A class action has been brought to enjoin the City of Indianapolis from setting up roadblocks to catch drug offenders, a practice that the plaintiffs claim violates the Fourth Amendment. The plaintiffs' motion for a preliminary injunction was denied on the ground that the City's practice is lawful, precipitating this interlocutory appeal under 28 U.S.C. sec. 1292(a)(1). The legality of drug roadblocks has divided the other courts that have been asked to decide the issue. Compare United States v. Huguenin, 154 F.3d 547, 554-55 (6th Cir. 1998); United States v. Morales-Zamora, 974 F.2d 149 (10th Cir. 1992); Galberth v. United States, 590 A.2d 990 (D.C. 1991), and Wilson v. Commonwealth, 509 S.E.2d 540 (Va. App. 1999), which held them illegal, with Merrett v. Moore, 58 F.3d 1547 (11th Cir. 1995), and State v. Damask, 936 S.W.2d 565 (Mo. 1996), which held them legal. This is our first case. Because it was decided by the district court on a very skimpy stipulation of facts, our ruling on the legality of the City's program is necessarily tentative.

Six times between August and November of last year, the City's police department set up roadblocks on Indianapolis streets to catch drug offenders. A total of 1,161 cars were stopped at these roadblocks--for how long is unclear but the police endeavor to operate the checkpoints in such a manner that the stop does not exceed five minutes. During the stop, the police demand the driver's license and car registration, peer through the car's windows into its interior, and lead a drug-sniffing dog around the car. The stopping of the 1,161 vehicles resulted in 55 drug-related arrests, meaning that 5 percent of the total number of stops resulted in successful drug "hits," and 49 arrests for conduct unrelated to drugs, such as driving with an expired driver's license, for an overall hit rate of 9 percent. The City is continuing the program.

Stopping a car at a roadblock is a seizure within the meaning of the Fourth Amendment, Whren v. United States, 517 U.S. 806, 809-10 (1996); Michigan Dept. of State Police v. Sitz, 496 U.S. 444, 450 (1990), even though the sequel--the peering into the car windows and the sniffing of the car by the dog outside--does not rise to the level of a search as that term of the amendment has been interpreted by the Supreme Court. United States v. Place, 462 U.S. 696, 707 (1983); Texas v. Brown, 460 U.S. 730, 739-40 (1983); United States v. Ware, 914 F.2d 997, 1000 (7th Cir. 1990); United States v. Rodriguez-Morales, 929 F.2d 780, 788-89 (1st Cir. 1991). Whether the seizures effected by Indianapolis's drug roadblocks are reasonable may depend on whether reasonableness is to be assessed at the level of the entire program or of the individual stop. If the former, these roadblocks probably are legal, given the high "hit" rate and the only modestly intrusive character of the stops. In many Fourth Amendment contexts, the reasonableness of a practice is held to depend on the balance between its benefits (usually nonpecuniary) and its costs (ditto). E.g., Wyoming v. Houghton, 119 S. Ct. 1297, 1300 (1999); Whren v. United States, supra, 517 U.S. at 817; Skinner v. Railway Labor Executives' Ass'n, 489 U.S. 602, 619 (1989); Camara v. Municipal Court, 387 U.S. 523, 536-37 (1967); Dimeo v. Griffin, 943 F.2d 679, 681 (7th Cir. 1991) (en banc). The benefits of a random system of searches or seizures, such as vehicle stops pursuant to a roadblock system, are a function of, first, the probability that the stop will result in an arrest or a seizure of contraband or evidence of crime, and, second, the gain to the achievement of a lawful governmental goal that such an arrest or seizure will produce. The costs are a function of the harm that the stop will cause to the property or privacy of the people whose cars are stopped. In the case of Indianapolis's drug-roadblock program, the probability of a "hit" is high (vastly higher than, for example, the probability of a hit as a result of the screening of embarking passengers and their luggage at airports, see National Treasury Employees Union v. Von Raab, 489 U.S. 656, 675 n. 3 (1989)), and the deterrence of drug offenses produced by these hits advances the strong national, state, and local policy of discouraging the illegal use of controlled substances. The cost--in delay, anxiety, and invasion of privacy--to the drivers and passengers stopped for five minutes at a roadblock and subjected to a visual inspection of the interior and a sniff by a dog is small, though it is greater than the cost of the normal airport screening and (like that screening) is incurred in all stops while the benefit from the program is obtained only when there is a hit.

