法学资料大全二（英文）：刑事审判和行政规章

Harvard Law School Harvard Law School Public Law Research Paper No. 126 This paper can be downloaded without charge from the Social Science Research Network Electronic Paper Collection at: http://ssrn.com/abstract=783565 The Political Constitution of Criminal Justice William J. Stuntz *Professor, Harvard Law School. I owe a long list of colleagues and friends thanks for very helpful comments and conversations: Rachel Barkow, Dick Fallon, Heather Gerken, Jack Goldsmith, Joe Hoffmann, Pam Karlan, Orin Kerr, Mike Klarman, Daryl Levinson, Debra Livingston, Dana Mulhauser, Wes Oliver, John Rappaport, Dan Richman, Mike Seidman, Carol Steiker, Matthew Stephenson, and Adrian Vermeule. I also benefitted enormously from workshops at Columbia and Stanford Law Schools, the Hoffinger Colloquium at NYU Law School, and the Criminal Justice Roundtable at Harvard. Heather McNaught, Dana Mulhauser, Warren Postman, and John Rappaport provided excellent research assistance. Errors that remain are my responsibility. 1 THE POLITICAL CONSTITUTION OF CRIMINAL JUSTICE William J. Stuntz* INTRODUCTION I. CRIME, POLITICS, AND THE CONSTITUTION A. The Allocation of Power B. Legislative Politics and the Constitution 1. Policing and Procedure 2. Defining Crimes and Sentences 3. Spending C. Executive Branch Politics and the Constitution II. REPRESENTATION REINFORCEMENT: A BLUEPRINT A. The Content of Constitutional Regulation B. The Method of Constitutional Regulation III. A REFORM PROGRAM A. Policing the Police B. Defining Crimes and Adjudicating Guilt C. Punishment D. Federalism E. Imagining Constitutional Reform IV. CONCLUSION 1Jim Whitman uses this phrase as the title of his brilliant book on the history of criminal punishment in America and Europe. JAMES Q. WHITMAN, HARSH JUSTICE: CRIMINAL PUNISHMENT AND THE WIDENING DIVIDE BETWEEN AMERICA AND EUROPE (2003). 2See United States v. Booker, 125 S.Ct. 738 (2005); Blakely v. Washington, 124 S.Ct. 2531 (2004); Apprendi v. New Jersey, 530 U.S. 466 (2000). 3Cf. Blakely, 124 S.Ct. at 2546 (O’Connor, J., dissenting) (“While not a constitutional prohibition [of] guidelines schemes, the majority’s decision today exacts a substantial constitutional tax”). Until Blakely, Supreme Court Justices used the phrase 2 INTRODUCTION Large literatures discuss the constitutional law of criminal justice and the politics of crime. To date, no substantial literature addresses the relationship between the two. At first blush, that relationship seems straightforward: politicians ignore the interests of criminal suspects and defendants, so the Supreme Court steps in to protect those interests. On this view, politics is to constitutional law as a disease is to the medicine that cures it. America’s politics of crime is indeed diseased. But the metaphor may get causation backward. The constitutional proceduralism of the 1960s and after helped to create the harsh justice1 of the 1970s and after. Overcriminalization, excessive punishment, racially skewed drug enforcement, overfunding of prisons and underfunding of everything else — these familiar political problems are as much the consequences of constitutional regulation as the reasons for it. The medicine is reinforcing the disease. Political incentives are the mechanism. Constitutional law creates a series of political taxes and subsidies, making some kinds of legislation and law enforcement more expensive and others cheaper. Since the 1960s, the Supreme Court has regulated policing and trial procedure aggressively, while leaving substantive criminal law and (until the past few years)2 noncapital sentencing to the politicians. Consequently, legislators find it easy to expand criminal codes and raise sentences but harder to regulate policing and the trial process. These incentives apply to spending as well. Prison budgets receive a constitutional subsidy. Budgets for criminal adjudication and (especially) local police are subject to a constitutional tax.3 “constitutional tax” only in tax cases. 4No one knows how many searches and seizures the police conduct each year, but the number must be huge. According to the Statistical Abstract, 22.7 million people were the targets of motor vehicle stops in 1999. U.S. CENSUS BUREAU, STATISTICAL ABSTRACT OF THE UNITED STATES — 2002, at 195 No. 309 [hereinafter 2002 STATISTICAL ABSTRACT]. This is more than double the number of felony and misdemeanor charges filed per year, combined. See infra note 5. 5The figures on felony and misdemeanor charges are imprecise. The Justice Department lists arrests “by offense charged.” BUREAU OF JUSTICE STATISTICS, U.S. DEP’T OF JUSTICE, SOURCEBOOK OF CRIMINAL JUSTICE STATISTICS — 2003, at 353 tbl. 4.6 [hereinafter 2003 SOURCEBOOK]. The relevant table covers the majority of the United States but not the entire country, and some of the charged offenses may be categorized differently in different jurisdictions. Assuming the proportions hold nationally, there were at least 2.3 million felony charges in 2002. See id. at 344 tbl. 4.1, 353 tbl. 4.6. On the same assumption, the number of misdemeanor charges was at least 9 million. Id. 6In the 75 largest counties in 2002, 27% of those charged with felonies were not convicted, nearly always because charges were dropped. See 2003 SOURCEBOOK, supra note 5, at 457 tbl. 5.57. Of those who were convicted, 31% were not incarcerated; nearly all of those defendants were sentenced to probation. Id. at 458 tbl. 5.59. 7Paige M. Harrison & Allen J. Beck, Bureau of Justice Statistics, U.S. Dep’t of Justice, Prison and Jail Inmates at Midyear 2004, at 6 tbl. 7 (2005) (showing 686,437 prison admissions in 2003). 3 To see how perverse those taxes and subsidies are, it helps to visualize the criminal justice system as a giant funnel. Entering the broad end of the funnel are the tens of millions of men and women whom the police search or seize each year, most of them guilty of nothing worse than a traffic offense.4 Slide down the funnel, and that broad pool of suspects narrows considerably, producing a smaller pool of criminal defendants: about two million per year charged with felonies, and several million more charged with misdemeanors.5 Most of these are guilty, but not all, and a sizeable fraction of even the felons are let off with essentially no punishment.6 Slide down a bit farther and the pool narrows more, to the 700,000 who enter prison each year.7 Sadly, there are innocents here too — but presumably their number is small, and no one knows who they are. Notice the pattern: as one proceeds from policing to adjudication to punishment, the system’s targets grow fewer, less 8For a good discussion of the rule barring voting by convicted felons and challenges to it, see Pamela S. Karlan, Convictions and Doubts: Representation, Retribution, and the Debate over Felon Disenfranchisement, 56 STAN. L. REV. 1147 (2004). For an argument that few felons would vote even if the rule were changed, see Thomas J. Miles, Felon Disenfranchisement and Voter Turnout, 33 J. LEGAL STUD. 85 (2004). 9King was the Los Angeles motorist who, after a high-speed police chase, was beaten savagely by four officers; the beating was videotaped by a nearby civilian. The officers who beat King were tried and acquitted in state court, after which Los Angeles saw the second-worst urban riots in American history. Horton was the Massachusetts inmate who was freed on a prison furlough program during Michael Dukakis’s governorship. While on release, Horton committed assault, rape, and auto theft. In the 1988 presidential campaign, Vice President George Bush’s campaign manager, Lee Atwater, used Horton’s case to attack Governor Michael Dukakis for being soft on crime. 10From 1972 to 2001, overall police spending rose 148% in constant dollars. This figure is taken from 2003 SOURCEBOOK, supra note 5, at 5 tbl. 1.4, and CRIMINAL JUSTICE RESEARCH CENTER, U.S. DEP’T OF JUSTICE, SOURCEBOOK OF CRIMINAL JUSTICE STATISTICS — 1974, at 33 tbl.1.2 [hereinafter 1974 SOURCEBOOK]. The inflation adjustment is taken from U.S. CENSUS BUREAU, STATISTICAL ABSTRACT OF THE UNITED STATES: 2004-2005, 4 politically attractive, and less likely to vote.8 Constitutional law inverts the funnel. Suspects receive the most constitutional protection, criminal defendants get less, and prisoners get least of all. Politically speaking, that structure is upside-down. Tens of millions of mostly innocent criminal suspects can win political battles, at least sometimes. Two million mostly guilty felony defendants will find those battles harder to win. Several hundred thousand already-convicted prisoners may find victory impossible. To put the point in concrete terms, abused suspects like Rodney King have a lot more political appeal than prisoners like Willie Horton9 — and the Kings outnumber the Hortons by a considerable margin. Yet constitutional law chiefly protects the suspects, not the prisoners. Politicians are freest to regulate where regulation is most likely to be one-sided and punitive. A similar pattern characterizes criminal justice spending. Over the past generation — the time when constitutional law has played a large role in American criminal justice — state legislators and members of Congress have dramatically shifted the distribution of law enforcement dollars. Police spending has risen a little faster than other government spending.10 Spending on the adjudication process has at 461 No. 697 [hereinafter 2004-2005 STATISTICAL ABSTRACT]. All subsequent spending comparisons are adjusted for inflation, using this table. By comparison, total federal government spending rose 90% in constant dollars during the same period — less than police spending, but not a lot less. Compare id. at 308 tbl. 461 (showing 2001 federal spending of $1.8639 trillion) with U.S. CENSUS BUREAU, STATISTICAL ABSTRACT OF THE UNITED STATES — 1974, at 221 No. 359 (showing 1972 federal spending of $231.9 billion). 11From 1972 to 2001, spending on criminal adjudication rose 298% in constant dollars. See 2003 SOURCEBOOK, supra note 5, at 5 tbl. 1.4; 1974 SOURCEBOOK, supra note 10, at 33 tbl. 1.2. 12From 1972 to 2001, spending on corrections rose 456% in constant dollars. See 2003 SOURCEBOOK, supra note 5, at 5 tbl. 1.4; 1974 SOURCEBOOK, supra note 10, at 33 tbl. 1.2. 13The phrase belongs, of course, to the late John Hart Ely, for whom criminal procedure was for him a prime example of the phenomenon. See, e.g., JOHN HART ELY, 5 risen a great deal more.11 Spending on prisons has skyrocketed.12 Legislators spend where they can also govern. Constitutional law gives them little room to govern policing, more control over adjudication, and nearly unlimited authority to dole out punishment. Budget dollars flow accordingly. The perverse effects do not end with legislation. The constitutional law of policing widens the gap between the cost of investigating middle- and upper-class suspects and the cost of investigating poor ones. The law of trial procedure does the same thing for criminal prosecution. That gap, combined with law enforcement agencies’ budget pressures, pushes local police and prosecutors to focus too much attention on the crimes of the poor and too little on the crimes of the middle class. Given the demographics of poverty in America, that makes for a more racially discriminatory system — especially with respect to drug crime, where law enforcers’ discretion matters most. There is no way to run a test, but it seems likely that because of the constitutional rules that govern policing and trial procedure, criminal law is broader, sentencing rules are harsher, key criminal justice institutions are more underfunded, and the population of arrestees and defendants is more racially skewed than would otherwise be the case. These are odd results indeed for law that is supposed to be “representation reinforcing,” a corrective to politicians’ tendency to ignore the interests of suspects and defendants.13 DEMOCRACY AND DISTRUST 97 (1980) (calling the Fourth Amendment “a harbinger of the Equal Protection Clause”). The basic idea dates (at least) to Chief Justice Stone’s opinion in United States v. Carolene Products Co., 304 U.S. 144, 153 n. 4 (1938) (“[P]rejudice against discrete and insular minorities may be a special condition, which tends seriously to curtail the operation of those political processes ordinarily to be relied upon to protect minorities, and which may call for a correspondingly more searching judicial inquiry.”). 14See Michael C. Dorf & Charles F. Sabel, A Constitution of Democratic Experimentalism, 98 COLUM. L. REV 267 (1998). 6 What would it take to make constitutional law genuinely representation reinforcing? The answer is: radical change. Constitutional law that reinforced healthy politics would be different in nearly every respect from current law. Its animating goals would be to ensure legality and fight discrimination, not to protect privacy and extend jury power. It would focus more on macro incentives — the kind that drive legislation and the allocation of money and manpower — and less on the micro kind that dominate the law today. It would be more flexible and open to innovation, hence more political market-friendly. To use Michael Dorf’s and Charles Sabel’s helpful terminology, such law would be both more democratic and more experimentalist.14 Though reformed constitutional law would be much less intrusive than the current regime, it would also better protect criminal suspects and defendants, who get little benefit from the legal doctrines that allegedly safeguard their rights. My goal in this article is two-fold: to explain why radical change is needed, and to sketch what it might look like. Part I addresses the first of those two tasks, exploring the relationship between constitutional law and the politics of crime. Part II lays out the key characteristics of a more democratic experimentalist approach (hat tip to Dorf and Sabel) in this area. Part III applies these ideas, sketching a reform program for four areas: policing, adjudication and crime definition (legally two different topics, though they should be seen as one), punishment, and federalism. Part III concludes with a brief discussion of how these reforms might happen — and why they probably won’t. All proposals have their downsides; different readers will prefer different reform packages. Yet the basic argument does not depend on the particulars of any package. Current constitutional law makes the politics of criminal justice worse: more punitive, more racist, and less protective of individual liberty. That state of affairs is not inevitable. A different approach to constitutional law in this area would 15These propositions are central themes of the many books and articles on the American politics of crime. For a sampling, see KATHARINE BECKETT, MAKING CRIME PAY (1997); WILLIAM J. CHAMBLISS, POWER, POLITICS, AND CRIME (1999); DAVID GARLAND, THE CULTURE OF CONTROL: CRIME AND SOCIAL ORDER IN CONTEMPORARY SOCIETY (2001); STUART A. SCHEINGOLD, THE POLITICS OF LAW AND ORDER: STREET CRIME AND PUBLIC POLICY (1984); WHITMAN, supra note 1; LORD WINDLESHAM, POLITICS, PUNISHMENT, AND POPULISM (1998); FRANKLIN E. ZIMRING ET AL., PUNISHMENT AND DEMOCRACY: THREE STRIKES AND YOU’RE OUT IN CALIFORNIA (2001); Sara Sun Beale, What’s Law Got to Do With It? The Political, Social, Psychological, and Other Non-Legal Factors Influencing the Development of (Federal) Criminal Law, 1 BUFF. CRIM. L. REV. 23 (1997); Donald A. Dripps, Criminal Procedure, Footnote Four, and the Theory of Public Choice; Or, Why Don’t Legislatures Give a Damn About the Rights of the Accused?, 44 SYRACUSE L. REV. 1079 (1993). 16Most of the works cited in the preceding note take this position. One common version of the claim is that voters are subject to cognitive error when assessing the risk of crime, thanks to the sensationalist character of the media’s treatment of the issue. See, e.g., Rachel E. Barkow, Administering Crime, 52 UCLA L. REV. 715, 747-54 (2005) [hereinafter Barkow, Administering Crime], and sources cited therein. 17As John DiIulio points out, the 1960s saw both a huge increase in crime and a substantial drop in the incarceration rate — an unprecedented combination. Seen in that light, the punitive turn in American politics since 1970 seems a good deal less surprising. See John J. DiIulio, Jr., Keeping Crime on the Run, BLUEPRINT MAGAZINE, Fall 2000, available at http://www.ppionline.org/ndol/print.cfm?contentid=2153. If, during the 1970s and after, the voters decided to reject the advice of criminal justice experts in favor of punitive populism, as Frank Zimring, Gordon Hawkins, and Sam Kamin have argued, see ZIMRING ET AL., supra note 15, perhaps that is because in the 1960s, the experts failed the voters. Cf. GARLAND, supra note 15, at 65-68 (discussing radical criminology of the late 1960s that urged “expressive freedom and the liberation of individuals from arbitrary authority”). 7 mean, over time, healthier politics — and a healthier criminal justice system. I. CRIME, POLITICS, AND THE CONSTITUTION Scholars generally agree that American politics is too punitive, discriminatory, and unconcerned with the interests of the criminal justice system’s targets.15 Politicians, voters, and the media get the bulk of the blame16 — more than they deserve.17 The law gets less than it deserves. Like the rest of us, politicians 18See 2003 SOURCEBOOK, supra note 5, at 42 tbl. 1.27. The figure in the text includes 605,631 full-time, sworn officers in local police forces and sheriffs’ offices, plus 37,623 part-time officers in those same agencies and 43,413 “special jurisdiction” officers policing transit systems, public housing complexes, college campuses, and the like. 19Id. 20Id. at 69 tbl. 1.72. 21See id. at 344 tbl. 4.1, 389 tbl. 4.34. I subtracted the federal arrests as listed in table 4.34 from the total arrests recorded in table 4.1. 22Id. at 389 tbl. 4.34. 23Carol J. DeFrances, Bureau of Justice Statistics, U.S. Dep’t of Justice, Prosecutors in State Courts, 2001, at 2 (2002) [hereinafter DeFrances, Prosecutors]. 