.
The order also states, in the ordinary form, that it creates no right or benefit enforceable at law or in equity. 90 Fed. Reg. at 18766 to 18767.
RETRACTING A FINDING IS NOT DISPROVING IT
The Civil Rights Division's May 21, 2025 announcement says the Division will "retract the Biden administration's findings of constitutional violations" as to eight law enforcement agencies. It describes the Louisville and Minneapolis suits as having "accused Louisville and Minneapolis of widespread patterns of unconstitutional policing practices by wrongly equating statistical disparities with intentional discrimination and heavily relying on flawed methodologies and incomplete data."
Read as a document, two things are visible on its face. First, retraction is an administrative act by the office that made the findings. It withdraws the government's assertion; it adjudicates nothing. No court found the earlier findings false, because the suits were not tried. Second, the release states a general methodological objection and does not identify a single specific finding, in any of the eight jurisdictions, that it contends was factually wrong. That is an observation about the release we read in full, not a claim that no detailed analysis exists anywhere; if one was published, this pass did not locate it.
The release also makes the affirmative case for what it did, and this page quoted the release three times without quoting the part that argues against us. It is the strongest statement of the opposing position anywhere in this analysis and it belongs here in full. Assistant Attorney General Harmeet K. Dhillon of the Civil Rights Division is quoted by name: "Overbroad police consent decrees divest local control of policing from communities where it belongs, turning that power over to unelected and unaccountable bureaucrats, often with an anti-police agenda," and "Today, we are ending the Biden Civil Rights Division's failed experiment of handcuffing local leaders and police departments with factually unjustified consent decrees." The release states its own rationale as a cost: "In short, these sweeping consent decrees would have imposed years of micromanagement of local police departments by federal courts and expensive independent monitors, and potentially hundreds of millions of dollars of compliance costs, without a legally or factually adequate basis for doing so." That is a real argument about local self-government and about the price of judicial supervision, it is made on the record by the official responsible for making it, and nothing in this analysis answers it.
The release also says the Department "stands ready to take all necessary action to address any resulting constitutional or civil-rights violations, including via criminal prosecution." Those are different instruments and the release does not claim otherwise. A prosecution under 18 U.S.C. 242 runs against an individual and requires proof beyond a reasonable doubt that he acted wilfully. A pattern-or-practice action under 34 U.S.C. 12601 runs against a system and seeks equitable relief. Substituting the first for the second leaves the systemic question unasked.
One thing the release did not do, and this page previously implied it did, is end structural oversight of the two named departments. Minneapolis remained bound by the state court-enforceable settlement agreement described above. Whether the Louisville and Minneapolis federal dismissals were entered, and on what terms, this analysis did not verify from the dockets; the release announces steps to dismiss rather than a completed dismissal, and we describe it that way.
THE GOVERNMENT'S TWO POSITIONS ON WHAT EXECUTING THE LAWS MEANS
In Trump v. Illinois the Government argued that "regular forces" in the federalisation statute refers to civilian law enforcement officers, such as those employed by Immigration and Customs Enforcement or the Federal Protective Service. In the same litigation it also maintained, consistent with the longstanding view of the Executive Branch, that performing protective functions for federal personnel and property "does not constitute 'execut[ing] the laws' within the meaning of the Posse Comitatus Act."
The Court set the two side by side. "If that is correct," it wrote, "it is hard to see how performing those functions could constitute 'execut[ing] the laws'" under the federalisation statute, citing the presumption that Congress does not silently attach different meanings to the same term in related statutes. That is not our characterisation of the Government's litigating position. It is the Court's, in the order denying the stay, quoted from the slip opinion.
Two qualifications, because this is a Hidden Agendas section and the standard of proof for one has to be higher than for an argument. The observation is the majority's and not the whole Court's: Justice Alito's dissent records that in the lower courts the parties agreed that "regular forces" meant civilian federal law enforcement officers, and that the Court raised the alternative reading itself. And a litigant taking a position in one statute and a different one in another is a normal feature of advocacy; what makes this worth recording is that the Supreme Court itself found the two hard to hold together, not that we do.
WHAT THESE COMPARISONS DO NOT ESTABLISH
They do not establish that any official knowingly said anything false, and no reader should take them that way. They do not establish the current state of federal troop deployments or the outcome of the deployment appeals; this analysis verified court orders through December 17, 2025 and stopped there. They do not establish the FBI's use-of-force participation rate after 2021, or whether deaths-in-custody reporting improved after fiscal year 2021. They record no position from any police union, police association, technology vendor, or reform or civil-liberties organisation, in either direction; where such bodies have disputed the findings quoted here, those disputes were not retrieved and are therefore not characterised. They establish no causal claim about crime: nothing on this page attributes any trend in crime, incarceration or overdose mortality to any statute, administration, technology or policy. And they establish no frequency: this analysis does not measure how often any doctrine on this page actually disposes of a case.
This analysis does not reach several sub-domains that a comprehensive criminal-justice review would cover: sentencing and mandatory minimums; capital punishment; juvenile justice; plea bargaining; the Sixth Amendment jury right as applied to criminal trials; the Controlled Substances Act and drug scheduling; dual-sovereignty doctrine (Bartkus v. Illinois, 359 U.S. 121 (1959); Gamble v. United States, 587 U.S. 678 (2019)); criminal forfeiture as distinct from civil forfeiture; collateral consequences of conviction; prosecutorial discretion as a doctrinal question; and non-federal bail-reform experience beyond what is cited above. Each is a substantial body of law in its own right. Their absence is a scope choice, not a claim that they are unimportant or that the constitutional analysis here would necessarily reach the same conclusions if extended to them.