But courts do not usually assess reasonableness at the program level when they are dealing with searches related to general criminal law enforcement, see, e.g., Whren v. United States, supra, 517 U.S. at 810, rather than to primarily civil regulatory programs for the protection of health, safety, and the integrity of our borders. E.g., Michigan v. Tyler, 436 U.S. 499, 504-06 (1978); United States v. Martinez-Fuerte, 428 U.S. 543 (1976); Camara v. Municipal Court, supra; Platteville Area Apartment Ass'n v. City of Platteville, 179 F.3d 574 (7th Cir. 1999). Because it is infeasible to quantify the benefits and costs of most law enforcement programs, the program approach might well permit deep inroads into privacy. In high- crime areas of America's cities it might justify methods of policing that are associated with totalitarian nations. Cf. Brown v. Texas, 443 U.S. 47 (1979). One can imagine an argument that it would be reasonable in a drug-infested neighborhood to administer drug tests randomly to drivers and pedestrians. Although there is nothing in the text of the Fourth Amendment to prevent dragnet searches (read literally, the text requires only that searches and seizures be "reasonable" and confines the requirement of "probable cause" to searches or seizures made pursuant to warrant), the Supreme Court has insisted that "to be reasonable under the Fourth Amendment, a search ordinarily must be based on individualized suspicion of wrongdoing," save in cases of "special need" based on "concerns other than crime detection." Chandler v. Miller, 520 U.S. 305, 313-14 (1997) (emphasis added); see also Vernonia School District 47J v. Acton, 515 U.S. 646, 653 (1995); United States v. Martinez- Fuerte, supra, 428 U.S. at 560-61; Terry v. Ohio, 392 U.S. 1, 27 (1968). Program-level justifications for searches in support of specific regulatory programs do not carry over to general criminal law enforcement. See, e.g., Chandler v. Miller, supra, 520 U.S. at 313-14; New York v. Burger, 482 U.S. 691, 716 n. 27 (1987); Michigan v. Tyler, supra, 436 U.S. at 508; Donovan v. Dewey, 452 U.S. 594, 598 n. 6 (1981); Abel v. United States, 362 U.S. 217, 226 (1960); Michigan v. Clifford, 464 U.S. 287, 294 (1984) (plurality opinion); United States v. $124,570 U.S. Currency, 873 F.2d 1240, 1244 (9th Cir. 1989).

The qualification in "ordinarily" must not be overlooked. When the police establish a roadblock on a route that they know or strongly suspect is being used by a dangerous criminal to escape, the probability is high not only of apprehending the criminal but also of preventing him from engaging in further criminal, or otherwise hazardous, activity incidental to his escape. See, e.g., United States v. Harper, 617 F.2d 35, 40-41 (4th Cir. 1980). So the roadblock is allowed even though it is likely to "seize" some individuals who are not suspected of wrongdoing.

But here the roadblock is meant to intercept a completely random sample of drivers; there is neither probable cause nor articulable suspicion to stop any given driver. Even so, we can imagine cases in which, although the police do not suspect anyone, a roadblock or other dragnet method of criminal law enforcement would be reasonable. We may assume that if the Indianapolis police had a credible tip that a car loaded with dynamite and driven by an unidentified terrorist was en route to downtown Indianapolis, they would not be violating the Constitution if they blocked all the roads to the downtown area even though this would amount to stopping thousands of drivers without suspecting any one of them of criminal activity. See Maxwell v. City of New York, 102 F.3d 664 (2d Cir. 1996); Norwood v. Bain, 143 F.3d 843, 845-50 (4th Cir. 1998), aff'd (so far as pertinent), 166 F.3d 243, 245 (4th Cir. 1999) (en banc) (per curiam); United States v. Williams, 372 F. Supp. 65 (D. S. Dak. 1974); Brinegar v. United States, 338 U.S. 160, 183 (1949) (Jackson, J., dissenting); 4 Wayne R. LaFave, Search and Seizure: A Treatise on the Fourth Amendment sec. 9.6(a) (3d ed. 1996); American Law Institute, A Model Code of Pre-Arraignment Procedure sec. 110.2(2) (1975). When urgent considerations of the public safety require compromise with the normal principles constraining law enforcement, the normal principles may have to bend. The Constitution is not a suicide pact. But no such urgency has been shown here.