8 respond to incentives, and constitutional law creates bad ones: political taxes and subsidies that push politicians in precisely the directions scholars criticize. I begin with a brief discussion of the lines of authority over policing, criminal adjudication, and punishment. The discussion then turns to constitutional law’s perverse effects on the politics of legislation, and on the politics of policing and prosecution. A. The Allocation of Power The horizontal division of labor in America’s criminal justice system is simple. Legislators write the laws that define crimes and sentences. They also write budgets for the agencies that enforce those laws: police forces, prosecutors’ offices, public defenders’ offices, courts, and prisons. Police officers decide when and whom to search and arrest. Prosecutors decide whom to charge and how severely. Judges interpret the laws legislators write, and exercise whatever sentencing discretion those laws give them. The vertical allocation of power is more complex. Begin with the odd truth that local governments do most criminal law enforcement. Local police departments and county sheriffs’ offices employ roughly 700,000 police officers,18 compared to only 56,000 state troopers19 and 11,000 FBI agents.20 Local cops and state police together make some 13 million arrests;21 FBI agents make only 11,000.22 The data on prosecution is similar. Roughly 27,000 local prosecutors23 prosecute 1.1 million 24Matthew R. Durose & Patrick A. Langan, Bureau of Justice Statistics, U.S. Dep’t of Justice, Felony Sentences in State Courts, 2002, at 1 (2003) (estimating 1,051,000 state court felony convictions per year). Recall that the number of felony charges filed — well over 2 million — is much higher. See supra note 5. In some of those cases prosecutors drop charges, and others are bargained down to misdemeanors. 25See 2003 SOURCEBOOK, supra note 5, at 74 tbl. 1.79. 26Id. at 418 tbl. 5.17. 27Id. at 449 tbl. 5.44. 28Id. at 419 tbl. 5.18. 29Id. at 449 tbl. 5.44. 30Id. at 419 tbl. 5.18. A large fraction of the gun violators probably deserve to be classified as violent felons, but it is impossible to say how large a fraction. 31For the best discussion of the phenomenon to date, see Frank O. Bowman III, The Failure of the Federal Sentencing Guidelines: A Structural Analysis, 105 COLUM. L. REV. 1315, 1328-49 (2005). 9 felony cases each year.24 By contrast, the 5300 lawyers who work in United States Attorneys’ offices25 bring only 66,000 felony prosecutions per year.26 Each year sees roughly 320,000 state court felony drug convictions,27 compared with 25,000 in federal court.28 Local prosecutors convict 173,000 violent felons each year;29 their federal counterparts convict a mere 2600, plus another 4900 for weapons offenses.30 The bottom line is clear: Local law enforcement is the main course. Federal law enforcement is, by comparison, small potatoes. Nearly all local arrests and prosecutions are governed by state criminal codes and state sentencing rules. State legislators are responsible for those codes and rules, though they sometimes delegate a large measure of their sentencing power to sentencing commissions or judges. Congress defines federal crimes and sentences, with the help of the Federal Sentencing Commission. Over time, members of Congress have tended to exercise more power and the Commission less.31 The Supreme Court’s decision in United States v. Booker reshuffled the cards, shifting 32See 125 S.Ct. 738, 756 (Breyer, J., opinion of the Court in part) (retaining the Guidelines but with a new, more flexible reasonableness standard of review). 33In 1972, local governments paid 62% of the bill for criminal adjudication. By 2001 that share had fallen to 42%. These figures include prosecution and defense expenditures as well as judicial budgets. See 2003 SOURCEBOOK, supra note 5, at 5 tbl. 1.4; 1974 SOURCEBOOK, supra note 10, at 33 tbl. 1.2. 34See Carol J. DeFrances, Bureau of Justice Statistics, U.S. Dep’t of Justice, State- Funded Indigent Defense Services, 1999 (2001) [hereinafter DeFrances, Indigent Defense Services]. This report notes that 21 states pay all or nearly all the tab for indigent defense; in 9 states, county governments pay the full cost. Elsewhere, funding is divided between states and localities. Id. at 1. 35DeFrances, Prosecutors, supra note 23, at 4 (“Half of the prosecutors’ offices received 85% or more of their funding from the county government”); Carol J. DeFrances, Bureau of Justice Statistics, U.S. Dep’t of Justice, State Court Prosecutors in Small Districts, 2001, at 4 (2003) (“Half of the full-time offices serving districts with a population under 250,000 received 76% or more of their funding from the county government”). 36See infra note 38 and accompanying text. 37Local responsibility for jail budgets sometimes encourages state legislators to shift prisoners from penitentiaries, which the states fund, to jails. See David Boerner & Roxanne Lieb, Sentencing Reform in the Other Washington, 28 CRIME & JUST. 71, 94-96 (2001). 10 power from Congress to federal judges.32 Budgets are more straightforward. Congress and state legislatures decide how much to spend on federal and state law enforcement agencies, and on federal and state prisons. The same is mostly true of federal and state courts, though local governments pay a large share of the bill for state court systems.33 Public defenders are funded differently in different places — sometimes locally, sometimes state-wide, sometimes a mix of the two.34 Local governments pay the large majority of local prosecutors’ budgets,35 and an even larger share of local police budgets.36 Finally, the locals pay most of the cost of local jails, which house about one-third of America’s inmate population.37 Divide these various expenditures into three chronological phases — policing, adjudication, and punishment — and some interesting patterns emerge. Local governments pay 70% of the total cost of policing (and 92% of the cost of local 38The data cover fiscal year 2001. See 2003 SOURCEBOOK, supra note 5, at 5 tbl. 1.4. 39See supra note 10. 40See supra notes 11-12. 41The prison population in 1970 was 198,831. MARGARET WERNER CAHALAN, HISTORICAL CORRECTIONS STATISTICS IN THE UNITED STATES, 1850-1984, at 29 tbl. 3-2 (1986). By 2003, the number stood at 1,387,269. 2003 SOURCEBOOK, supra note 5, at 478 tbl. 6.1. 42In part, that is because federal and state dollars fund a steadily larger share of the system. In 1972, local governments paid for 56% of all criminal justice expenses; federal and state governments paid for the other 44%. 1974 SOURCEBOOK, supra note 10, at 33 tbl. 1.2. By 2001, those percentages were reversed. 2003 SOURCEBOOK, supra note 5, at 5 tbl. 1.4. 43See infra notes 92-94, 100-05 and accompanying text. 44The jury right applies where the maximum sentence exceeds six months’ incarceration. See Baldwin v. New York, 399 U.S. 117 (1970). It also applies in some other cases, never clearly specified by the Court. See, e.g., Lewis v. United States, 518 U.S. 322 11 policing, which is the kind that matters most), 42% of the cost of criminal adjudication, and 29% of the cost of corrections.38 Though spending on all three phases has risen over the past thirty years, spending in the latter two categories has risen substantially more. Nationwide, spending on policing rose 148% from 1972 to 2001.39 Spending on adjudication and corrections rose 298% and 456%, respectively.40 The prison population multiplied seven-fold.41 As those numbers suggest, federal and state spending is reshaping the system, changing its focus from policing to punishment.42 The final aspect of the allocation of power over American criminal justice is constitutional law itself. Fourth and Fifth Amendment law largely governs police searches and seizures and interrogation of suspects. State and federal statutes play a significant role in those areas, but most of those statutes deal with subjects that the Supreme Court has expressly declined to regulate.43 The story with respect to trial procedure is similar. Constitutional law governs the most important procedures: the scope of the right to trial by jury44 and jury selection,45 burdens of proof,46 the finality (1996). The right has recently been expanded to include a large fraction of sentencing factors. See Blakely v. Washington, 124 S.Ct. 2531 (2004); Apprendi v. New Jersey, 530 U.S. 466 (2000). 45The key doctrines are the Sixth Amendment right to a jury venire that represents a fair cross-section of the community, see Duren v. Missouri, 439 U.S. 357 (1979), and the Fourteenth Amendment right to be free from the discriminatory use of peremptory challenges, see Batson v. Kentucky, 476 U.S. 79 (1986). The latter requirement binds defendants as well as the government. See George v. McCollum, 505 U.S. 42 (1992). 46In re Winship, 397 U.S. 358 (1970). For the best discussions of the tortured caselaw on the meaning of Winship’s constitutionalized reasonable doubt rule, see Ronald J. Allen, Structuring Jury Decisionmaking in Criminal Cases: A Unified Constitutional Approach to Evidentiary Devices, 94 HARV. L. REV. 321 (1980); John Calvin Jeffries, Jr. & Paul B. Stephan III, Defenses, Presumptions, and Burden of Proof in the Criminal Law, 88 YALE L.J. 1325 (1979). 47On the basic rule that acquittals are unappealable, see Fong Foo v. United States, 369 U.S. 141 (1962). On the application of collateral estoppel to criminal cases, see Ashe v. Swenson, 397 U.S. 436 (1970). On the permissibility (or not) of motions to reconsider defense verdicts, see Smith v. Massachusetts, 125 S.Ct. 1129 (2005). 48See Crawford v. Washington, 541 U.S. 36 (2004). 49The past thirty-five years have seen a revolution in the nonconstitutional law of discovery, with vastly greater compelled disclosure than had been the norm. For the best account of the early stages of this revolution, see Robert P. Mosteller, Discovery Against the Defense: Tilting the Adversarial Balance, 74 CAL. L. REV. 1569 (1986). The best analyses of the current landscape appear in a pair of articles by John Douglass: Balancing Hearsay and Criminal Discovery, 68 FORDHAM L. REV. 2097 (2000); Beyond Admissibility: Real Confrontation, Virtual Cross-Examination, and the Right to Confront Hearsay, 67 GEO. WASH. L. REV. 191 (1999). 50The privilege matters in both directions. The discovery revolution mentioned in the preceding note was possible only because the Supreme Court decided that compelling disclosure from the defense did not automatically violate the Fifth Amendment. See 12 of verdicts,47 the right to confront opposing witnesses,48 and so on. Though nonconstitutional law plays a large role in shaping discovery,49 so do a pair of constitutional doctrines: the privilege against self-incrimination limits defendants’ disclosure,50 while Brady doctrine expands the government’s.51 Last but not least, the Williams v. Florida, 399 U.S. 78 (1970). On the other hand, the Fifth Amendment law of subpoenas substantially restricts compelled disclosure from white-collar defendants, either directly or through their lawyers. See, e.g., United States v. Hubbell, 530 U.S. 27 (2000). 51The canonical case is Brady v. Maryland, 373 U.S. 83 (1963). For two arguably inconsistent applications of Brady’s right to disclosure of material exculpatory evidence, see Strickler v. Greene, 527 U.S. 263 (1999); Kyles v. Whitley, 514 U.S. 419 (1995). 52E.g., Rompilla v. Beard, 125 S.Ct. 2456 (2005); Strickland v. Washington, 466 U.S. 668 (1984). 53In addition to the categories listed in this sentence, constitutional law bars vagrancy laws and seriously limits loitering statutes. See Chicago v. Morales, 527 U.S. 41 (1999); Papachristou v. Jacksonville, 405 U.S. 156 (1972). These “crimes” are chiefly used as means of granting police authority over low-level street disorder. Consequently, vagueness doctrine is best seen as an adjunct to Fourth Amendment law, not as a serious check on crime definition. 54E.g., Texas v. Johnson, 491 U.S. 397 (1989) (invalidating criminal flag burning statute). 55See Lawrence v. Texas, 539 U.S. 558 (2003) (invalidating criminal sodomy statute). 56See Roe v. Wade, 410 U.S. 113 (1973 (invalidating criminal abortion statute). 57530 U.S. 466 (2000) (requiring jury determination of all facts necessary to authorize defendant’s sentence). 58For an excellent analysis, see Kevin R. Reitz, The New Sentencing Conundrum: Policy and Constitutional Law at Cross-Purposes, 105 COLUM. L. REV. 1082 (2005) 13 Sixth Amendment right to the effective assistance of counsel regulates the quality of representation defendants receive.52 By contrast, constitutional law places few limits on crime definition,53 save for crimes that involve speech,54 consensual sex,55 or reproduction.56 (The large majority of crimes involve none of those things.) Until Apprendi v. New Jersey,57 the constitutional law of noncapital sentencing was similarly restrained. Today, Apprendi doctrine restricts sentencing procedure, but only where sentencing is done by rule.58 Substantive sentencing doctrine remains unregulated,59 as do both [hereinafter Reitz, Sentencing Conundrum]. 59See Ewing v. California, 538 U.S. 11 (2003); Harmelin v. Michigan, 501 U.S. 957 (1991). 60See Williams v. New York, 337 U.S. 241 (1949). For an analysis of the tension between this proposition and Apprendi, see Reitz, Sentencing Conundrum, supra note 58, at 1094-96. 61For much the best discussion, see Carol S. Steiker & Jordan M. Steiker, Sober Second Thoughts: Reflections on Two Decades of Constitutional Regulation of Capital Punishment, 109 HARV. L. REV. 355 (1995). For a recent — and exceptional — case of substantive regulation, see Roper v. Simmons, 125 S.Ct. 1183 (2005) (invalidating death penalty for offenders who were under 18 at the time of their crimes). 62See Whren v. United States, 517 U.S. 806 (1996). 63See Inmates of Attica Correctional Facility v. Rockefeller, 477 F.2d 375 (2d Cir. 1973). The leading Supreme Court cases do not involve criminal prosecution. See Heckler v. Chaney, 470 U.S. 821 (1985) (upholding FDA authority to decline to initiate enforcement proceedings against officials who administer lethal injections); Allen v. Wright, 468 U.S. 737 (1984) (denying standing to parents seeking to enforce IRS ban on tax exemptions for 14 substance and process where sentencing is discretionary.60 Broadly speaking, capital punishment follows the same pattern, regulating procedure heavily and substance lightly — though both kinds of regulation (especially procedure) are more intensive in death penalty cases.61 In three contexts — police discretion, prosecutorial discretion, and plea bargaining — the absence of constitutional regulation plays a central role in the system. Police officers are free to choose whom to search and arrest as long as they satisfy the relevant Fourth Amendment standards.62 Technically, they may not discriminate on the basis of race or other protected characteristics, but as a practical matter their discretion is nearly total. That matters little for crimes like murder or armed robbery. Investigating those crimes is politically mandatory, and the people and places investigated depend on the people and places victimized. But police discretion matters a great deal when it comes to drug crime. Illegal drugs are scattered through the population. Whom the police catch depends on where they look. Where they look is largely up to them. Prosecutorial discretion is, if anything, even broader.63 Official racially discriminatory private schools). Yet reading those cases, one senses that it is criminal prosecutors whose discretion the Court most wishes to protect. 64The key decisions are United States v. Armstrong, 517 U.S. 456 (1996), and McCleskey v. Kemp, 481 U.S. 279 (1987). 65See Wayte v. United States, 470 U.S. 598 (1985) (16 out of 674,000 draft nonregistrants prosecuted; Court held that prosecution did not violate due process or equal protection). 66Thus, the Court in Bordenkircher v. Hayes, 434 U.S. 357 (1978), did not inquire into the frequency (or not) of Kentucky prosecutors’ enforcement of the state’s three-strikes law, even though that law’s application to Paul Hayes seemed strange at best, outrageous at worst. See infra notes 278-280 and accompanying text. 67In Brady v. United States, 397 U.S. 742, 743-44 (1970), the defendant pled guilty in response to a threat to seek the death penalty. The statute authorizing the death penalty was later invalidated; the Supreme Court nevertheless held that the defendant’s plea was voluntary. Id. at 756-58. The prosecutor in Bordenkircher offered the defendant a plea with a five-year sentence; if the defendant refused, the prosecutor would seek a life sentence under a state three-strikes law. See 434 U.S. at 358-59. The defendant did refuse, the prosecutor carried out his threat, and the Supreme Court found the threat permissible. Id. at 364-65. 68The case most nearly on point is United States v. Ruiz, 536 U.S. 622 (2002), which rejected the defendant’s claim but left open the possibility that other claimants might fare better. 15 discrimination is barred but even where it exists, proof of discriminatory intent is impossible.64 Laws need not be regularly enforced. Prosecutors can charge a handful of defendants and ignore hundreds of thousands of violators.65 The same proposition appears to hold for sentencing rules.66 Which leads to the third example of a constitutional dog not barking: the (non-)regulation of plea bargaining. The Supreme Court has consistently refused to place even modest limits on the threats the government may use to extract guilty pleas.67 And thus far, the Court has declined to require that the government disclose exculpatory evidence prior to entry of a guilty plea.68 In criminal trials, the Constitution is omnipresent. In guilty pleas, it is nearly invisible. 69For the general argument, see Corinna Barrett Lain, Countermajoritarian Hero or Zero? Rethinking the Warren Court’s Role in the Criminal Procedure Revolution, 152 U. PA. L. REV. 1361 (2004). Kerr’s article nicely complements Lain’s: Kerr shows that legislative privacy protection has often worked better than the judicially mandated kind; along the way he offers a telling history of Congressional regulation of wiretapping, both before and during the Warren era. See Orin S. Kerr, The Fourth Amendment and New Technologies: Constitutional Myths and the Case for Caution, 102 MICH. L. REV. 801, 839- 50 (2004) [hereinafter Kerr, Constitutional Myths]. 16 B. Legislative Politics and the Constitution When the Supreme Court constitutionalized criminal procedure in the 1960s the conventional wisdom, evidently shared by the Justices, held that elected legislators would never adequately protect the interests of criminal suspects and defendants. That view seemed plausible at the time — though the contrary view was also plausible, as Corinna Lain and Orin Kerr have shown.69 Today, the Justices’ political prophecy looks either wrong or self-fulfilling. Members of Congress and state legislators can and do protect the interests of the criminal justice system’s targets. That protection follows some basic patterns. Legislators are more likely to craft fair procedures than to write fair substantive laws, and more likely to guard against abusive policing than to prevent excessive punishment. The Court drove legislators, along with the dollars they control, away from those areas where legislation might have done the most good (policing and procedure), and into those areas where it is bound to do the most harm (crime definition and sentencing). Three propositions are key. First, constitutional law makes legislative regulation of constitutionalized subjects politically costly. Second, for the past few decades, constitutional law has made legislation in constitutionally unregulated areas politically cheap. Third, constitutional law distorts state and federal spending. Congress and state legislatures spend where they can also govern, and constitutional law makes governing policing and trial procedure difficult. 