WHERE WE BELIEVE THE LAW SHOULD GO
This section, like the position above it, is editorial rather than descriptive. It sets out what we think would work, with the honest limits of each.
Qualified immunity should be abolished by statute rather than waited out in the courts, and the two existing state models show both the shape and the shortfall. Colorado's Enhance Law Enforcement Integrity Act, S.B. 20-217, added Colo. Rev. Stat. 13-21-131, creating a cause of action against a peace officer who under colour of law deprives a person of rights secured by article II of the state constitution, including by failing to intervene; it provides that qualified immunity is not a defence, disapplies the Colorado Governmental Immunity Act, and requires a fee award to a prevailing plaintiff. The statute reaches further than an earlier draft of this page said, and that correction runs our way: as enacted in 2020 the section reached only a peace officer "employed by a local government", and House Bill 21-1250 struck those five words, effective July 6, 2021, so the current text runs against "[a] peace officer, as defined in section 24-31-901 (3), who, under color of law, subjects or causes to be subjected" any person to such a deprivation. State officers are inside it now, not only local ones. Its limits deserve stating just as plainly, and they run the other way. The officer's employer must indemnify; personal exposure arises only if the employer itself determines, "on a case-by-case basis", that the officer did not act on a good-faith and reasonable belief that the action was lawful; and it is then capped at five percent of the judgment or settlement or twenty-five thousand dollars, whichever is less. Even that residue is backstopped, and this page omitted the backstop: the same subsection provides that "if the peace officer's portion of the judgment is uncollectible from the peace officer, the peace officer's employer or insurance shall satisfy the full amount of the judgment or settlement." Colo. Rev. Stat. 13-21-131(4)(a). A statute that abolishes qualified immunity and then guarantees the officer will not personally pay is a weaker model than the headline suggests. New Mexico's Civil Rights Act, enacted in 2021, prohibits the qualified-immunity defence for deprivations of rights secured by the bill of rights of the New Mexico constitution and reaches "a public body or person acting on behalf of, under color of or within the course and scope of the authority of a public body" — the agent principle written into a state statute, which is why we point to it. Its limits: claims lie exclusively against the public body, the body pays the judgment, and liability is capped per claimant inclusive of costs and fees at a figure enacted at two million dollars which the Act itself raises for the cost of living each July 1, so the current cap is higher than the enacted number. Both statutes reach only their own state constitutions. Other states and localities have taken narrower steps, which this analysis did not survey. Federal qualified immunity is untouched by any of it, and only Congress or the Supreme Court can reach that.
Congress should give 34 U.S.C. 12601 and CRIPA a private right of action, and should make 12601's conduct clause as wide as its defendant clause. The prohibitions already exist and already name agents and persons acting on a government's behalf; what is missing is anyone other than the Attorney General who can invoke them, and a conduct clause that plainly covers a contractor's staff. Adding a private action with fees would leave the Attorney General's power entirely intact and would stop the remedy from vanishing between administrations.
Congress should make the data mandatory and self-enforcing. The Death in Custody Reporting Act already conditions grants on reporting, and GAO found states taking the grants and reporting nothing, with no consequence. The statute is built to produce that result: non-compliance is punishable only "at the discretion of the Attorney General" and only by "not more than a 10-percent reduction" of the grant, 34 U.S.C. 60105(c)(2), so the maximum price of reporting nothing is a tenth of one funding stream, imposed at the option of the same officer who holds both structural remedies described above. A mandatory, larger and automatic withholding is the change that would make the condition operate as one. The use-of-force collection has the opposite defect and it is a defect of design rather than of anyone's diligence: participation is voluntary and what may be published is gated on how many agencies volunteer, so the agencies being measured decide how complete the measurement is. A right that depends on proving a pattern cannot be vindicated against a record built that way.
Revenue and enforcement should be separated by statute. Tumey and Ward already forbid the adjudicator with a stake. The rule should be extended to the institution: fines, fees and forfeiture proceeds should be remitted to a fund on which neither the enforcing agency nor its court draws. Ferguson is what the alternative looks like, and Ferguson was documented by the federal government itself rather than by its critics.
Forfeiture should require a conviction, or at minimum a prompt hearing on retention. Culley holds the Constitution does not require that hearing. A legislature may require it anyway, and this is the ordinary pattern in this field: where the Constitution has been read to permit something, the enforceable limits that exist come from statutes.
The agent rule should be enacted rather than litigated case by case. A statute providing that any private party performing a policing, custodial, prosecutorial, adjudicative or supervisory function under contract with or at the direction of a government is subject to the same constitutional constraints and the same causes of action as the government would close in one sentence the gap that Malesko, Minneci and Egbert left open, and would make the Filarsky line irrelevant to the question. New Mexico's formula is a working model of the drafting, and CRIPA's definition of an institution that "provides services on behalf of" a State is a working model of the coverage.
And the domestic use of the military should be tightened by legislation rather than left to emergency litigation. Trump v. Illinois read the federalisation statute narrowly on a stay application, over two dissenting opinions carrying three Justices, and expressly reserved the reviewability of the President's findings. Six days earlier the D.C. Circuit had allowed a separate deployment to continue under a different statute, 32 U.S.C. 502(f), while saying its own reasoning did not bind the merits panel. A statute that defined the predicate, required a finding reviewable on a record, set a duration and a reporting obligation, closed the Title 32 route to what Title 10 forbids, and made the Posse Comitatus Act enforceable by the people subjected to a violation of it would put the question where it belongs.
None of that is current law. We do not present any of it as current law, and on most of it the courts have gone the other way.