The Supreme Court has upheld the validity of roadblocks in less extreme cases, however, and it is on these that the City pitches its defense of its program. The Court upheld sobriety checkpoints--roadblocks at which drivers are checked for being under the influence of alcohol or (other) mind-altering drugs--in Michigan Dept. of State Police v. Sitz, 496 U.S. 444 (1990), and roadblocks designed to intercept illegal immigrants, in United States v. Martinez-Fuerte, supra. The Court has not, however, ever held or stated that all roadblock programs (even those not vulnerable to a charge of delegating too much discretion to individual police officers) are consistent with the Fourth Amendment. On the contrary, the amendment would be violated if "the roadblock was a pretext whereby evidence of narcotics violation might be uncovered in 'plain view' in the course of a check for driver's licenses." Texas v. Brown, supra, 460 U.S. at 743 (plurality opinion); see also United States v. Ortiz, 422 U.S. 891 (1975).

Randomized search programs have been upheld that involved the compelled provision of urine samples for drug testing of law enforcement officers, jockeys, railroad workers, and other classes of employee, e.g., Vernonia School District 47J v. Acton, supra; National Treasury Employees Union v. Von Raab, supra, 489 U.S. at 665-66; Skinner v. Railway Labor Executives' Ass'n, supra; Dimeo v. Griffin, supra, as well as administrative searches conducted without any basis to suspect any particular individual of wrongdoing, but rather pursuant to a program of inspections incidental to a general scheme of licensing or other regulation. E.g., New York v. Burger, supra; Michigan v. Tyler, supra; Camara v. Municipal Court, supra; In re Establishment Inspection of Skil Corp., 846 F.2d 1127 (7th Cir. 1988). But again the Court has not granted carte blanche.

Many of the cases we have cited do involve criminal prosecutions, however, and we must consider how they can be squared with the principle that the requirement of individualized suspicion is to be relaxed only on the basis of (as the Supreme Court said in the Chandler case, an example of a systematic search program that did not pass constitutional muster) "concerns other than crime detection." The answer is that the concern which lies behind the randomized or comprehensive systems of inspections or searches that have survived challenge under the Fourth Amendment is not primarily with catching crooks, but rather with securing the safety or efficiency of the activity in which the people who are searched are engaged. Consider employment drug tests for transport workers, as in the Skinner case. The employee who uses drugs will perform badly in his work, to the detriment of fellow employees or the general public; and the most effective preventive measure may be to test all or a random selection of applicants or employees. Similar are sobriety checkpoints, which are designed to protect other users of the road from the dangers posed by drunk drivers; administrative searches that are ancillary to concededly lawful systems of inspection; and the use of metal detectors and x-ray machines to screen entrants to government buildings and embarking air travelers. United States v. Herzbrun, 723 F.2d 773, 775-76 (11th Cir. 1984); United States v. Edwards, 498 F.2d 496 (2d Cir. 1974) (Friendly, J.); United States v. Davis, 482 F.2d 893, 908-10 (9th Cir. 1973). These cases rest on the commonsense principle that employers and other proprietors (such as the state as the owner of public roads), including the quasi- proprietor that is the government in heavily regulated industries, see, e.g., United States v. Biswell, 406 U.S. 311, 316-17 (1972), have a right to take reasonable measures to protect the safety and efficiency of their operations. These measures, moreover, usually make only limited inroads into privacy, because a person can avoid being searched or seized by avoiding the regulated activity, though we hesitate to put much weight on this point; people are unlikely to feel they can afford to "ground" themselves in order to avoid airport searches.

Indianapolis does not claim to be concerned with protecting highway safety against drivers high on drugs. Its program of drug roadblocks belongs to the genre of general programs of surveillance which invade privacy wholesale in order to discover evidence of crime. Imagine if the government set up a metal detector outside each person's home and required the person to step through it whenever he entered or left, in order to determine whether he was carrying a gun for which he lacked a permit. A principle that justified a drug roadblock would justify such surveillance.

We mentioned cases that allow the police or the Border Patrol to set up roadblocks to intercept illegal immigrants, a form of "contraband" to which illegal drugs might be analogized. Other cases allow custom searches of the luggage of people entering the United States. For examples of both types of case, see United States v. Montoya de Hernandez, 473 U.S. 531, 537 (1985); United States v. Martinez-Fuerte, supra; United States v. Ramsey, 431 U.S. 606, 616-19 (1977); United States v. Johnson, 991 F.2d 1287, 1290-92 (7th Cir. 1993). But such cases depend ultimately on sovereign powers over foreign relations, foreign commerce, citizenship, and immigration (see, e.g., Harisiades v. Shaughnessy, 342 U.S. 580, 586-89 (1952) (Jackson, J.)) that states and cities do not possess. Martinez-Fuerte involved searches well inland from the border, but the Court emphasized the infeasibility of preventing illegal immigration by border checks alone. 428 U.S. at 552.