1. Policing and Procedure Constitutional rules of policing and trial procedure are mandatory; legislators may add to but cannot replace them. At a glance, that seems obviously right — of course legislators cannot substitute their own preferred rules for constitutional mandates — but it is far from inevitable, and as a matter of political economy it is probably perverse. Default rules would encourage legislative protection of 70Earl Warren hinted that it might be a default. See Miranda v. Arizona, 384 U.S. 436,467 (1966) (inviting legislators to “exercise . . . their creative rule-making capacities”). But the hint was never serious, as the Court confirmed in Dickerson v. United States, 530 U.S. 428 (2000) (invalidating a congressional effort to overrule Miranda). 71Mandatory taping probably would be the equilibrium regulatory strategy were Miranda a default rule rather than a mandatory rule. Illinois, Maine, Texas, and the District of Columbia require the taping of interrogation sessions by statute; Alaska and Minnesota have a similar requirement via judicial decision. More than a dozen other states have considered mandatory taping bills in the past two years. See Joelle Anne Moreno, Faith- Based Miranda?: Why the New Missouri v. Seibert “Bad Faith” Test is a Terrible Idea, 47 ARIZ. L. REV. 395, 418 & nn. 176-201 (2005). It seems reasonable to suppose that, were Miranda’s restrictions relaxed, a great many more such laws would have been enacted. 72For a similar argument in a different context, see Einer Elhauge, Preference- Eliciting Statutory Default Rules, 102 COLUM. L. REV. 2162 (2002). 73For a discussion of how crime rates and resource constraints affect the choice of substitutes, see William J. Stuntz, The Uneasy Relationship Between Criminal Procedure 17 constitutional interests. If Miranda were a default rather than a mandatory rule,70 legislators who dislike Miranda might join with legislators who want better limits on police coercion to pass, say, laws requiring video- and audiotaping of interrogation sessions.71 That is how constitutional defaults work: they expand support for alternative regulatory strategies.72 Mandatory rules have the opposite effect. The existence of any given constitutional rule tends to reduce support for friendly legislation. Like the marginal benefit of most things, the marginal benefit of regulation declines; each new increment is worth less than the one before. One can imagine a number of different legal rules aimed at stopping coercive police interrogation: Miranda-style warnings, mandatory taping, a list of forbidden interrogation tactics, and no doubt there are others. Any of those rules would likely stop the most egregious police misconduct. Once one rule is in place, therefore, the benefit of adding another is bound to shrink. That pattern holds throughout criminal procedure. As marginal benefit falls, marginal cost rises. The government pays for criminal procedure rules in the coin of foregone arrests and convictions. When a particular rule turns winning cases into losers, prosecutors seek substitutes: other cases to take the place of the ones lost.73 Naturally, the highest-value substitutes go and Criminal Justice, 107 YALE L.J. 1, 23-27 (1997) [hereinafter Stuntz, Uneasy Relationship]. 74Roughly 20% of interrogated suspects invoke their Miranda rights, thus barring further questioning. Only 1% of suspects invoke their rights after questioning has begun — the rest of the invocations come when the Miranda warnings are first read to the suspect. For the leading study, see Richard A. Leo, The Impact of Miranda Revisited, J. CRIM. L. & CRIMINOLOGY 621, 653 (1996). 18 first. As regulation piles up and more cases fall, the substitutes are worth less and less. The cost of the lost cases grows. At some point, the substitutes run out and the cost jumps sharply — lost cases are simply lost; there are no replacements. So the marginal cost curve looks like a partially flattened S: [Figure 1] The more constitutional law regulates, the more likely legislators are to find themselves in the expensive portion of that curve. In theory, the costs of different regulatory options might overlap; if the overlap were large enough, the marginal cost of adding new regulation would be small. The reality is different. Miranda’s costs are felt when suspects invoke their rights immediately after hearing the famous warnings.74 Mandatory taping would lead to lost confessions in cases involving long, drawn-out questioning of vulnerable suspects. A list of banned tactics would cost the police confessions in cases in which those tactics would be especially useful, or perhaps especially tempting to the police. These sets of lost cases overlap only slightly. Take another example. One might regulate street stops by mandating individualized suspicion, as current Fourth Amendment law does, or by an antidiscrimination rule. Requiring individualized suspicion encourages the police to shift resources from middle-class neighborhoods, where drug crime takes place 75For a more extended argument, see William J. Stuntz, Race, Class, and Drugs, 98 COLUM. L. REV. 1795, 1804-15 (1998) [hereinafter Stuntz, Race and Drugs.] 76For the classic treatment, see WILLIAM JULIUS WILSON, THE TRULY DISADVANTAGED (1987). For more recent discussions, see Robert J. Sampson & William Julius Wilson, Toward a Theory of Race, Crime, and Urban Inequality, in CRIME AND INEQUALITY 41 (John Hagan & Ruth D. Peterson eds., 1995); Michael H. Schill & Susan M. Wachter, The Spatial Bias of Federal Housing Law and Policy: Concentrated Poverty in Urban America, 143 U. PA. L. REV. 1285 (1995). 772002 STATISTICAL ABSTRACT, supra note 4, at 195 No. 309. 19 indoors, to poor ones, where open-air drug markets and street gangs operate.75 (The government can more easily show the requisite suspicion when officers witness some portion of the crime.) Because of the strong correlation between race and urban poverty,76 a serious antidiscrimination rule would cut in the opposite direction, making street stops cheaper in wealthy areas and more expensive in poorer ones. These two forms of regulation have no substantial cost overlap. Or, consider trial procedure. A broad jury trial right costs the government convictions of defendants who are most appealing to jurors. Raising the burden of proof costs cases in which the government can meet a lower burden but not the higher one. Both rules seek greater accuracy, but they seek that goal in different ways, so their cost is felt in different cases. If constitutional law adopts one of these rules, the marginal cost of the second is higher than it would have been given a constitutional blank slate. The bottom line is simple. Constitutional regulation raises the political price of legislative regulation. But how much does the higher price matter? Perhaps the answer is, not much. Legislative regulation of policing and trial procedure might be politically impossible regardless of what the Supreme Court does. If so, the political costs of criminal procedure doctrines are beside the point. Both theory and evidence suggest otherwise. Legislators respond to powerful interest groups. Contrary to the conventional wisdom, criminal suspects are a powerful interest group. The police stop 23 million motorists per year.77 No one knows how many pedestrians the police stop, but that number is probably higher 78In New York City, the ratio of police stops to arrests is 8 to 1. OFFICE OF THE ATTORNEY GENERAL OF NEW YORK, THE NEW YORK CITY POLICE DEPARTMENT’S “STOP & FRISK” PRACTICES 111 (1999) [hereinafter NYPD STOPS AND FRISKS]. Nationwide, the ratio of traffic stops to arrests is less than 2 to 1. See supra note 5. 79See Janice Nadler, No Need to Shout: Bus Sweeps and the Psychology of Coercion, 2002 SUP. CT. REV. 153, 208-13. As the rest of Nadler’s article notes, most suspects do not, in fact, consent to what Fourth Amendment cases call “consent searches” — another reason for suspects to desire greater regulation. Especially since, as Nadler emphasizes, the large majority of targets of consent searches are innocent. Id. at 210 & n. 198. 80See supra notes 4-5 and accompanying text. 81See supra note 78. 20 still.78 Even if suspects vote at a lower rate than the general population (as they surely do), no politician can afford to ignore the interests of that many constituents. And there is evidence that even consensual police-citizen encounters are often traumatic,79 which suggests a fair amount of latent popular demand for regulating policing. If those doing the demanding were all criminals, legislators might shrug their shoulders and do nothing. But the data suggest that most suspects in cars are guilty of nothing more than a traffic offense,80 while most suspects on foot are guilty of nothing, period.81 Plus, legislators can protect some suspects without protecting all. Search and seizure regulation can be fine-grained. Drug suspects, white-collar suspects, and violent crime suspects need not be treated alike. And regulation can protect not only suspects but the police as well, given that regulatory legislation can be (and extensive regulation probably must be) coupled with spending. Advancing the interests of both the police and the victims of police misconduct sounds like a winning political package. Legislative regulation of the adjudication process — with the goal of protecting defendants’ interests, not just the government’s — is likewise more politically attractive than the conventional wisdom would have it. Compare criminal procedure legislation with legislation changing substantive criminal law. Adding new crimes makes it cheaper to prosecute offenders whom prosecutors can charge with the new crimes. Making the criminal process more efficient lowers the cost of prosecuting all offenders. Better crime definition raises prosecutors’ incentive to charge the newly defined crime properly. More accurate procedures raise 82See William J. Stuntz, The Pathological Politics of Criminal Law, 100 MICH. L. REV. 505, 557-58 (2001) [hereinafter Stuntz, Pathological Politics] (noting that courts, not legislatures, have been responsible for nearly all the significant doctrinal moves narrowing criminal liability). 83See Craig S. Lerner, Legislators as the “American Criminal Class”: Why Congress (Sometimes) Protects the Rights of Defendants, 2004 U. ILL. L. REV. 599. 84The McDade example is discussed extensively in Lerner’s article. In the wake of Representative Joseph McDade’s failed prosecution, Congress passed both the McDade and Hyde Amendments. The former applies state ethical rules to federal prosecutors, and thereby bars prosecutors from questioning represented defendants outside the presence of their lawyers. The latter offers compensation for some acquitted federal criminal defendants. See id. at 640-56. Lerner points to several other examples of this phenomenon, including the congressional decision to allow the independent counsel statute to lapse. See id. at 656-61. 21 prosecutors’ incentive to charge properly across the board. Changes in substantive criminal law affect those whose conduct is close to the borderline of criminal liability: those whose conduct is just barely or almost-but-not-quite criminal. Procedural legislation affects them too, but it also affects people whom the authorities wrongly suspect — some of whom are wholly innocent, not just dancing close to the legal line. Innocents suspected of others’ crimes are a more politically attractive interest group than borderline crooks. That proposition may explain why, historically, legislatures have been a good deal quicker to expand criminal procedure protections than to contract criminal liability.82 Craig Lerner notes that Congress regularly criminalizes acts that its members might commit. Senators and Representatives protect their interests not through careful definition of the relevant crimes, but through carefully regulated procedures.83 In theory, Joseph McDade might have persuaded his House colleagues to narrow the federal bribery laws under which he was prosecuted. (He beat the rap.) In practice, McDade knew that procedural limits on white-collar investigations were an easier political sell.84 Finally, if legislators care about deterring crime, as they surely do, theory suggests they should value procedures and budgets more than substantive laws. Consider a recently proposed reform of the law of sexual assault. Ian Ayres and Katharine Baker suggest criminalizing “reckless sex,” defined as failure to use a 85Ian Ayres & Katharine K. Baker, A Separate Crime of Reckless Sex, 72 U. CHI. L. REV. 599 (2005). 86Thus, they begin their article with a hypothetical based on the Kobe Bryant prosecution. Id. at 599-600. 87See id. at 630-58. 88See Kirk Johnson, As Accuser Balks, Prosecutors Drop Bryant Rape Case, N.Y. TIMES, Sept. 2, 2004, at A1. 89See Steve Chawkins, Stuart Pfeifer & Megan Garvey, Jackson Acquitted on All 10 Counts, L.A. TIMES, June 14, 2005, at A1. 90See Jonathan D. Glater, Jury Finds Ex-Tyco Lawyer Not Guilty of All Charges, N.Y. TIMES, July 16, 2004, at C1. 91Tom Tyler is the leading authority on this point. See TOM R. TYLER, WHY PEOPLE OBEY THE LAW (1990); E. ALLAN LIND & TOM R. TYLER, THE SOCIAL PSYCHOLOGY OF PROCEDURAL JUSTICE (1988). For a good, brief discussion of how Tyler’s insights apply to the criminal process, see Tracey L. Meares, Neal Katyal & Dan M. Kahan, Updating the 22 condom in an initial sexual encounter.85 Ayres and Baker defend their proposal as a winnable alternative to acquaintance rape prosecutions, especially for prosecutions of upper-class defendants;86 they also suggest that reckless sex laws would reduce the spread of sexually transmitted diseases.87 These are worthy goals. But no criminal prohibition is likely to achieve them. Prosecutors decline to prosecute upscale sexual assault not because substantive law immunizes the relevant conduct — it doesn’t — but because the criminal process provides a host of litigation opportunities for well- off defendants, while prosecutors’ budgets are tight. That is why wealthy defendants tend to defeat local district attorneys regardless of the crime charged: think of Kobe Bryant,88 Michael Jackson,89 or Tyco lawyer Mark Belnick.90 Changing that pattern would require a combination of streamlined procedures and bigger prosecutorial budgets, not broader criminal laws. One can put the point more generally. Accurate procedures make deterrent signals clearer and more powerful. Fair procedures (not the same thing) are more likely to be perceived as legitimate — and, if the relevant social science evidence is to be believed, perceptions of legitimacy are critical to the system’s ability to command obedience.91 Efficient procedures allow law enforcers to get more bang Study of Punishment, 56 STAN. L. REV. 1171, 1193-97 (2004). 92See Kerr, Constitutional Myths, supra note 69, at 850-57, and sources cited therein. 93See Smith v. Maryland, 442 U.S. 735 (1979) (government access to “pen registers” that recorded the phone numbers defendant dialed did not infringe defendant’s reasonable expectation of privacy); Zurcher v. Stanford Daily, 436 U.S. 547 (1978) (government may search newsrooms under the same Fourth Amendment standards that apply to other searches and seizures); United States v. Miller, 425 U.S. 435 (1976) (government access to defendant’s bank records did not infringe defendant’s reasonable expectation of privacy). 94For the most part, the privacy interests Congress has protected tend to be infringed through contacts with intermediaries — banks, internet service providers, telephone companies, and the like. The Court has repeatedly held that no constitutionally protected privacy interest is infringed when the police obtain information from a third party. See California v. Greenwood, 486 U.S. 35 (1988); United States v. White, 401 U.S. 745 (1971); cf. Rakas v. Illinois, 439 U.S. 128 (1978) (denying standing to contest searches other than those that infringe claimants’ reasonable privacy expectations). 95A recent headline on the ACLU’s website read: “House Patriot Act Proposals Would Make Sweeping Government Powers Permanent” and called the Administration’s proposed legislation “a power grab.” The website is http://www.aclu.com/ (last visited July 17, 2005). 23 for the buck. Legislative regulation of policing and procedure is perfectly natural. It is also common, where the Supreme Court has not already occupied the relevant field. Consider privacy-based regulation of police investigations. Federal statutes protect privacy interests in government records, newsrooms, phone conversations and numbers called, bank records, cable television, video rentals, and e-mail — and that is a partial list.92 In several of those areas, Congress acted shortly after the Supreme Court expressly declined to protect the relevant activity through the Fourth Amendment.93 As to most of the others, the Justices had not spoken, though their decisions suggested such privacy interests would get short shrift.94 Judging from the overheated rhetoric surrounding debates over its reauthorization, the Patriot Act might seem to have seriously undermined these protections.95 It 96For a good discussion, see Orin S. Kerr, Internet Surveillance Law After the USA Patriot Act: The Big Brother That Isn’t, 97 NW. U. L. REV. 607 (2003). 97Carl Hulse, House Rejects One Provision of Patriot Act, THE NEW YORK TIMES, June 16, 2005, at A1. 98See, e.g., Kyllo v. United States, 533 U.S. 27 (2001). 99See Christopher Slobogin, The Poverty Exception to the Fourth Amendment, 55 FLA. L. REV. 391 (2003); William J. Stuntz, The Distribution of Fourth Amendment Privacy, 67 GEO. WASH. L. REV. 1265 (1999) [hereinafter Stuntz, Distribution]. 10042 U.S.C. §14141. Section 14141 was part of the Violent Crime Control and Law Enforcement Act of 1994, Pub. L. No. 103-322, 108 Stat. 1796. 