We are mindful of the paradoxical implication that the Fourth Amendment, though originally limited to federal law enforcement, may pinch the states more tightly. But the paradox need not detain us. Indianapolis makes no attempt to defend its roadblocks on the basis that it is trying to exclude a harmful substance or dangerous persons. Though that may be the ultimate aim, the City concedes that its proximate goal is to catch drug offenders in the hope of incapacitating them, and deterring others, by criminal prosecution. The program has no regulatory purpose that might be compared to that of the immigration laws, which seek to exclude and deport illegal immigrants rather than just to prosecute them for criminal violations of the immigration laws.

It is true that in the course of looking for drugs in vehicles stopped at its drug roadblocks, the Indianapolis police often discover violations of the traffic laws. If the purpose of the roadblock program were to discover such violations, and if a program having such a purpose could be justified under the cases that allow searches and seizures without individualized suspicion of wrongdoing, then the seizure, in the course of such searches, of drugs that were in plain view would be lawful. Texas v. Brown, supra; United States v. Trevino, 60 F.3d 333, 336 (7th Cir. 1995). But the first "if" has not been shown. It is necessary in this regard to distinguish between two kinds of purpose, that of the program's designers and that of the police officers manning the roadblocks. The test for the lawfulness of a particular search or seizure is an objective one; the motives of the officer carrying out the search or seizure are irrelevant. Whren v. United States, supra, 517 U.S. at 811-13; United States v. Villamonte- Marquez, 462 U.S. 579, 584 n. 3 (1983); Scott v. United States, 436 U.S. 128, 138 (1978); cf. Graham v. Connor, 490 U.S. 386, 397 (1989). But the purpose behind the program is critical to its legality. The program must be a bona fide effort to implement an authorized regulatory policy rather than a pretext for a dragnet search for criminals. New York v. Burger, supra, 482 U.S. at 716 n. 27. "[T]he exemption from the need for probable cause (and warrant), which is accorded to searches made for the purpose of inventory or administrative regulation, is not accorded to searches that are not made for those purposes." Whren v. United States, supra, 517 U.S. at 811-12 (emphasis in original); see also Texas v. Brown, supra, 460 U.S. at 743 (plurality opinion). Leading a drug-sniffing dog around a car cannot be justified by reference to a desire to detect traffic violations, and so the use of the dog at the City's roadblocks shows--what is anyway not contested--that the purpose of the roadblocks is to catch drug offenders. We are not asked to decide whether, if the primary purpose were to detect drunken drivers, the dog could be added to the roadblock scenario on the theory that since a sniff is not a search, the incremental invasion of privacy would be negligible, or at least would not violate the Fourth Amendment.

It can be objected that requiring consideration of purpose injects too large an element of uncertainty into the interpretation of the amendment, and that purpose may be difficult to determine when it is corporate in nature. But law like politics is the art of the possible and often requires imperfect compromises. Inquiry into purpose is one method of identifying and banning the most flagrantly abusive governmental conduct without handcuffing government altogether. The alternative would be to rule that either all roadblocks are illegal or none are, which would be akin to punishing all killings identically because the "objective" fact is that someone has died.

To summarize, we have identified four exceptions to the principle that a search or seizure is forbidden by the Fourth Amendment unless there is a basis for believing that a particular search or seizure, as distinct from a program of universal or randomized searches or seizures, will yield evidence or fruits or instrumentalities of crime. The first exception, illustrated by the roadblock set up to catch a fleeing criminal, is where there is a suspect--the police have identified the criminal and have only to find him--but it is infeasible to avoid an indiscriminate search or seizure of other persons, persons not suspected of crime, as well. The second exception, illustrated by the hypothetical dynamite case, is where no specific person is under suspicion but the circumstances make it impossible to prevent a crime without an indiscriminate search. The third exception is the regulatory search, the objective of which is to protect a specific activity rather than to operate as an adjunct to general criminal law enforcement. The last exception is the prevention of illegal importation whether of persons (a power limited to the federal government, Saenz v. Roe, 119 S. Ct. 1518 (1999)) or of goods. On the basis of the record compiled in the preliminary-injunction proceedings--a record essentially limited to the parties' stipulation of facts--the Indianapolis roadblock program has not been shown to fit any of these exceptions, and thus the lawfulness of the program has not, as the district judge believed, been established. As that was the only ground on which she denied the preliminary injunction, her order cannot stand.

Whether there may be other grounds for denying the preliminary injunction, or whether on a fuller record the Indianapolis program might pass Fourth Amendment muster, are issues for the district court to decide in the first instance. We are not enthusiastic about the use of the Constitution to squelch experiments in dealing with serious social problems. The high hit rate of Indianapolis's roadblock scheme suggests that Indianapolis has placed the roadblocks in areas of the city in which drug use approaches epidemic proportions; and if so the roadblocks might be justified by reference to the second exception, as illustrated by such cases as Maxwell (involving a flurry of drive-by shootings), Norwood (threat of violence at a rally of motorcycle gangs), and Williams (Indian insurrection). But this is not argued either.

Reversed.

Easterbrook, Circuit Judge, dissenting.

Roadblocks in Indianapolis check for both driving and drug offenses. Someone driving a car without a license, or with drugs, can expect arrest. The program is spectacularly successful as roadblocks go; 9.4% of those stopped are arrested, with the reason equally divided between driving and drug crimes. Roadblocks with much lower rates of success have been held consistent with the fourth amendment. United States v. Martinez-Fuerte, 428 U.S. 543 (1976) (0.12% success rate); Michigan Department of State Police v. Sitz, 496 U.S. 444 (1990) (1.6% success rate). Yet my colleagues declare that the fourth amendment forbids what Indianapolis has done, because its primary purpose is to enforce the criminal laws. Martinez-Fuerte approved a roadblock to search for alien smuggling, a violation of a criminal law; Sitz approved a roadblock to search for drunk driving, a violation of a criminal law. So how can the fact that possessing drugs violates the criminal laws doom this program? One would suppose that our case is a fortiori from Sitz, because alcohol, marijuana, and cocaine all are drugs, any of which can impair a driver's performance. If the Constitution allows a roadblock to intercept alcohol users, how can it condemn a roadblock to intercept marijuana and cocaine users?

My colleagues' answer is that everything depends on the "primary" or "real" motive for the roadblock. Thus if Indianapolis set out to find people driving without licenses and only later added a dog to sniff for drugs (a step that does not entail a search or seizure of any kind, see United States v. Place, 462 U.S. 696 (1983)) in cars that already were stopped, then the program would pass constitutional muster. But if the City first decides to search for drugs, then adds license checks to make better use of the time while the dog does its work, then the program is invalid. If a city starts a license + drug program, then its validity depends on the primary motivation: if to search for people not legally entitled to drive, the program is valid; if to search for people not legally entitled to carry drugs, invalid. If a program is designed primarily to search for people using drugs in the car, and only secondarily to locate drugs in the trunk, then it is valid; if it is designed primarily to search for carried drugs, and only secondarily for ingested drugs, then it is invalid.

Why should the constitutionality of a roadblock program turn on what its promoters think (or the order in which its components were approved), rather than on what happens to the citizenry? Over and over, the Supreme Court says that the reasonableness inquiry under the fourth amendment is objective; it depends on what the police do, not on what they want or think. Whren v. United States, 517 U.S. 806, 811-13 (1996); Graham v. Connor, 490 U.S. 386, 397 (1989); Maryland v. Macon, 472 U.S. 463, 470-71 (1985); Scott v. United States, 436 U.S. 128, 136-38 (1978). The majority believes that things are otherwise when a program's design is in issue: "the purpose behind the program is critical to its legality. The program must be a bona fide effort to implement an authorized regulatory policy rather than a pretext for a dragnet search for criminals." 183 F.3d at 665 (emphasis in original). Where does "purpose" come into the fourth amendment? Not from its text; reasonableness fairly screams an objective inquiry. Not from its history; my colleagues do not mention the amendment's genesis. Not from the Supreme Court's cases. None of the opinions my colleagues cites requires a tour through the heads of the programs' sponsors. None suggests that the Constitution blesses a program in which a criminal-investigation component is added to a regulatory-enforcement one, while condemning an identical program in which the criminal- investigation component comes first. When the Supreme Court speaks of "regulatory" programs, such as the searches of business premises in New York v. Burger, 482 U.S. 691 (1987), or Marshall v. Barlow's, Inc., 436 U.S. 307 (1978), it asks what the programs do, not what the sponsors of the programs had in mind. So far as the fourth amendment is concerned, there is no difference between a roadblock originally designed to catch drug peddlers and also used to catch drunk drivers, and a roadblock originally designed to catch drunk drivers and also used to catch drug peddlers. Only observable differences in police behavior enter into the calculus of "reasonableness."