101For the best discussions to date, see Barbara E. Armacost, Organizational Culture and Police Misconduct, 72 GEO. WASH. L. REV. 453, 525-31 (2004); Debra Livingston, 24 hasn’t.96 Indeed, the pendulum seems to be swinging in the other direction today — witness the House of Representatives’ recent vote to protect records of libraries and bookstores.97 These federal statutes protect mostly (though not exclusively) middle- and upper-class suspects. Perhaps constitutional law’s role is to protect poorer targets of police attention. If so, the law fails badly. The Fourth Amendment protects “reasonable” privacy expectations.98 Expectations are deemed reasonable when they conform to existing social arrangements — people in houses enjoy more Fourth Amendment protection than apartment dwellers (who in turn enjoy more than the homeless), suspects in cars are better protected than suspects who use public transportation, and the list goes on.99 Legislative privacy protection is no more class- biased than its constitutional counterpart. Probably less so. Privacy aside, legislators tend to fill whatever regulatory space constitutional law leaves open — even when those whom the legislation aims to protect are poor, black, or both. That description fits the most important legal initiative of the past twenty years in the sphere of police regulation: section 14141, the federal statute that authorizes broad injunctive relief if federal government lawyers prove a pattern of constitutional violations.100 Thus far, that statute has been used to redress a litany of blue-collar police misconduct: racial profiling, corruption, police brutality, and concocting fake evidence.101 That statute was passed eleven years after the Supreme Police Reform and the Department of Justice: An Essay on Accountability, 2 BUFF. CRIM. L. REV. 815 (1999); Samuel Walker, The New Paradigm of Police Accountability: The U.S. Justice Department’s “Pattern or Practice” Suits in Context, 22 ST. LOUIS U. PUB. L. REV. 3 (2003). 102In Los Angeles v. Lyons, 461 U.S. 95 (1983), the Court ruled that a federal court could not enjoin unconstitutional police conduct even if the plaintiff showed a pattern of such conduct, unless the plaintiff personally faced a substantial risk of being victimized by that conduct in the future. Because no individual plaintiff could make that showing, Lyons effectively barred injunctions as a remedy for police misconduct. 103See, e.g., CAL. PENAL CODE § 13519.4 (banning profiling); COLO. REV. STAT. § 42-4-115 (requiring data collection); CONN. GEN. STAT. ANN. § 54-11 (both); KY. REV. STAT. ANN. § 15A.195 (banning profiling); MD. TRANSP. CODE ANN. § 25-113 (requiring data collection); 2000 MASS. ACTS 228 (requiring data collection); MO. REV. STAT. § 590.650 (requiring data collection); NEB. REV. STAT. § 20-504 (both); N.J. STAT. ANN. § 2C:30-5 (both); R.I. GEN. LAWS § 31-21.1 (both); TENN. CODE § 38-1-402 (requiring data collection); TEXAS CRIM. PROC. CODE ANN. art. 2.132 (both); WASH. REV. CODE § 43.43.480 (requiring data collection). For a compilation of state laws as of March 2005, along with pending legislative proposals, recent litigation, and some local government initiatives, see Police Foundation, Racial Profiling: The State of the Law, available at http://www.policefoundation.org/pdf/racialprofiling.pdf (listing 24 states that had enacted anti-profiling legislation). 104517 U.S. 806 (1996). Whren held that, if a police officer had sufficient cause to believe a suspect had violated a traffic rule, the officer could stop the suspect, even if the real motivation for the stop was suspicion of illegal drug activity. That holding barred nearly all constitutional challenges to profile-based traffic stops. 105See supra note 71. 25 Court held that victims of police brutality were not entitled to injunctive relief against the police departments that victimized them, thereby ruling out injunctions as a significant tool for addressing police misconduct.102 More than a dozen states have passed legislation banning racial profiling on the highways, requiring that police keep detailed records of traffic stops (including the demographics of detained motorists), or both.103 That state legislation followed the 1996 Supreme Court decision in Whren v. United States, effectively barring Fourth Amendment claims in profiling cases.104 Notwithstanding Miranda, several states mandate the taping of police interrogation sessions.105 These moves cannot fairly be characterized as protecting only wealthy 106Both the claims and the statutes are due in part to the work of the Innocence Project. See http://www.innocenceproject.org/. For a good, reasonably current discussion of the relevant state statutes, see Daina Borteck, Note, Pleas for DNA Testing: Why Lawmakers Should Amend State Postconviction DNA Testing Statutes to Apply to Prisoners who Pled Guilty, 25 CARDOZO L. REV. 1429, 1452-60 (2004). For a good discussion of the recently enacted federal DNA testing statute, see Ronald Weich, The Innocence Protection Act of 2004: A Small Step Forward and a Framework for Further Reforms, THE CHAMPION, Mar. 2005, at 28. For an excellent discussion of the issues these laws raise, see Seth F. Kreimer & David Rudovsky, Double Helix, Double Bind: Factual Innocence and Postconviction DNA Testing, 151 U. PA. L. REV. 547 (2002). 107See Herrera v. Collins, 506 U.S. 390 (1993). 108For an excellent discussion, see Darryl K. Brown, The Decline of Defense Counsel and the Rise of Accuracy in Criminal Adjudication, 93 CALIF. L. REV. ___ (forthcoming Dec. 2005). Notice that both crime labs and DNA testing are examples of procedural innovation that helps both the prosecution and the defense. Legislation defining crimes and sentences does not have this character. 109On indigent defense spending, see DeFrances, Indigent Defense Services, supra note 34, at 2 (showing 164% inflation-adjusted increase from 1982 to 1999 in 21 states studied). Police spending rose 99% in real dollars during the same period. See BUREAU OF JUSTICE STATISTICS, U.S. DEP’T OF JUSTICE, SOURCEBOOK OF CRIMINAL JUSTICE STATISTICS — 1985, at 2 tbl.1.1 [hereinafter 1985 SOURCEBOOK] (showing total police spending of $19.02 billion in 1982); BUREAU OF JUSTICE STATISTICS, U.S. DEP’T OF JUSTICE, SOURCEBOOK OF CRIMINAL JUSTICE STATISTICS — 2002, at 3-4 tbl. 1.2 (showing total police spending of $65.36 billion in 1999) [hereinafter 2002 SOURCEBOOK]. 26 suspects. A similar pattern holds for legislation governing trial procedure. Recent state and federal legislation protecting and encouraging DNA-based innocence claims106 benefits poor defendants, not rich ones, as the latter can afford to pay for such tests themselves. Those laws were enacted after the Supreme Court barred habeas corpus relief for freestanding innocence claims.107 State legislatures have created and funded state crime labs that foster better use of forensic evidence, partly for the purpose of identifying and exonerating innocent suspects.108 State spending on appointed counsel for indigent defendants increased more than two-and-one-half times in inflation-adjusted dollars between 1982 and 1999, a much bigger increase than police budgets saw during the same years.109 These phenomena are inconsistent with the 110Kevin Reitz is the leading authority on state sentencing legislation. For excellent overviews, see Kevin R. Reitz, Sentencing: Allocation of Authority, in 4 ENCYCLOPEDIA OF CRIME AND JUSTICE 1400 (Joshua Dressler et al. eds., 2002); Kevin R. Reitz, Sentencing: Guidelines, id. at 1429 [hereinafter Reitz, Guidelines]. 111On the contrary: As Rachel Barkow and Kevin Reitz have shown, state determinate sentencing legislation has led to much more moderate sentences than the federal variety. See Barkow, Administering Crime, supra note 16; Reitz, Guidelines, supra note 110. 112For a comprehensive and critical discussion, see Michelle J. Anderson, From Chastity Requirement to Sexuality License: Sexual Consent and a New Rape Shield Law, 70 GEO. WASH. L. REV. 51 (2002). 113See Robert P. Mosteller, Victims’ Rights, in 4 ENCYCLOPEDIA OF CRIME AND JUSTICE 1639 (Joshua Dressler et al. eds., 2002). 27 notion that state legislators and members of Congress attach no value to the procedural interests of criminal defendants, or value the interests of wealthy defendants only. The same is true of the mass of legislation since the mid-1970s dealing with sentencing procedure.110 Of all aspects of the criminal process, sentencing has seen the most legal innovation over the past three decades. Pre- Apprendi, it was also the least constitutionally regulated part of the process. And while federal sentencing legislation has seemed notably one-sided (more on that below), that criticism cannot fairly be leveled at state sentencing legislation.111 There is more. Constitutional law addresses only the interests of law enforcers and those whom they target. Of course — constitutional law is made by courts through litigation, and judges are trained to focus on the interests of the litigants. Legislators are not so limited. Consequently, criminal procedure legislation often takes account of interests that the litigants do not represent, as with rape shield laws112 and victims’ rights statutes.113 Most of the examples in the last few paragraphs are recent; the legislation in question has been enacted since the late 1960s, much of it in the past decade. In part, that is probably due to changed political preferences: Americans seem to value both privacy and process more than they once did. The rising power of black voters has 114Cf. Dan M. Kahan & Tracey L. Meares, Foreword: The Coming Crisis of Criminal Procedure, 86 GEO. L.J. 1153 (1998) (arguing that the rising power of urban black voters negates the need for constitutional regulation of policing). 115That is the date when Congress passed the federal wiretapping statute, Title III of the Omnibus Crime Control and Safe Streets Act of 1968, Pub. L. No. 90-351, 82 Stat. 213. 116An appendix to the Wickersham Commission’s report on the third degree lists statutes banning coercive police interrogation in 27 states. Zechariah Chafee, Walter Pollak & Carl S. Stern, The Third Degree, in WICKERSHAM COMMISSION, NO. 11: REPORT ON LAWLESSNESS IN LAW ENFORCEMENT 213-24 (1931). The first two decades of the twentieth century are the most commonly reported dates of enactment. For example, Colorado, Kentucky, Louisiana, Montana, Nevada, and Washington enacted their statutes between 1908 and 1912. Id. at 214, 217-19, 223. When the report was issued, constitutional regulation of police questioning was still several years in the future. See Brown v. Mississippi, 297 U.S. 278 (1936). 117Chafee et al., supra note 116, at 216. 118Id. at 215 (dating passage of Georgia ban as “in or before 1833"). 1192 REV. STAT. 708, §§ 14-15 (1829). I am grateful to Wes Oliver for uncovering this statute, and for exploring and explaining the fascinating history of the New York law of pretrial interrogation. 120367 U.S. 643 (1961). Mapp held that the Fourth Amendment’s exclusionary rule, which had previously applied only in federal cases, applied in state court as well. 121See Lain, supra note 69, at 1379-82. 28 no doubt also played an important role.114 But the history of legislation on policing and procedure did not begin in 1968.115 American police interrogation law began with state statutes banning the “third degree.” Most of those statutes were passed early in the twentieth century,116 though some came sooner: Illinois criminalized coercive interrogation in 1874,117 Georgia did so by the 1830s,118 and New York required that suspects be warned of their rights as early as 1829.119 The years before Mapp v. Ohio120 saw the exclusionary rule enacted by statute in several states.121 The first federal legislation regulating wiretapping — Congress tried to ban it but the 122The ban was part of the Communications Act, 48 Stat. 1064 (1934), later codified at 47 U.S.C. §605. The Justice Department read the statute to allow government wiretapping, as long as the taped conversations and their fruits were not later used in court. See Kerr, Constitutional Myths, supra note 69, at 845-46 & nn. 258-64. 123397 U.S. 358 (1970). For a critical discussion of the rise of the reasonable doubt standard, published 94 years before Winship was decided, see Some Rules of Evidence: Reasonable Doubt in Civil and Criminal Cases, 10 AM. L. REV. 642, 651-64 (1876). 124372 U.S. 335 (1963). The five states were Alabama, Mississippi,Florida, North Carolina, and South Carolina. See Lain, supra note 69, at 1391-92. 125See id., passim. This conclusion applies even to Miranda, which is generally seen as the most countermajoritarian of the Warren Court’s criminal procedure decisions. As Lain notes, the most reliable polling data showed that 65% of respondents agreed with the Court’s decision. Id. at 1421-24. A few months before Miranda was decided, one house of the New York legislature passed a Miranda-like statute by a lopsided vote. Id. at 1412 & n. 268. 29 Justice Department did not cooperate — was enacted in 1934.122 State law required proof beyond a reasonable doubt for most criminal convictions a century before the Supreme Court mandated it in In re Winship.123 Appointed counsel for indigent felony defendants was the policy of all but five states before the Supreme Court ordered it in Gideon v. Wainwright.124 As these examples suggest and as Corinna Lain’s excellent article on the subject demonstrates, the Warren Court’s criminal procedure decisions were, on the whole, majoritarian.125 The fact that legislation on subjects like search and seizure remedies, police interrogation, and burdens of proof grew less common after the 1960s, while legislation on constitutionally unregulated subjects grew more so, suggests that constitutional law has cast a large shadow on legislative regulating of policing and criminal adjudication. 2. Defining Crimes and Sentences Just as constitutional law makes some kinds of legislation politically more difficult, it makes other kinds politically easier. Crime definition and sentencing fall into the latter category — partly because policing and trial procedure fit the former. Different kinds of criminal legislation are substitutes for one another. Anything that makes one more costly necessarily makes others cheaper. 126For fuller versions of the argument in this paragraph, see Stuntz, Pathological Politics, supra note 82, at 529-57; William J. Stuntz, Plea Bargaining and Criminal Law’s Disappearing Shadow, 117 HARV. L. REV. 2548 (2004) [hereinafter Stuntz, Criminal Law’s Shadow]. 127WHITMAN, supra note 1, at 55. 30 Substitution effects are just the beginning. The law of criminal procedure makes criminal trials more expensive than they otherwise would be. At the same time, the law leaves prosecutors’ charging discretion and plea negotiation strategies unregulated. That combination — heavily regulated, expensive trials and cheap, regulation-free bargaining — obviously encourages bargaining. Less obviously, the same combination also encourages legislatures to expand criminal codes and to enact tougher sentencing rules. Because of the breadth of prosecutors’ charging discretion, legislators know that new crimes and sentences will not be enforced as written. Instead, a longer list of criminal prohibitions and sentencing rules allows prosecutors to extract more guilty pleas, thereby evading inconvenient trial procedures. Why should prosecutors’ convenience matter to legislators? If constitutional law were structured differently, it wouldn’t. Legislators have no reason to help prosecutors evade procedures that the legislators themselves crafted. When courts do the crafting, though, the odds that legislators will dislike some of the judges’ handiwork are high. Judicially mandated procedures make legislators and prosecutors natural allies. Prosecutors treat laws defining crimes and sentences as bargaining chips, while legislators liberally supply the chips. Together, they nullify most of the law of criminal procedure and change the character of the substantive law of crimes and sentences. Criminal law and the substantive law of sentencing become not a body of rules that define banned conduct and its consequences, but a means of extracting guilty pleas and expressing public outrage. Anger and a desire to drive hard bargains are not sentiments one associates with balanced lawmaking. Unsurprisingly, the bodies of law that define crimes and sentences have been anything but balanced.126 The academic trend these days is to blame democracy for that state of affairs. Jim Whitman contends that “[h]arshness and democratization go hand in hand,” that the punitive turn in American criminal justice has coincided with “a larger shift away from the judicial and toward the legislative.”127 The rise of guidelines sentencing, tougher rules for white-collar and drug offenses, three-strikes laws, the death penalty 128See id. at 43-57. 129See Franklin E. Zimring, Populism, Democratic Government, and the Decline of Expert Authority: Some Reflections on “Three Strikes” in California, 28 PAC. L.J. 243 (1996). 130See SAMUEL WALKER, POPULAR JUSTICE: A HISTORY OF AMERICAN CRIMINAL JUSTICE (2d ed. 1998). 131See GARLAND, supra note 15, at 96-102, 131-37. 31 for teenage murderers — all these are examples, in Whitman’s view.128 Other scholars make similar arguments. Frank Zimring attributes the run-up in America’s prison population to populist politics.129 Samuel Walker gives the same idea a different label: “popular justice.”130 David Garland’s wonderful book, The Culture of Control, emphasizes the politicization of criminal justice in the United States and Great Britain and the coincident rise of punishment in both countries.131 Professors Garland, Walker, Whitman, and Zimring are all fine scholars, and all are right — to a point. But their picture is incomplete. The last few decades did indeed see a “shift away from the judicial and toward the legislative” — but the shift was limited to substantive criminal law and sentencing, and it happened after an even more marked shift in the opposite direction in the law of policing and procedure: the 1960s Supreme Court decisions constitutionalizing those fields. The later shift is, at least in part, a response to the earlier one, an effort by legislators to recapture some of the authority constitutional law took from them. Too, legislators behave differently when writing substantive laws than when regulating policing and procedure. Crimes and sentences are every bit as harsh as contemporary scholarship suggests. Laws banning racial profiling, facilitating DNA-based innocence claims, and protecting the privacy of criminal suspects are a good deal more defendant- friendly. Democracy and severity do not always travel hand in hand. Two factors explain the difference between substantive and procedural legislation. The first goes to subject matter. Legislative regulation of policing and procedure tends to be about tradeoffs and allocating limited resources. In these areas, regulation is multidimensional: both the government and those it targets can benefit from better procedures and better funding for law enforcement agencies. Consequently, moderation is both natural and common. Substantive criminal law 132Cf. WHITMAN, supra note 1, at 21-22 (discussing moral philosophy tradition of justifying degrading punishments). As Whitman explains, that tradition has strong roots both in ancient Greek and Christian thought. Id. 133514 U.S. 549 (1995). For the pre-Lopez doctrine, see Perez v. United States, 402 U.S. 146, 149 (1971) (treating constitutional challenge to federal jurisdiction over a loan shark defendant as “substantial”). The discussion in Perez presupposed that Congress’s power over criminal law was much more limited than state power. 