To be consistent, therefore, my colleagues should say that the fourth amendment would not permit the Michigan Department of State Police to add a drug-detection dog to the roadblock program sustained in Sitz. That conclusion would be so jarring, given received doctrine that a dog's sense of smell is not a search and requires no justification, that it could not be sustained. Yet if a dog may be added to the program sustained in Sitz, it can't matter to "reasonableness" whether some of the program's sponsors thought the dog more important than the breathalyzer. The trial envisaged by my colleagues--one at which officials of Indianapolis will testify about their motivations in approving the roadblock program, and the district judge must make credibility findings to resolve the fourth amendment objection--has no relation to the objective standard that a "reasonableness" benchmark demands.

What has led the majority to its search for the "primary purpose" behind a program, and thus to the startling conclusion that a given program may be constitutional or not today depending on what the Mayor thought last year, is its belief that "[p]rogram-level justifications for searches . . . do not carry over to general criminal law enforcement." 183 F.3d at 662. Indianapolis has adopted a program that is objectively reasonable given its minimal intrusion and substantial success (183 F.3d at 661), but my colleagues say that only "regulatory" programs may be justified in this manner. Without a distinction between criminal and regulatory searches, it would be impossible to understand Camara v. Municipal Court, 387 U.S. 523 (1967), which holds that a warrant may be issued without probable cause, despite the text of the fourth amendment, provided the government really isn't looking for anything in particular. But none of the Supreme Court's cases equates "regulatory search" with "regulatory purpose"; the Court's line is objective, while my colleagues' is subjective.

For every statement suggesting that criminal law enforcement may not be justified at the program level, it is easy to find another to the contrary--often in the same opinion. For example, my colleagues quote a few words from Chandler v. Miller, 520 U.S. 305, 313-14 (1997). That very opinion contains this passage (id. at 308): "Searches conducted without grounds for suspicion of particular individuals have been upheld, however, in 'certain limited circumstances.' See Treasury Employees v. Von Raab, 489 U.S. 656, 668 (1989). These circumstances include brief stops for questioning or observation at a fixed Border Patrol checkpoint, United States v. Martinez- Fuerte, 428 U.S. 543, 545-550, 566-567 (1976), or at a sobriety checkpoint, Michigan Dept. of State Police v. Sitz, 496 U.S. 444, 447, 455 (1990), and administrative inspections in 'closely regulated' businesses, New York v. Burger, 482 U.S. 691, 703-704 (1987)." This treats the roadblock and regulatory cases as independent, undermining the proposition that roadblocks are improper when they seek evidence of crime. Consider what Brown v. Texas, 443 U.S. 47, 51 (1979), had to say: "the Fourth Amendment requires that a seizure must be based on specific, objective facts indicating that society's legitimate interests require the seizure of the particular individual, or that the seizure must be carried out pursuant to a plan embodying explicit, neutral limitations on the conduct of individual officers." (Emphasis added.) Lots of avowedly criminal searches are justified at a program level. After making a custodial arrest, the police may conduct a complete search of the person, including the contents of any packages he may be carrying. United States v. Robinson, 414 U.S. 218 (1973). After entering a dwelling to make an arrest, police may conduct a visual "protective sweep" of other rooms to ensure that armed occupants there do not pose a risk. Maryland v. Buie, 494 U.S. 325 (1990). The list of searches and seizures justified in the aggregate, without regard to person-specific cause, is quite long. Roadblocks are just another example. See Wayne R. LaFave, 4 Search and Seizure sec.sec. 9.4(j), 9.6(b) (3d ed. 1996).

Interpretation of the fourth amendment is not a model of intellectual consistency. See Akhil Reed Amar, Fourth Amendment First Principles, 107 Harv. L. Rev. 757, 757-61 (1994). Cases create oodles of cubbyholes. My colleagues want to take as a fundamental doctrine that criminal investigations require person-specific cause, and conform all fourth amendment law to it. I suppose one could go in a different direction and treat auto searches or housing inspections as the paradigm. But our job as an inferior court is not to pick favorite passages from the hundreds of fourth amendment opinions the Supreme Court has issued, but to apply the principles devised for the most closely analogous cases. This is a roadblock case. To figure out how to handle a roadblock case, we must look at how the Supreme Court has handled other roadblock cases.