134See supra notes 92-97, 100-02 and accompanying text. Congress has broader power over policing than over substantive law because policing falls within Section Five of 32 and the substantive law of sentencing seem more conducive to moral absolutism.132 Legislation on those subjects is one-dimensional, zero-sum. Criminal liability either expands or contracts; sentences rise or fall. Policy choices that help both sides are hard to find. Compromise seems morally obtuse. Moderation is rare. The second factor concerns interest groups. The groups with the largest stake in substantive law are both smaller and less sympathetic than the groups with the largest stakes in policing and procedure. There are many more criminal suspects than criminal defendants, and a much larger proportion of the suspects are innocent of serious crime. Limiting police authority is thus politically easier than limiting the scope of criminal liability or reducing sentences. The same is true, albeit less dramatically, of procedural reform. If changing the process improves the system’s accuracy, the biggest winners are innocents wrongly suspected of serious crime — not a large group, but a very attractive one. By contrast, narrower crime definition helps those whose conduct places them near the borderline of criminal liability. More lenient sentences help only convicted criminals. These groups are politically unattractive. Legislators do a better job of protecting the interests of the more attractive claimants. The lesson is simple: crafting fair procedures, including police procedures, is politically easier than writing fair substantive laws. Yet constitutional law discourages procedural legislation and encourages the substantive kind. These tendencies are especially extreme in federal lawmaking, partly because Congressional power over criminal justice is more limited than state legislative power. Congress has no constitutional authority to define crimes and fix sentences for state court cases — a widely accepted proposition even before United States v. Lopez.133 Congress can regulate state and local police, and has done so.134 Until the Fourteenth Amendment. Most of federal criminal law, by contrast, is based on the commerce power, which is limited to conduct that is both interstate and commercial. See United States v. Morrison, 529 U.S. 598 (2000); Lopez, 514 U.S. 549. 135Before Boerne v. Flores, 521 U.S. 507 (1997), Congress’s power to enforce Fourteenth Amendment rights — including all the Bill of Rights provisions that the Supreme Court has incorporated into Fourteenth Amendment due process — was near-total. See Katzenbach v. Morgan, 384 U.S. 641 (1965). Boerne circumscribed that power, but Nevada Dep’t of Human Resources v. Hibbs, 538 U.S. 721 (2003) broadened it again, at least somewhat. Its current scope is unclear, but it is clear that Congress has significant legislative authority over state and local criminal procedure. For the best discussion of Section Five and its potential, see Robert C. Post & Reva B. Siegel, Legislative Constitutionalism and Section Five Power: Policentric Interpretation of the Family and Medical Leave Act, 112 YALE L.J. 1943 (2003). And Section Five is not the only source of Congressional power in this area. Congress can also write rules of policing and procedure and condition federal aid on compliance with those rules. See South Dakota v. Dole, 483 U.S. 203 (1987). 136In fiscal 2001, states spent $38.4 billion on corrections. 2003 SOURCEBOOK, supra note 5, at 5 tbl. 1.4. Total state spending for fiscal 2001 was $1.186 trillion. 2004- 2005 STATISTICAL ABSTRACT, supra note 10, at 272 No. 429. 137Federal corrections spending in fiscal 2001 was $5.2 billion. 2003 SOURCEBOOK, supra note 5, at 5 tbl. 1.4. The total federal budget for fiscal 2001 was $1.864 trillion. 2004-2005 STATISTICAL ABSTRACT, supra note 10, at 308 No. 461. 33 1997, Congress could also define state court criminal procedure if it wished.135 Even today, Congress has substantial power in that area. Federal legislation defining crimes and sentences, by contrast, applies only to federal cases. Power breeds responsibility, and its absence breeds the opposite. When Congress legislates about policing, it tends to legislate sensibly — because politically important interests (700,000 local cops, millions of innocent suspects) will be harmed if it doesn’t. The stakes are much lower when Congress defines federal crimes and sentences. Budget data are telling in this regard. State corrections spending is more than three percent of state budgets.136 Federal spending on corrections is less than three-tenths of one percent of the federal budget.137 The upshot is that Congress has much less incentive to worry about the severity of its substantive rules than does any 138This is the central point of Rachel Barkow’s excellent work on the political economy of sentencing. See Barkow, Administering Crime, supra note 16; Rachel E. Barkow, Federalism and the Politics of Sentencing, 105 COLUM. L. REV. 1276 (2005) [hereinafter Barkow, Federalism and Sentencing]. 139David A. Sklansky, Cocaine, Race, and Equal Protection, 47 STAN. L. REV. 1283, 1297 (1995) [hereinafter Sklansky, Cocaine and Race]. The phenomenon is not limited to drug sentencing, though it is especially marked there. As Frank Bowman notes, federal sentencing law contains more gradations than any state’s sentencing rules. More gradations, Bowman explains, mean more opportunities for sentencing legislation. See Bowman, supra note 31, at 1324-25, 1341-45. That would not be especially valuable if the point of such legislation was to make rules for litigated cases. But if the point is to make symbolic statements it is very valuable indeed, for it offers Congress a host of opportunities to announce how strongly it opposes various sorts of crime. 140For a discussion of Gotti’s criminal biography, see United States v. Locascio, 6 F.3d 924 (2d Cir. 1993). 34 state legislature — so Congress writes harsher sentencing rules than any state’s.138 When federal sentences for crack possession and distribution were under consideration, different members of Congress suggested different weight ratios between cocaine powder and crack — one-to-one, twenty-to-one, fifty-to-one. The ratio Congress enacted, one-hundred-to-one, was the highest anyone suggested.139 When severity is politically costless, one can expect to see severe laws. Severity is not costless when Congress legislates about procedure, so one sees less of it there. Bidding wars like the one that led to harsh federal crack sentences are not an inevitable consequence of democratic governance. Rather, they are the offspring of a bad marriage between an institutional design that makes federal crimes and sentences unimportant, and constitutional rules that make it easy for Congress to define crimes and sentences and hard to do anything else. The politicians just do what comes naturally. Often, what comes naturally is racism. To most American politicians of the past generation, crime has had a black face: Willie Horton, not John Gotti.140 Constitutional law did not create that phenomenon. By removing policing from the political center stage and placing crimes and sentences on it, however, the law might have caused politicians and voters to focus more attention on defendants and less on 141Diallo was an unarmed African immigrant shot by police near his New York apartment building in 1999. Michael Cooper, Officers in Bronx Fire 41 Shots, And An Unarmed Man Is Killed, N.Y. TIMES, Feb. 5, 1999, at A1. 142For the classic discussions of the criminal justice system’s failure to protect black crime victims, see Stephen L. Carter, When Victims Happen to Be Black, 97 YALE L.J. 420 (1988); Randall L. Kennedy, McCleskey v. Kemp: Race, Capital Punishment, and the Supreme Court, 101 HARV. L. REV. 1388 (1988). 143Dewey was America’s first celebrity prosecutor, the Rudy Giuliani of his day. He first won national fame as the Manhattan District Attorney in the late 1930s, when he successfully prosecuted Lucky Luciano. Later, as New York’s Governor, he supported and signed one of the most advanced anti-discrimination laws then in effect. See RICHARD NORTON SMITH, THOMAS E. DEWEY AND HIS TIMES 176-206, 443-48 (1982). Dewey’s top aide, Herbert Brownell, later became Dwight Eisenhower’s Attorney General, and in that capacity chose to intervene on the side of the plaintiffs in Brown v. Board of Education — a position about which Eisenhower had mixed feelings. STEPHEN E. AMBROSE, EISENHOWER: THE PRESIDENT 124-26, 142-43, 189-92 (1984). Kefauver’s name became a household word in 1950 and 1951, when he chaired a Senate committee investigating links among organized crime, gambling, and big-city Democratic machines. Kefauver ran for the Democratic presidential nomination in both 1952 and 1956. In the latter year, he was the leading candidate of civil rights supporters, and as such lost the crucial Florida primary to the then-more-conservative Adlai Stevenson. In the wake of Brown, Kefauver was one of only three Southern Senators (the others were his Tennessee colleague Albert Gore, Sr. and then-Majority Leader Lyndon Johnson of Texas) who refused to sign the segregationist Southern Manifesto, which urged defiance of the Court’s ruling. See ROBERT A. CARO, THE YEARS OF LYNDON JOHNSON: MASTER OF 35 suspects. Rodney King and Amadou Diallo had black faces too,141 but their images stand for something very different than Horton’s. A politics that paid more attention to them, and to suspects more generally, would be quicker to see blacks as victims — not just of police violence but also of crime itself142 — and less quick to see them as perpetrators. If black suspects and crime victims were as politically visible as black criminals, we might see more politicians willing to be tough on both crime and racism. That was once a common political package. Before the 1960s, two national politicians were prominently identified with crime-fighting: Republican Thomas E. Dewey and Democrat Estes Kefauver. Both were also leaders of their parties’ liberal, pro-civil rights wings.143 After the 1960s, politicians like Dewey and Kefauver all THE SENATE 785-88 (2002); JOSEPH BRUCE GORMAN, KEFAUVER: A POLITICAL BIOGRAPHY (1971). 144The death sentence was carried out just before the 1992 New Hampshire primary. The defendant, Rickey Ray Rector, had been sentenced by an all-white jury. “Logs at the prison show that in the days leading up to his execution, Rector was howling and barking like a dog, dancing, singing, laughing inappropriately, and saying that he was going to vote for Clinton.” Stephen B. Bright & Patrick J. Keenan, Judges and the Politics of Death: Deciding Between the Bill of Rights and the Next Election in Capital Cases, 75 B.U. L. REV. 760, 773 (1995). 145According to FBI data, the U.S. crime index rose from 1887 per 100,000 population in 1960 to 5282 in 1975. The murder rate rose from 5.1 in 1960 to 9.8 in 1974. Auto thefts rose from 183 in 1960 to 469 in 1975. U.S. CENSUS BUREAU, STATISTICAL ABSTRACT OF THE UNITED STATES — 1976, at 153 No. 252. Some of the index crime data may be subject to significant reporting errors, but that is not likely true of murders and auto thefts. 36 but disappeared. Bill Clinton, the closest contemporary American politics comes to the Dewey-Kefauver model, is not that close. To win the White House, Clinton had to uphold the execution of a mentally retarded black Arkansan — a graphic display of hostility to black criminals.144 Even in the more racist America of the 1950s, no national politician had to go to such lengths to prove his anti-crime bona fides. It would be a mistake to assign constitutional law the sole responsibility for that sad transformation. But the law does deserve some share of the blame. Maybe a lot. 3. Spending Members of Congress and state legislators were bound to spend a great deal more on criminal justice after the 1960s than before. Rising crime rates in the 1960s and early 1970s145 naturally produced rising demand for legislative action. Voters weren’t the only ones making demands; local governments were seeking more money as well. Criminal law enforcement is a redistributive service — crime disproportionately victimizes the poor, but the taxes that fund police departments and district attorneys’ offices are disproportionately paid by the rich — that local agencies largely provide. Local governments cannot easily fund redistributive services: if they tax too progressively, wealthy taxpayers will move to the neighboring town or 146For the classic discussion, see Charles M. Tiebout, A Pure Theory of Local Expenditures, 64 J. POL. ECON. 416 (1956). Tiebout’s theory is nicely summarized in John E. Chubb, The Political Economy of Federalism, 79 AM. POL. SCI. REV. 994, 995 (1985): “In Tiebout’s terms, competitive local governments have strong incentives to satisfy, at minimum cost, heterogeneous tastes in non-redistributive public services, but few incentives to tax progressively [or] assist the poor . . . .” 147See supra note 11-12, 40 and accompanying text. 148For Levitt’s analyses of the deterrent value of hiring more police and incarcerating more prisoners respectively, see Steven D. Levitt, Using Electoral Cycles in Police Hiring to Estimate the Effect of Police on Crime, 87 AM. ECON. REV. 270 (1997) [hereinafter Levitt, Police Hiring]; Steven D. Levitt, The Effect of Prison Population Size on Crime Rates: Evidence From Prison Overcrowding Litigation, 111 Q.J. ECON. 319 (1996). For Levitt’s analysis of the relative value of the two strategies, see Steven D. Levitt, Understanding Why Crime Fell in the 1990s: Four Factors that Explain the Decline and Seven that Do Not, 18 J. ECON. PERSP. 163, 176-79 (2004) [hereinafter Levitt, Understanding Why Crime Fell]. Levitt concludes that “a dollar spent on prisons yields an estimated crime reduction that is 20 percent less than a dollar spent on police.” Id. at 179. There is some evidence that given the high incarceration rates in American cities today, not only does incarceration fail to reduce crime substantially, it may increase it. See Jeffrey Fagan, Valerie West, & Jan Holland, Reciprocal Effects of Crime and Incarceration in New York City Neighborhoods, 30 FORDHAM URB. L.J. 1551 (2003). Cf. John J. Donohue III & Peter Siegelman, Allocating Resources Among Prisons and Social Programs in the Battle Against Crime, 27 J. LEGAL STUD. 1 (1998) (concluding that, at current high incarceration levels, more social spending is a better crime fighting strategy than more 37 county.146 When crime rises, local budget pressures rise too. After 1970, federal and state legislators were bound to hear appeals for more funds. So more legislation and fatter budgets were inevitable. But what would the legislation look like? Where would the budget dollars go? Today we know the answers: Congress and state legislatures expanded criminal codes, created new sentencing rules and procedures, appropriated more money for federal and state law enforcement agencies (but not much for local ones), and spent vastly more on prisons. Those answers seem natural, as the past always does. But they are not so natural as they seem. More money for prisons was easy to predict — but not five- and-one-half times more.147 If anything, local police seem a more attractive object of legislative largesse than prison systems. Steven Levitt has shown that marginal dollars spent on police do more to reduce crime than those spent on prison beds.148 incarceration). 149See 2003 SOURCEBOOK, supra note 5, at 5 tbl. 1.4. In 1972, local governments spent 69% of criminal justice dollars on policing, 19% on adjudication, and only 11% on corrections. See 1974 SOURCEBOOK, supra note 10, at 33 tbl. 1.2. 150See 2003 SOURCEBOOK, supra note 5, at 5 tbl. 1.4. 151See id. 152See id.; 1974 SOURCEBOOK, supra note 10, at 33 tbl.1.2. 153From 1972 to 2001, total spending on policing grew 148% (spending on local police rose much less), while adjudication spending grew 298%. See supra notes 10-11. 38 And the citizenry has more contact with cops than with prison guards; one might expect legislators to spend the most where budget dollars are most visible. Policing might have worked like public schools, with local governments providing the service but federal and state legislators paying most of the tab, and supplying a generous dollop of regulation on the side. (Call it “No Cop Left Behind.”) It didn’t happen that way. Instead, growing state and federal budget responsibility led to a growing focus on criminal punishment, not on crime prevention. Local governments spend 63% of criminal justice budget dollars on policing and 18% each on adjudication and corrections.149 State and federal spending is allocated very differently: 27% on policing, 26% on adjudication, and 47% on corrections.150 The policing percentage is misleadingly high, for federal and state policing dollars go mostly to federal and state police, not to the local cops who are responsible for day-to-day law enforcement. Only 15% of such spending is classified as “intergovernmental”: less than $4 billion, compared to total federal and state police spending of $25.5 billion and total local police spending of over $50 billion.151 And intergovernmental spending has actually fallen as a share of total criminal justice spending, from an already-modest 7% in 1972 to a mere 6% in 2001.152 That sounds surprising. So does this: spending on the adjudication process has grown substantially more than police budgets.153 Even spending on the one aspect of criminal litigation that most helps defendants — state-paid counsel for the 154The best recent study shows a 164% increase in spending on indigent defense between 1982 and 1999. Police spending rose 99% during those same years. See sources cited supra note 109. Notice that the comparison begins in 1982, nineteen years after Gideon v. Wainwright made appointed counsel for indigent defendants mandatory. So the increase in spending for indigent defense cannot be explained as a catch-up phenomenon; by 1982, the catching up should have been complete, or as complete as legislators wanted it to be. 155See 1974 SOURCEBOOK, supra note 10, at 33 tbl. 1.2. 