Neither Sitz nor Martinez-Fuerte involved a regulatory inspection, yet in each the Court assessed reasonableness at the program level. Michigan searched for drunk driving and the United States for alien smuggling. Because both programs were designed to enforce the criminal laws, a simple criminal-regulatory distinction won't fly. This impels my colleagues to proclaim a multivariate approach under which the reasonableness of criminal investigations will be assessed at the program level if some other condition holds--if there is a really pressing need (the search for a terrorist), if the reason for the stop is closely related to the dangers of driving (the search for drunk drivers), if the stop is justified by some "special" governmental power. So much for the organizing principle with which the majority begins. Why use a principle that disintegrates at first application?

Neither Sitz nor Martinez-Fuerte uses the approach my colleagues devise. Let's work through the line of reasoning that actually appears in these opinions.

First, the privacy interest of drivers is diminished relative to the interests of people at home or in the office. E.g., Wyoming v. Houghton, 119 S. Ct. 1297 (1999); Carroll v. United States, 267 U.S. 132 (1925). Carroll is close to the mark, because there the Court sustained the search of a vehicle for closed bottles of alcohol, which during Prohibition was treated the same way marijuana and cocaine are treated today. Recognition that drivers have a diminished expectation of privacy is missing from my colleagues' opinion. Instead they analogize a roadblock to a program under which "the government set up a metal detector outside each person's home and required the person to step through it whenever he entered or left, in order to determine whether he was carrying a gun for which he lacked a permit. A principle that justified a drug roadblock would justify such surveillance." 183 F.3d at 664. No, it wouldn't; the special treatment of automobiles provides a stopping point. One may question this treatment as an original matter, but it is entrenched in the Supreme Court's jurisprudence and is the key to analyzing roadblocks.

Second, the invasion of privacy at a roadblock is slight. Detention is short, the search superficial. Indeed, the use of a dog is not a "search" at all. A roadblock administered the way Indianapolis handles its program is less intrusive than the search of one's person and belongings at an airport, another familiar kind of "roadblock" (and one that, like the Indianapolis program, is designed to find evidence of crime, such as carrying weapons aboard an aircraft). Cases since Sitz and Martinez-Fuerte describe roadblocks in these terms--as reasonable in light of the minimal intrusion, not because they are "regulatory" or conducted with an approved "motive." See, e.g., United States v. Villamonte-Marquez, 462 U.S. 579, 587-88 (1988), and the passages from Chandler and Brown quoted above.

Third, a small invasion can be justified by aggregate success. "Probable cause," the level of suspicion adequate to support a custodial arrest that may last for days, is something less than a 50% likelihood; "reasonable suspicion," enough to support a frisk, means substantially less than "probable cause," see United States v. Chaidez, 919 F.2d 1193 (7th Cir. 1990); and as a brief stop is less intrusive than a frisk, an even lower probability of detecting crime suffices. Martinez-Fuerte holds a probability under 1% will do for a roadblock, and in Indianapolis the probability is much greater. (The probability of finding a bomb in an airport search must be less than one in ten million, given the volume of air traffic, but judges think this sufficient given the gravity of the evil being detected. Can my colleagues believe that drugs are so harmless that a 5% detection rate makes a search unreasonable even when the intrusion is slight?)

Fourth, the principal risk in allowing stops of vehicles without person-specific cause is that the officers will abuse the discretion thus conveyed. Some officers will stop people for the "offense" of DWB ("driving while black"); other officers, though pure of heart, may make stops at times or in places that leave the drivers apprehensive about their safety. Delaware v. Prouse, 440 U.S. 648 (1979), holds that the fourth amendment does not tolerate standardless discretion in making stops, and the Court suggested that Delaware use instead the roadblock or checkpoint approved in Martinez-Fuerte. Cf. Chicago v. Morales, 67 U.S.L.W. 4415 (U.S. June 10, 1999). Indianapolis was at pains to establish a rigorous protocol for its stops. Roadblocks are fixed; the number of cars selected for inspection is fixed; the procedure following a stop is fixed. So the concern that led to Prouse is missing, and the first three considerations show that the roadblock is reasonable.