1562003 SOURCEBOOK, supra note 5, at 5 tbl. 1.4. 157The best discussions of this proposition focus on federal spending on services administered by state and local governments. See, e.g., Chubb, supra note 146, at 997-99, 1004-11 (analyzing agency costs that attend federal aid to public education); David A. Super, Rethinking Fiscal Federalism, 118 HARV. L. REV. 2544, 2568-71, 2586-93 (2005) (exploring key features of the fiscal policy “game” federal and state governments play). 39 indigent — has risen more than police spending, by a large margin.154 In 1972, state legislatures spent twice as much on policing ($4.4 billion in 2001 dollars) as on adjudication ($2.2 billion).155 By 2001, states spent one-third more on adjudication ($14.4 billion) than on policing ($10.5 billion).156 Surely the police are more attractive objects of state beneficence than judges and lawyers — especially defense lawyers. Yet state legislatures have spent more and increased spending faster on the lawyers than on the cops. What explains these surprising numbers? The best answer, the one most consistent both with the data and with legislative behavior in other contexts, is that spending follows regulation: the ability to regulate makes spending more politically attractive, just as the ability to spend makes regulation more valuable. State legislators and members of Congress have spent where they could govern. Constitutional law made governing policing hard, governing litigation somewhat easier, and governing punishment very easy indeed. Legislators have spent accordingly. Take these points in turn. Agency costs always attend state and federal spending on government services.157 Legislators have preferences about how the relevant service is provided, but they cannot completely control those who provide it. Legislative oversight and regulation lower agency costs by increasing legislators’ control. That makes spending more valuable (at least to the legislators). 158See Joseph L. Hoffmann, Protecting the Innocent: The Massachusetts Governor’s Council Report, 95 J. CRIM. L. & CRIMINOLOGY 561, 570, 572-74 (2005). 159The proposal was enacted as the Violent Crime Control and Law Enforcement Act of 1994, Pub. L. No. 103-322, 108 Stat. 1796. Title I of the Act appropriated $8.8 billion over six years to help local police departments hire another 100,000 officers. For a good discussion of the Act, including the “100,000 cops” provision, see Harry A. Chernoff, Christopher M. Kelly, & John R. Kroger, The Politics of Crime, 33 HARV. J. LEGIS. 527 (1996). 160See supra notes 100-02, 159 and accompanying text. 40 Constitutional law, by contrast, raises agency costs. The more regulatory choices constitutional law takes off the table, the less control state and federal legislators can exercise over how their budget dollars are spent. That makes spending less valuable. The effect runs both ways: just as spending is worth more when accompanied by legislative regulation, regulation is worth more when accompanied by spending. Suppose legislators in a death penalty state want to minimize the risk that innocents will wind up housed on the state’s death row. The most obvious way to achieve that goal is to employ a higher burden of proof in capital cases, like the “no doubt” standard recently proposed in Massachusetts.158 That higher evidentiary standard will have some value standing alone, but it will mean more if it is coupled with bigger budgets for district attorneys and public defenders. More money leads to more and better personnel, who can spend more time and energy finding and sifting evidence in capital cases. Spending and regulation reinforce one another. The synergies may be especially large when the objects of spending and regulation are local police forces. By coupling aid to local police with regulation, legislators can help both the police and victims of police misconduct — a win-win proposition. The Clinton-era legislation that aimed to “put 100,000 cops on the street”159 was a perfect example. Congress made federal funds contingent on local police agencies embracing community policing and added section 14141, the provision authorizing injunctions against departments that violate citizens’ rights.160 That is a natural legislative package. But not a common one. From the late 1960s to the early 1980s, the Law Enforcement Assistance Administration funneled modest amounts of federal money 161The standard history of the LEAA is MALCOLM M. FEELEY & AUSTIN D. SARAT, THE POLICY DILEMMA: FEDERAL CRIME POLICY AND THE LAW ENFORCEMENT ASSISTANCE ADMINISTRATION, 1968-1978 (1980). 162In the late 1960s — shortly before the LEAA was established — the federal share of spending on primary and secondary education doubled, from 4% to 8%; in the following years, the federal share of education spending stayed roughly at that level as state spending soared. See, e.g., Joel D. Sherman, Changing Patterns of School Finance, 33 PROCEEDINGS OF THE ACADEMY OF POLITICAL SCIENCE 69, 69-72 (1978). The number of dollars involved is substantial. In fiscal 2001, the federal government spent $33 billion on primary and secondary education. See Crecilla Cohen & Frank Johnson, National Center for Education Statistics, U.S. Dep’t of Education, Revenues and Expenditures for Public Elementary and Secondary Education: School Year 2001-02, at 1. By comparison, federal spending on state and local criminal justice institutions was a mere $4.2 billion in 2001. 2003 SOURCEBOOK, supra note 5, at 5 tbl. 1.4. 163In fiscal 2001, the federal government spent $15 billion on policing, the states spent $12.5 billion, and local governments spent $50.7 billion. Only $2.5 billion of the federal dollars were “intergovernmental” (presumably the lion’s share of that money went to local police forces). States spent a mere $1.3 billion on aid to local police. See 2003 SOURCEBOOK, supra note 5, at 5 tbl. 1.4. 164The classic example is the federal No Child Left Behind Act; Pub. L. No. 107- 110, 115 Stat. 1425 (2002). For a good analysis, see James E. Ryan, The Perverse Incentives of the No Child Left Behind Act, 79 NYU L. REV. 932 (2004). But the basic phenomenon long predates NCLB. See, e.g., Chubb, supra note 146, at 999-1001; Sherman, supra note 162, at 69-70. 41 to local police with some mild regulatory strings attached.161 The LEAA never took off in the way that (for example) federal education spending did;162 it was canceled by one of the first Reagan Administration’s budget cuts. Aside from the LEAA and Clinton’s program, there have been no significant infusions of federal cash into local police forces. States spend even less on policing than does the federal government — and an even lower share of state spending goes to local cops.163 Had Fourth and Fifth Amendment law left legislators more space to regulate, those statistics might look very different. We might have seen more “100,000 cops”-style proposals, with budget dollars attached to regulatory strings — much the way education spending has been coupled with state and federal regulation of curricula, teacher hiring and tenure, and the like.164 165Current Fourth Amendment law allows police officers to stop vehicles whenever the officers reasonably believe traffic rules have been violated — a condition that, given the breadth of traffic rules and the laxity with which they are enforced, is almost always satisfied. Under Whren v. United States, 517 U.S. 806, 810-13 (1996), this standard applies regardless of the motive for the stop. White-collar investigations are subject to the same restrictions on police searches that apply when police investigate street or drug crime. But those restrictions matter little in the white-collar context, given the breadth of prosecutors’ subpoena power. See United States v. R. Enterprises, 498 U.S. 292, 301 (1991) (motion to quash a subpoena “must be denied unless the district court determines that there is no reasonable possibility that the category of materials the Government seeks will produce information relevant to the general subject of the grand jury’s investigation”). 166See sources cited supra note 103. 167See Lerner, supra note 83, at 640-56. 168Constitutional law does play a small (“tiny” might be a better word) negative role in crime definition. See, e.g., Lawrence v. Texas, 539 U.S. 558 (2003); Texas v. Johnson, 491 U.S. 397 (1989). Evidently, Texas is more constrained than other jurisdictions. 42 Instead, state legislators and members of Congress spent heavily on their own police agencies. Constitutional law reinforces that tendency. The Fourth Amendment leaves two forms of policing largely unregulated: traffic stops and white-collar investigations.165 The chief job of state police is to regulate highway traffic. The FBI is primarily responsible for white-collar investigations. In both instances, legislative regulation has taken the place of the constitutional kind. State anti-profiling laws constrain traffic stops,166 just as the federal McDade Amendment limits federal officials’ investigative power in white-collar cases.167 Again, legislative dollars travel together with legislative regulation. The same basic pattern holds with respect to spending on adjudication and corrections. Two bodies of law regulate criminal adjudication: the law of criminal procedure and substantive criminal law. The first is partly defined by legislation, but the most important parts are constitutionalized. The second is defined almost exclusively by state and federal legislation.168 Courts and legislatures thus share the task of regulating criminal litigation, while courts dominate regulation of policing. Since legislators have more control over litigation than over policing, they buy the former more readily than the latter. That extends even to indigent criminal defense, 169See DeFrances, Indigent Defense Services, supra note 34, at 2. 170For the classic discussion, see Stephen B. Bright, Counsel for the Poor: The Death Sentence Not for the Worst Crime but for the Worst Lawyer, 103 YALE L.J. 1835 (1994). 171See Rompilla v. Beard, 125 S.Ct. 2456 (2005); Wiggins v. Smith, 539 U.S. 510 (2003); and Williams v. Taylor, 529 U.S. 362 (2000). 172The key decisions were issued by two federal district judges: Holt v. Sarver, 309 F. Supp. 362 (E.D. Ark. 1970), and Ruiz v. Estelle, 503 F. Supp. 1265 (S.D. Tex. 1980). Holt and Ruiz are discussed in MALCOLM M. FEELEY & EDWARD L. RUBIN, JUDICIAL POLICYMAKING AND THE MODERN STATE: HOW COURTS REFORMED AMERICA’S PRISONS 51-95 (1998). 173452 U.S. 337 (1981). Rhodes held that double celling did not amount to unconstitutional overcrowding. Over the next five years, the Court barred or severely limited most constitutional claims prisoners might make against the prison system. See Davidson v. Cannon, 474 U.S. 344 (1986) (due process not violated where guards negligently allow one prisoner to attack another); Daniels v. Williams, 474 U.S. 327 (1986) (due process not violated where prison officials negligently damage or destroy prisoner’s property); Hudson v. Palmer, 468 U.S. 517 (1984) (prisoners have no constitutionally protected privacy). 174From 1972 to 1982, state and federal corrections spending rose 63% in inflation- adjusted dollars. See 1985 SOURCEBOOK, supra note 109, at 2 tbl.1.1; 1974 SOURCEBOOK, 43 funding for which rose substantially in the 1980s and 1990s.169 During those decades, ineffective assistance doctrine was infamously lax.170 Since 2000 the law has changed course, or seemed to.171 If the theory is right, this decade’s spending increases should be smaller. Until the last few years, constitutional law played no appreciable role in noncapital sentencing. (Obviously, it does now — but it is too soon to tell what effect Apprendi and its progeny will have on sentencing levels and prison budgets.) For a time, it seemed that Eighth Amendment doctrine would impose very substantial limits on prison conditions; a constitutional code of prison design and management seemed in the offing.172 Beginning in 1981 in Rhodes v. Chapman, the Supreme Court withdrew the threat.173 Spending on corrections accelerated sharply after that date.174 supra note 10, at 33 tbl. 1.2. From 1982 to 2001, it rose 292%. See 2003 SOURCEBOOK, supra note 5, at 5 tbl. 1.4; 1985 SOURCEBOOK, supra note 109, at 2 tbl.1.1. 175For a good summary discussion, see FINAL REPORT OF THE INDEPENDENT PANEL TO REVIEW DOD DETENTION OPERATIONS 10-13 (August 2004), available at http://www.defenselink.mil/news/Aug2004/d20040824finalreport.pdf. 176See Ronald Wright & Marc Miller, The Screening / Bargaining Tradeoff, 55 STAN. L. REV. 29 (2002). 177See Levitt, Police Hiring, supra note 148; Levitt, Understanding Why Crime Fell, supra note 148, at 176-77, 183-84. 178In 2001 dollars, total spending on local police rose from $31.3 billion in fiscal 1990 to $50.7 billion in fiscal 2001. Local governments provided $16.3 billion of the $19.4 billion in increased spending. The federal government spent $2.5 billion, and the states added the remaining $600 million. See BUREAU OF JUSTICE STATISTICS, U.S. DEP’T OF JUSTICE, SOURCEBOOK OF CRIMINAL JUSTICE STATISTICS — 1992, at 2 tbl. 1.1; 2003 SOURCEBOOK, supra note 5, at 5 tbl. 1.4. 44 All of which suggests that constitutional regulation deters, and its absence promotes, state and federal spending. That proposition is especially problematic given another: spending is itself a productive regulatory strategy. More money for policing means more and better trained police, and more personnel and better training make for less police violence and abuse. (Notice that lack of training and personnel contributed significantly to the abuses in Iraq’s Abu Ghraib prison.175) More defense lawyers allow for more defense litigation, leaving fewer defendants forced to plead guilty. More money for local prosecutors would permit wider use of innovative screening systems like the New Orleans regime that Marc Miller and Ron Wright made famous.176 If such screening systems work as advertised, fewer innocents would be punished for crimes that others committed. If constitutional law has deterred even a modest amount of spending on the local agencies that dominate America’s unsystematic criminal justice system, that system is bound to be less just than it could be, and should be. And crime rates are probably higher than they could and should be. Steven Levitt has shown that expanding city police departments played a large role in the 1990s crime drop.177 The expansion came from cities’ own budgets; federal and state aid played a small role.178 Had Congress and state legislatures spent some of what 179See CAHALAN, supra note 41, at 65 tbl. 3-31. 180See Michael Tonry, Obsolescence and Immanence in Penal Theory and Policy, 105 COLUM. L. REV. 1233, 1255 & tbl. 3 (2005). Another study finds that 45% of inmates are black and 16% Latino. See 2003 SOURCEBOOK, supra note 5, at 505 tbl. 6.34. The figures Tonry uses are better bases for cross-time comparison. 181CAHALAN, supra note 41, at 30 tbl. 3-3.. 1822003 SOURCEBOOK, supra note 5, at 501 tbl. 6.29. 183Id. at 505 tbl. 6.33. 184Id. 185See Robert Sampson & Janet Lauritsen, Racial and Ethnic Disparities in Crime and Criminal Justice in the United States, in 21 CRIME & JUST. 311 (1997). 186U.S. SENTENCING COMMISSION, 2001 SOURCEBOOK OF FEDERAL SENTENCING STATISTICS tbl. 34, available at http://www.ussc.gov/ANNRPT/2001/table34.pdf. 45 they lavished on prisons on local police instead, crime would have dropped even more. Had constitutional law left legislators more room to govern policing, that spending might have happened. C. Executive Branch Politics and the Constitution In 1960, 37% of America’s inmate population was African American.179 By 1995, that percentage had risen to 50; it has remained at that level since.180 In 1960, the incarceration rate — the number of prisoners per 100,000 population — stood at 126.181 In 2003, it was 482.182 Among black males, the number was 3405.183 Among black males in their late 20s, the rate exceeded 9000.184 To some degree, these numbers reflect different crime rates in different demographic groups.185 But they also reflect a substantial amount of discrimination. In fiscal year 2001, 83% of federal defendants convicted of selling or possessing crack were black; only 7% were white.186 Yet in 2003, the federal government’s own data showed that 47% of those who had used crack in the past month were white and only 41% were black; of those 187OFFICE OF APPLIED STUDIES, 2003 NATIONAL SURVEY ON DRUG USE & HEALTH, at tbl. 1.1A, available at http://oas.samhsa.gov/NHSDA/2k3tabs/LOTSect1pe.htm#DCrack. 188Civil Rights Restoration Act of 1991, Pub. L. No. 102-166, 105 Stat. 1071. 1892003 SOURCEBOOK, supra note 5, at 519 tbl. 6.56. 190Id. at 495 tbl. 6.20. 191Paige M. Harrison & Allen J. Beck, U.S. Dep’t of Justice, Prisoners in 2002, at 10 tbl.15 (2003), available at http://www.ojp.usdoj.gov/bjs/pub/pdf/p02.pdf. 192As of the late 1990s, most state laws treated crack and cocaine powder possession equally — a stark contrast to the federal 100:1 ratio. See Andrew N. Sacher, Note, Inequities of the Drug War: Legislative Discrimination on the Cocaine Battlefield, 19 CARDOZO L. REV. 1149, 1170 (1997) 46 who used the drug in the past year, the percentages were 55 and 29.187 These data should surprise. America in 1960 was still the land of Jim Crow. Even outside the South, police forces, prosecutors’ offices, and judges’ chambers were all-white or nearly so, and a lot of the white men who staffed those offices were openly racist. Those things are not true today. Legal change has reinforced racial progress. The law forbids discriminatory practices that it once required. Four years after passing the federal law that mandated draconian sentences in crack cases, a nearly identical Congress passed (and a Republican President signed) legislation extending affirmative action and expanding legal remedies in antidiscrimination suits.188 It would appear that America has become more racially just even as its criminal justice system has become less so. Consider one more trio of statistics. The ratio of black to white inmates incarcerated on federal drug charges is 5 to 6.189 In local jails, the analogous ratio is 5 to 3.190 In state prisons it is 5 to 2.191 By all accounts, federal drug sentencing laws are much more discriminatory than their state counterparts,192 yet the federal inmate population is much whiter. More puzzling still, a large portion of inmates in state prisons and local jails come from cities where African Americans cast a sizeable fraction of the jurisdiction’s votes. It seems more than a little strange that the 193Cf. Kahan & Meares, supra note 114 (arguing that African American power in large, high-crime cities vitiates the need for constitutional regulation of urban policing). 194The “generally accepted indigency rate” for state court felony cases in the early 1980s was 48%. BUREAU OF JUSTICE STATISTICS, U.S. DEP’T OF JUSTICE, NATIONAL CRIMINAL DEFENSE SYSTEMS STUDY 33 (1986). By 1992, it was 80%. Steven K. Smith & Carol J. DeFrances, Bureau of Justice Statistics, U.S. Dep’t of Justice, Indigent Defense 4 (1996). The best recent study does not report the percentage of cases receiving appointed counsel, but it does show that budgets for such counsel increased substantially — more than two-and-one-half times, in inflation-adjusted dollars — throughout the 1980s and 1990s. DeFrances, Indigent Defense Services, supra note 34, at 2. Given the budget rises, it seems likely that the indigency rate has continued to rise, at least modestly, since 1992. 195Slobogin, supra note 99; Stuntz, Distribution, supra note 99. 196See William J. Stuntz, Local Policing After the Terror, 111 YALE L.J. 2137, 2169- 76 (2002). 47 system’s worst racial tilts arise in places where blacks have the most political clout.193 What gives? The answer begins with another kind of discrimination: class-based, not race- based. About half of all criminal defendants were eligible for appointed counsel in 1980; by 1992 the figure was 80%, and it is probably higher today.194 In part, that change is due to rising docket pressure. Cost is a bigger factor in discretionary decisions of police officers and prosecutors than thirty or forty years ago, because there is not enough money to go around. When money is scarce, police and prosecutors tend to look for easy cases. Poor people are easier to convict than rich ones. The law likewise plays a role, widening the gap between the cost of arresting and prosecuting poor offenders and the cost of arresting and prosecuting their wealthier counterparts. Fourth Amendment privacy protection is skewed toward the middle and upper classes and away from the urban poor.195 The absence of serious Fourth Amendment regulation of the manner of searches and seizures196 — how violent and humiliating they are — has a similar discriminatory effect, since police are more likely to inflict physical and dignitary harms on poor suspects than on rich ones. Given the different characters of drug markets in poor city neighborhoods and 197See Stuntz, Race and Drugs, supra note 75, at 1804-15. 198John M. Dawson et al., Bureau of Justice Statistics, U.S. Dep’t of Justice, Prosecutors in State Courts, 1990, at 2 (1992); DeFrances, Prosecutors, supra note 23, at 2. 199State court felony filings more than doubled between 1978 and 1991. See NATIONAL CTR. FOR STATE COURTS, STATE COURT CASELOAD STATISTICS: ANNUAL REPORT 1984, at 189-90 tbl. 35 (1986) (showing 36% rise from 1978 to 1984); NATIONAL CTR. FOR STATE COURTS, STATE COURT CASELOAD STATISTICS: ANNUAL REPORT 1991, at 37 tbl. 1.25 (1993) (showing 51% rise from 1985 to 1991). Data for 1984-85 is missing; if the increase that year was comparable to the increases before and after, the total increase between 1978 and 1991 was roughly 125%. The number of state court felony convictions in 1990 was 829,344; the number of felony cases was presumably a little higher. BUREAU OF JUSTICE STATISTICS, U.S. DEP’T OF JUSTICE, SOURCEBOOK OF CRIMINAL JUSTICE STATISTICS — 1992, at 527 tbl. 5.49. In 2002, there were 1.1 million state court felony cases, nationwide. Durose & Langan, supra note 24, at 1. That suggests roughly another 25% increase since the early 1990s. The years between 1974 and 1978 presumably saw at least a modest increase. If so, felony filings have more than tripled since 1974. 48 elsewhere,197 these aspects of Fourth Amendment doctrine amount to a significant legal subsidy of drug policing in the inner city and a significant tax on such policing elsewhere. In a budget-constrained system, police go where they can get the most bang for the buck. The interaction of Fourth Amendment law and the realities of urban drug markets makes for a bigger bang in poor neighborhoods than in rich ones. The story with respect to the litigation process is similar. Constitutional rules of procedure make trials more costly, especially given the elaborate rules for selecting juries. More expensive trials lead to rising docket pressure. That would be true even in normal times, but the past few decades have been anything but normal. Steeply rising crime rates in the 1960s and early 1970s led to demands for more prosecutions, and budgets have not kept up with those demands. The number of local prosecutors grew from 17,000 in 1974 to 27,000 in 2001.198 Felony dockets tripled during the same period.199 Prosecutors have been forced to process more cases in the midst of rising trial costs. In a system that invited procedural innovation, states and localities might have experimented with different trial processes. That is what we did see in the realm of sentencing procedure, as different forms of determinate 200For an excellent, brief treatment, see Reitz, Guidelines, supra note 110. 201See Schill & Wachter, supra note 76, at 1287-88 & tbls. 1-2 (discussing study that showed blacks comprise 57% of the population of poor urban neighborhoods and 68% of the population of “distressed” neighborhoods). 202In fiscal 1999, there were 1056 bench trials, out of a total of 4352 federal criminal trials. BUREAU OF JUSTICE STATISTICS, U.S. DEP’T OF JUSTICE, SOURCEBOOK OF CRIMINAL JUSTICE STATISTICS — 2000, at tbl. 5.16. In fiscal 2001, the total number of trials (4356) was almost identical, but the number of bench trials had risen to 1536. 2003 SOURCEBOOK, supra note 5, at 418 tbl. 5.17. 203Jury trials fell from 3296 to 2820, from 76% of criminal trials to less than 65%. See sources cited supra note 202. 49 sentencing sprung up in the federal government and about two dozen states.200 But constitutional law stifled experimentation with different kinds of criminal trials. Instead, prosecutors were given a single cost-effective exit option: plea bargains. The predictable result has been more pleas. Which in turn encourages prosecutors to prosecute those defendants who are most easily induced to plead (and who have the most trouble making bail): the indigent. In the major cities that face the most constrained budgets, the largest portion of the indigent are African Americans.201 The same perverse incentives are at work in sentencing, the site of the Supreme Court’s latest constitutional initiative: Apprendi doctrine’s right to a jury trial on (some) sentencing factors. Federal bench trials rose by 45% in the two years after Apprendi was decided,202 and that was before litigants even knew whether Apprendi applied to federal sentencing. Since the number of federal criminal trials held constant, a rise in bench trials meant a decline in the number of jury trials203 — an odd result for a doctrine that aims to protect jury power. But odd results are the norm when it comes to constitutional rules of trial procedure. New Supreme Court mandates increase the returns from bargaining around those mandates. Consequently, the constitutionalization of criminal trials has made criminal trials rare events. Trials are the system’s Potemkin village, a piece of pretty scenery for display on Court TV while real cases, and lives, are disposed of more casually off-camera. That effect leads to another: a sharp decline in transparency. In a healthy system the law is what it appears to be. The rules applied in court are the same as the rules on the street, and courts apply those rules often enough that citizens can tell 204For a brilliant but too celebratory account of this process in federal cases, see Gerard E. Lynch, Our Administrative System of Criminal Justice, 66 FORDHAM L. REV. 2117 (1998). There is less to celebrate in state cases because resources are more constrained there. 205See Stuntz, Criminal Law’s Shadow, supra note 127, at 2549, 2560-62, 2569. 50 what they are. In our system, substantive law is a tool for evading inconvenient procedures, and courtrooms are used for guilty pleas. Criminal punishment is allocated behind closed doors, where the lawyers dicker over charges and sentences.204 Criminal codes do not describe the behavior that will actually land one in a prison cell, and sentencing rules do not accurately predict how long one will stay there. Instead, the law of crimes and sentences serves as a menu of threats for police and bargaining options for prosecutors.205 The real law — the law that governs individual cases — arises from discretionary decisions to order off the menu: police officers’ arrests and lawyers’ plea bargains. That law is invisible to outsiders. Invisibility, in turn, makes effective political regulation difficult. Not only does the current system of constitutional regulation aggravate the political diseases it sets out to cure. Worse, the law makes political cures harder to effect. II. REPRESENTATION REINFORCEMENT: A BLUEPRINT How can things be set right? How can constitutional law and politics work together to produce a more just criminal justice system? How can “representation reinforcement” become a reality rather than a slogan? The answer requires fundamental change, both in constitutional law’s content and in its method — both in the interests the law protects and in the means by which it protects them. With regard to content, John Hart Ely had it about right: constitutional law adds the most value when it advances interests that the political process will not advance on its own. With respect to the method of constitutional regulation, the key changes involve generating and using information. Decisionmakers must know enough to make good decisions, working feedback systems must be in place so that errors can be corrected, and the system must produce healthy incentives for its intended targets. A. The Content of Constitutional Regulation 206In retrospect, the chief exception is Brady v. Maryland, 373 U.S. 83 (1963), which required prosecutors to turn over material exculpatory evidence. At the time, a trio of cases concerning the permissible scope of substantive criminal law — Lambert v. California, 355 U.S. 225 (1957), Robinson v. California, 370 U.S. 660 (1962), and Griswold v. Connecticut, 381 U.S. 479 (1965) — looked more important. Griswold turned out to be very important indeed, but not for the scope of substantive criminal law. Lambert and Robinson have produced no substantial constitutional offspring. 207Compare Griffin v. Illinois, 351 U.S. 12 (1956) (equal protection requires provision of free trial transcript to indigent defendants), and Douglas v. California, 372 353 1963 (equal protection requires provision of state-paid counsel on direct appeal for indigent defendants), with Swain v. Alabama, 380 U.S. 202 (1965) (setting impossibly high proof standards for equal protection challenges to prosecutors’ use of peremptory challenges). Swain is infamous, but what mattered more was what the Court did not do: decide to rein in discriminatory policing and prosecution. That non-event had a large effect on later decisions in McCleskey v. Kemp, 481 U.S. 279 (1987) and United States v. Armstrong, 517 U.S. 456 (1996). 208Those discretionary processes — charging discretion by professional prosecutors and plea bargaining — did not exist when the Bill of Rights was written and ratified. They did exist, at least in nascent form, by the time the Fourteenth Amendment was enacted. For the best discussion of their rise, see GEORGE FISHER, PLEA BARGAINING’S TRIUMPH (2003). 51 The constitutional law of criminal justice rests on two textual sources: the Bill of Rights and the Fourteenth Amendment. One of those texts is better suited to the task. Unfortunately, the other dominates the doctrine. The problem begins with the incorporation decisions of the 1960s. In the realm of criminal justice, Earl Warren and his colleagues did little to expand due process206 and even less to guarantee “the equal protection of the laws.”207 Instead, they used the Fourth, Fifth, and Sixth Amendments to ratchet up regulation of state and local criminal processes. That choice had three perverse consequences. First, it made the constitutional law of criminal justice primarily about criminal procedure. Procedural regulation inevitably encourages substantive overreaching, as legislators give prosecutors more threats to use to bargain around the procedures legislators dislike. The end result is procedures that don’t matter, and substantive law that gives more power to law enforcers who already have too much of it. Second, Bill of Rights-based criminal procedure focuses the law’s attention on criminal trials, not on the discretionary processes that decide most cases.208 209Akhil Reed Amar, The Bill of Rights as a Constitution, 100 YALE L.J. 1131 (1991). 210That was certainly Ely’s view. See, e.g., ELY, supra note 13, at 97 (calling the Fourth Amendment a “harbinger of the Equal Protection Clause” and emphasizing the “tremendous potential for the arbitrary or invidious infliction of ‘unusually’ severe punishments on persons of various classes other than ‘our own’”). 211“The main thrust of the Bill was not to downplay organization structure, but to deploy it; not to impede popular majorities, but to empower them.” Amar, supra note 209, at 1132; AKHIL REED AMAR, THE BILL OF RIGHTS: CREATION AND RECONSTRUCTION xii-xiii (1998) (discussing “genius” of the Bill in similar terms). 212The canonical cite is Katz v. United States, 389 U.S. 347, 361 (1967) (Harlan, J., concurring) — though Justice Harlan does not use the phrase “reasonable expectation of privacy,” which has become the doctrinal test. 52 Prosecutors prone to abuse their authority usually do so through low-visibility plea bargains and even lower-visibility charging discretion, not through public trials. Yet charging and plea bargaining are, respectively, unregulated and nearly so, while the more visible trial process is confined at every turn by constitutional rules. The law drives regulated actors toward precisely the places where overreaching and abuse are most likely. Third, the Warren Court’s approach used constitutional law to protect values that politics and politicians can protect on their own. This point is not well understood, though it follows naturally from the argument in Akhil Amar’s classic article on the original understanding of the Bill of Rights, penned fifteen years ago.209 Before that article, the provisions of the Bill were widely seen as countermajoritarian tools for protecting individual rights and political minorities.210 Amar turned this conventional wisdom on its head. He showed that the Bill of Rights was originally designed to guard not individual rights but constitutional structure — not minorities but majorities — from a potentially oppressive central government.211 This majoritarian text protects majoritarian norms. The Fourth Amendment regulates policing by ensuring that “reasonable” — meaning ordinary, average — privacy expectations are not infringed without proper justification.212 The literature on the Fifth Amendment privilege against self-incrimination touts its countermajoritarian virtues, but for most of American history politicians did more 213See supra notes 116-19 and accompanying text. 214On the early history of the right to counsel, see Faretta v. California, 422 U.S. 806, 821-34 (1975); on more recent history, see, e.g., WILLIAM M. BEANEY, THE RIGHT TO COUNSEL IN AMERICAN COURTS (1955); LAWRENCE M. FRIEDMAN, CRIME AND PUNISHMENT IN AMERICAN HISTORY 235-58, 386-97 (1993); Lain, supra note 69, at 1389- 99. 215Cf. Amar, supra note 209, at 1182-91 (emphasizing juries’ majoritarianism more than their localism). 216See Barron v. Baltimore, 32 U.S. 243 (1833). 53 than judges to protect the privilege, especially in the police station.213 The Sixth Amendment right to counsel played a trivially small role in state court criminal justice before 1963, but politics ensured that counsel played a large role long before that date.214 Of course, the Sixth Amendment right to a jury trial embodies majoritarianism, by (apparently) guaranteeing local democratic control over the allocation of criminal punishment.215 The same is true of the jury right’s constitutional offshoots in double jeopardy law, which protects the finality of jury acquittals, and the law of confrontation, which protects jurors’ ability to see the government’s witnesses and judge their credibility. As those examples suggest, the chief among these rights — the sun around which the Bill’s procedural planets revolve — is the criminal jury. That is the one lesson of Amar’s analysis with which the current Supreme Court evidently agrees: witness the dramatic expansion of the jury trial right via Apprendi and its progeny. Today, the text that receives broadest enforcement is the one most thoroughly grounded in local democracy. That is a strange state of affairs, for local democracy is already well protected in American criminal justice. Prosecutors and most trial judges are elected locally. Sheriffs are elected by the counties they police; in cities, elected mayors and council members rule police departments. Mandating broad jury power — in federal cases; the Bill of Rights did not apply to the states216 — might have made good sense in 1791, when George Washington’s Justice Department seemed disconnected from local communities in Massachusetts and South Carolina. It makes little sense in our world, least of all in state cases, for juries do badly what the political system does fairly well: ensure that criminal justice reflects the will of local electorates. 217For a good, brief account of the politics of the Fourteenth Amendment, see ERIC FONER, RECONSTRUCTION 251-61 (1988). 218Hurtado v. California, 110 U.S. 516, 535-36 (1884). 