Because Martinez-Fuerte so clearly involves criminal law enforcement, the majority creates still another special rule: "such cases depend ultimately on sovereign power over foreign relations, foreign commerce, citizenship, and immigration". 183 F.3d at 664. Any implication that the Court in Martinez-Fuerte relied on such a special power would be incorrect. The checkpoints were between 65 and 90 miles inland. The Court's point was not that Congress has extra powers to conduct domestic searches for aliens, but that "traffic-checking practices . . . appropriately are subject to less stringent constitutional safeguards." 428 U.S. at 555. Although Martinez- Fuerte involved immigration, never did the Court say that enforcement of this body of laws should be especially easy. See particularly 428 U.S. at 561-62:

[Here] we deal neither with searches nor with the sanctity of private dwellings, ordinarily afforded the most stringent Fourth Amendment protection. See, e.g., McDonald v. United States, 335 U.S. 451 (1948). As we have noted earlier, one's expectation of privacy in an automobile and of freedom in its operation are significantly different from the traditional expectation of privacy and freedom in one's residence. United States v. Ortiz, 422 U.S., at 896 n. 2; see Cardwell v. Lewis, 417 U.S. 583, 590-591 (1974) (plurality opinion). And the reasonableness of the procedures followed in making these checkpoint stops makes the resulting intrusion on the interests of motorists minimal. On the other hand, the purpose of the stops is legitimate and in the public interest, and the need for this enforcement technique is demonstrated by the records in the cases before us. Accordingly, we hold that the stops and questioning at issue may be made in the absence of any individualized suspicion at reasonably located checkpoints.

Everything the Court wrote about the checkpoints in Martinez-Fuerte can be said about the roadblocks in Indianapolis.

Treating Martinez-Fuerte as exemplifying a "border exception" or "immigration exception" to normal fourth amendment principles turns that amendment on its head. The fourth amendment reflects antipathy toward efforts to enforce the customs laws by searching warehouses for dutiable goods. See Leonard W. Levy, Original Intent and the Framers' Constitution 234-46 (1988); Telford Taylor, Search, Seizure and Surveillance, in Two Studies in Constitutional Interpretation 23-44 (1969). Inland searches based on the national government's power over immigration and importation (Martinez-Fuerte was not a border search) should be evaluated under the fourth amendment's normal "reasonableness" standard; instead my colleagues treat the immigration power as a reason to reduce the force of the fourth amendment. As a side effect, the national government, the object of the fourth amendment, winds up with greater freedom to conduct searches and seizures than state and local governments, which have been brought within the fourth amendment only indirectly. Any approach that carries this "paradoxical implication" (183 F.3d at 664) must be rejected. Giving the national government more leeway than the states to conduct searches, and treating immigration and customs searches as especially easy to justify, are so at war with the text and history of the Constitution that they cannot be sustained.

Searches by the national government pose a serious threat to the citizenry; searches by local governments pose less, because movement within the country is easy. Some cities enforce their drug laws by heavy reliance on spies, infiltrators, informers, turncoats, wiretaps, and nighttime searches where battering rams smash through doors; others may substitute more civil methods, such as roadblocks where the only imposition is a five-minute wait with man's best friend outside. Which of these is most like the "methods of policing that are associated with totalitarian nations" (183 F.3d at 662)? Police and prosecutors today ply people with favors so that friends and family become informers; lying about their identity, agents wheedle their way into strangers' confidences; they search trash in the hope of finding an incriminating scrap. Scaling back these tactics (none of which requires person-specific justification) in favor of roadblocks would make enforcement of the drug laws a good deal more reasonable. Or so at least the people may conclude.

One glory of a federal society is that the people may choose for themselves not only laws but also law-enforcement methods. State A may employ extra police to follow a high-probability- of-detection and low-sentence approach; State B may choose fewer police, fewer intrusions on privacy, but higher sentences for those who are caught. Each may be reasonable. Indianapolis selected a roadblock system, one that may catch any of its drivers. If this strikes the wrong balance, the people may throw out of office those who adopted it. Given the modest intrusion that roadblocks create for personal privacy, this is a legitimate choice for the public to make. The real threat to civil liberties comes from the national government, not from law-enforcement variations that can be avoided by driving a few miles to the east or west. Local governments should have more, not less, leeway than does the national government to decide how the tradeoff between privacy and effective law enforcement shall be handled.