54 The Sixth Amendent’s jury right is, in this respect, like the criminal procedure provisions of the Bill of Rights as a whole. Instead of democratically crafted privacy protections, the Fourth Amendment protects unelected judges’ views of what are or aren’t “reasonable” expectations of privacy. Instead of a trial process that legislators define and litigants actually use, the Bill of Rights mandates an idealized version of eighteenth-century English trials — or at least the aspects of those trials that most appealed to James Madison’s constituents — for the few, and plea bargains by democratically elected district attorneys for the many. There is no reason to think that the system is more democratic, and every reason to believe it is less just, than if legislators were allowed to do as they pleased. Democracy needs protecting; the Elyesque argument for constitutional regulation remains strong. But one must first understand what it needs protecting from. Even — especially? — the most democratic institutions are prone to ignore the interests of minorities and the poor. Majorities often prefer one set of rules for themselves, and a different and harsher set for those who live on the other side of the tracks. The proper role of constitutional law is to reject that majoritarian preference, to ensure that all parts of the citizenry live by the same rules and bear the same punishments when those rules are violated. That is not an enterprise to which the criminal procedure provisions of the Bill of Rights are well suited, since it is not the enterprise for which those provisions were designed. The Fourteenth Amendment’s Due Process and Equal Protection Clauses, on the other hand, were designed for precisely that purpose. The Amendment’s authors wrote those provisions to correct a democracy deficit: the tendency of Southern officials to deny newly freed slaves the “protection of the laws.”217 Unsurprisingly, those clauses offer a much better textual tool for addressing current democratic deficits. Two ideas are key. First is the link between due process and the rule of law. That is an old equation; the Supreme Court emphasized the connection between the two concepts in its first case applying Fourteenth Amendment due process to state court criminal litigation. As Justice Matthews put the point in 1884, “[l]aw is something more than mere will exerted as an act of power,” “not a special rule for a particular person or a particular case” and not “[a]rbitrary power.”218 That reading 219See Twining v. New Jersey, 211 U.S. 78, 107 (1908) (declining to apply the Fifth Amendment privilege against self-incrimination to the states because the right was insufficiently “fundamental in due process”); Palko v. Connecticut, 302 U.S. 319, 328 (1937) (reaching the same conclusion with respect to the Fifth Amendment’s Double Jeopardy Clause, on the ground that double jeopardy does not “subject[] the defendant [to] a hardship so acute and shocking that our polity will not endure it”); Rochin v. California, 342 U.S. 165, 172 (1952) (forcibly pumping the defendant’s stomach to obtain a small quantity of illegal narcotics violates due process, on the ground that “[t]his is conduct that shocks the conscience”); Darden v. Wainwright, 477 U.S. 168, 179-83 (1986) (prosecutor’s closing argument that referred to defendant as “an animal” who “shouldn’t be out of his cell unless he has a leash on him” did not require reversal of defendant’s conviction since it did not render the trial “fundamentally unfair”) (internal quotation marks omitted). 220See FONER, supra note 218, at 119-23, 425-44. 221See, e.g., Chicago v. Morales, 527 U.S. 41 (1999). For telling criticisms, see id. at 99-101 (Thomas, J., dissenting); Dan M. Kahan & Tracey L. Meares, The Wages of Antiquated Procedural Thinking: A Critique of Chicago v. Morales, 1998 U. CHI. LEGAL FORUM 197; Debra Livingston, Gang Loitering, the Court, and Some Realism About Police 55 of due process tends to focus constitutional attention on official discretion — “mere will” and “arbitrary power” — rather than on the details of trial procedure. Unfortunately, rule-of-law-based due process dropped out of criminal procedure doctrine early in the twentieth century, to be replaced by thoroughly unlawlike formulas such as “fundamental fairness” and “shocks the conscience.”219 The older tradition needs reviving. Criminal punishment will always be governed by a mix of law and discretion, but today, the mix is dangerously tilted toward discretion. It needs to be tilted back. Due process is the right constitutional hook. In criminal justice as elsewhere, discretion and discrimination travel together. Hence the need to ensure that the government offers the “protection of the laws” — a great constitutional phrase, often lost in the shadow cast by the modifier “equal” — to all its citizens. That need was especially salient to the men who wrote and ratified the Fourteenth Amendment. Discrimination and lawlessness dominated post-Civil War Southern justice, as private vigilantes enforced vicious racial codes and government officials refused to stop them.220 Offering the law’s protection to ex- slaves was one of the chief goals of the Amendment’s drafters. But like the rule of law in due process doctrine, that principle soon fell by the wayside. Current law often limits police protection, even in neighborhoods plagued by gang violence.221 Patrol, 1999 SUP. CT. REV. 141. 222Castle Rock v. Gonzales, 125 S.Ct. 2796 (2005). 223See RANDALL KENNEDY, RACE, CRIME, AND THE LAW 29-75 (1997). 224As Kelo v. New London, 105 S.Ct. 2655 (2005), shows, compensation is not a complete solution. Still, it reduces the size of the problem considerably. But cf. Daryl J. Levinson, Making Government Pay: Markets, Politics, and the Allocation of Constitutional Costs, 67 U. CHI. L. REV. 345, 364-67, 375-77 (2000) (arguing that compensation requirement may lead to inefficiently high or low level of takings). 225For an interesting contrary argument, see Jeffrey Manns, Liberty Takings: A Framework for Compensating Pretrial Detainees, 26 CARDOZO L. REV. 1947 (2005). 56 But the law never requires such protection, even when the government has notice that a violent crime may be in progress.222 Regulating police action but leaving inaction alone encourages more inaction. That message would be wrong in any context, but it is especially so given America’s history of discrimination against disfavored classes of crime victims.223 The poor disproportionately suffer the costs of both crime and police misconduct. Sometimes, the worst misconduct consists of the failure to stop the depredations of criminals. Here too, an older legal tradition needs reviving. Readers will likely agree that legality and equal protection are attractive norms for our, or any, criminal justice system. But in order to be attractive constitutional norms, there must be good reason to believe the political process will not protect them adequately. There is. Criminal punishment raises the same potential for strategic behavior by government officials as do takings. The monetary costs of punishment are of course spread across the citizenry, as are its benefits. But the nonmonetary costs are highly concentrated, borne by those who are punished and by their families. Just as the government may be tempted to take property from a few landowners and use it to benefit a large bloc of voters, it may also be tempted to punish too readily and too severely, given that the hardest-to-bear costs of punishment are so concentrated. Takings law solves this problem with a compensation requirement.224 For obvious reasons, that solution does not work with respect to criminal punishment.225 Another solution does: the rule of law, together with a strong equal 226For example, Brannon Denning and Glenn Reynolds studied lower courts’ response to United States v. Lopez, 514 U.S. 549 (1995), and found evidence of “willful judicial foot-dragging.” Brannon P. Denning & Glenn H. Reynolds, Rulings and Resistance: The New Commerce-Clause Jurisprudence Encounters the Lower Courts, 55 ARK. L. REV. 1253, 1256 (2003). 227It does. Both prongs of the current legal test for ineffective assistance — unreasonable attorney performance and a reasonable probability that attorney errors produced an adverse outcome, see Strickland v. Washington, 466 U.S. 668, 687-98 (1984) — require judges to determine causal connections between defense attorney tactics and 57 protection requirement. If the majority must live by the same rules and bear the same punishments as the minority, the majority’s rules are likely to be fairer and less severe than if one law applies in comfortable neighborhoods and another reigns in poorer places. Democracy will not reliably protect the ideal of one law for all, as the persistent racial gaps in drug enforcement and sentencing prove. Constitutional law should aim its guns at that problem, instead of guaranteeing protections that politicians would provide without the guarantees. B. The Method of Constitutional Regulation Not only does current constitutional law seek the wrong ends, it pursues those ends through means that guarantee regulatory failure. Lawmaking power is centralized, but ineffectively so. Because the Supreme Court reviews only a tiny fraction of criminal cases, lower courts are often free to pursue their own legal agendas without fear of adverse consequences.226 The Court does the bulk of its regulating via conduct rules, but the Justices lack the information needed to craft good conduct rules. To know which searches represent a reasonable balance of individual privacy and law enforcement need, judges must know the nature and distribution of local crime, how stretched police resources are, and how much search targets value different privacy interests. At best, trial judges have rough instincts about these things; appellate judges lack even that. No one has the information required to make such judgments for the nation as a whole. That is just one example; there are many others. The Justices do not know which defense attorney tactics are most likely to produce defense victories, and which ones are bound to lead to defeat. Probably no one knows, since the answers depend on context and (not incidentally) on the identity of defense counsel. To the extent that ineffective assistance doctrine requires such judgments,227 the Justices cannot do litigation outcomes. That is explicit in the second prong, which asks whether counsel’s mistakes prejudiced the defendant. But it is true of the performance prong as well. Defense litigation tactics are constitutionally unreasonable only if they tend to produce defeat — in other words, only if they are likely to be prejudicial. Strickland’s two prongs thus amount to prejudice in general and prejudice in the particular case at hand. Both kinds of prejudice are probably beyond judges’ capacity to determine accurately. The first is especially problematic in these terms, since good data on the effects of different defense tactics do not exist. 228On the consequences of asymmetric criminal appeals, see Kate Stith, The Risk of Legal Error in Criminal Cases: Some Consequences of the Asymmetry in the Right to Appeal, 57 U. CHI. L. REV. 1 (1990). 58 what their own doctrine requires. Meanwhile, the threat of reversal operates only when appeal is possible — which it isn’t in the many cases that end in guilty pleas. It follows that the largest effects of the Court’s rules are invisible to the Justices who define those rules. That is not a recipe for sensible regulation. The key ingredient of that recipe is information. In the existing system, the actors who need the most have the least. Remedying that mismatch requires changing the identity of the key decisionmakers, the nature of most liability rules, and the remedies used when those liability rules are violated. Constitutional regulation is a top-down enterprise; it needs to be more bottom-up. That requires taking power away from appellate judges (especially the nine who sit at the top of the judicial pyramid), and giving more of it to trial judges. Liability rules nearly always define prohibited conduct instead of banning impermissible outcomes. Where possible, that should change. Remedying constitutional violations usually means reversing criminal convictions. That too should change. Wherever feasible, civil injunctions should be the primary remedy for constitutional wrongs, in order to reward criminal justice institutions that perform well and punish those that do badly. Take these points in turn. Trial judges have several large informational advantages over appellate judges. Trial judges see not only tried cases but also guilty pleas, and among tried cases they see acquittals as well as convictions. Appellate judges see only convictions.228 Representatives of the same police force, district attorney’s office, and public defender’s office (often the same individual cops, prosecutors, and defense lawyers) appear repeatedly before any judge who tries criminal cases. Trial judges are therefore well positioned to see patterns of good and bad conduct by those institutions. Appellate judges aren’t. And a set of trial judges 229Or, much the same as criminal jury verdicts shape the incentives of law enforcement agencies. For the best discussion, see Daniel C. Richman, Old Chief v. United States: Stipulating Away Prosecutorial Accountability?, 83 VA. L. REV. 939, 956-69 (1997). 230Experience with the much-studied Federal Sentencing Guidelines confirms this point. The chief justification for the Guidelines was the elimination of unjustified sentencing disparity. But the Guidelines’ effect was not to eliminate disparity but to shift its source, from federal judges to federal prosecutors. For the classic discussion, see Ilene H. Nagel & Stephen J. Schulhofer, A Tale of Three Cities: An Empirical Study of Charging and Bargaining Practices Under the Federal Sentencing Guidelines, 66 S. CAL. L. REV. 501 (1992). For a recent study showing large disparities under state sentencing guidelines based on the choice of process — a matter that is often primarily under the control of prosecutors — see Nancy J. King, David A. Soule, Sara Steen, & Robert R. Weidner, When Process Affects Punishment: Differences in Sentences After Guilty Pleas, Bench Trial, and Jury Trial in Five Guidelines States, 105 COLUM. L. REV. 959 (2005). 59 making discretionary decisions in a given jurisdiction can create something like a market, to which government institutions can then respond — much as civil jury verdicts establish a market price of tortious behavior.229 Trial judges can then respond to the response; a satisfactory equilibrium can emerge fairly quickly. Appellate judges destroy this process by turning local discretionary judgments into jurisdiction-wide legal mandates. Change and adaptation take too long and, given the vagaries of stare decisis, may be impossible. Legal errors become uncorrectable. When trial judges err the errors are less long-lasting, hence less costly. An obvious disadvantage accompanies these advantages. More discretionary power for trial judges means more outcome disparity. Different trial judges exercise power differently, so similar cases in neighboring courtrooms may yield opposite results. That seems at odds with the rule of law. But such disparities exist already, through the much less transparent discretion of police officers and prosecutors.230 Giving trial judges more discretionary power would create competitive discretion, in place of the concentrated discretion that exists now. Given a choice between having a single prosecutor exercising discretionary authority over a criminal defendant’s charge and sentence and having both the prosecutor and a trial judge exercise such authority, the latter seems the better option. That is how checks and balances work: discretion checks discretion. Changed decisionmakers should go hand in hand with changed liability rules and remedies. Existing law defines prohibited conduct and reverses criminal 231See Miller-El v. Dretke, 125 S.Ct. 2317 (2005). Justice Souter’s majority opinion in Miller-El could have emphasized the number and distribution of strikes. Out of 20 black venire members, 9 were struck for cause and the prosecution used peremptory challenges to strike another 10. Only one black served on the jury. Id. at 2325. Instead, the bulk of the Court’s opinion is devoted to a detailed analysis of the voir dire of particular would-be jurors. Id. at 2325-32. 232See Fox Butterfield, When the Police Shoot, Who’s Counting?, N.Y. TIMES, April 29, 2001, Sec. 4, p. 5, col. 1. 60 convictions when the prohibited conduct happens, a process that focuses too much attention on particular transactions with individual regulated actors. Judges applying the Fourth Amendment examine how one or two police officers treated a single suspect. When deciding ineffective assistance claims, judges assess the conduct of individual defense lawyers in particular cases. As the Supreme Court has recently reminded us, the most minute details of individual voir dire examinations govern Batson claims.231 One is tempted to say: of course. What else could the law regulate besides conduct? Actually, a better option is readily available: regulating harm. Take police use of deadly force. Current Fourth Amendment law tries to draw a line between reasonable and unreasonable police shootings. As anyone who has read the case law knows, the line is badly drawn. That may be unavoidable, since no one knows how to specify in advance when police officers should try to kill suspects. But it would be fairly easy to identify trigger-happy police departments — all one need do is look at the data on rates of violent crime and violent policing, jurisdiction by jurisdiction. In many places, the data on police shootings do not exist — which is a scandal.232 Courts could require police departments to keep and publish such data. Outlier jurisdictions could be subject to injunctions requiring better training and more stringent regulation. A large fraction of the law of trial procedure could work the same way. When DNA testing gets cheap enough (maybe it already has), courts could require testing in all cases in which such evidence can be obtained. One could then measure the error rate, and impose more stringent procedures where that rate is high. The point does not depend on technological fixes. Instead of elaborate review of trial procedure and lax review of the evidence — the norm today — appellate courts could flip the coin: review the sufficiency of the evidence carefully but offer lax review of the trial 233See Ian Ayres & Robert Gertner, Strategic Contractual Inefficiency and the Optimal Choice of Legal Rules, 101 YALE L.J. 729 (1992); Robert E. Scott, A Relational Theory of Default Rules for Commercial Contracts, 19 J. LEGAL STUD. 597, 609 (1990). 234Chafee et al., supra note 116, at 213-24. In addition, state courts played a much larger role in regulating police interrogation than did federal courts. See id. at 52-83 (discussing reported cases). 61 process. After awhile, one could check to see which courthouses and which district attorneys’ offices err most often, and again, impose tougher procedural standards on the offending judges and prosecutors. That approach avoids the need to do something appellate courts routinely do now and always do poorly: decide what conduct tends to produce bad outcomes, with no information about how often and in what circumstances the outcomes are bad. Sometimes, conduct rules are unavoidable. Police interrogation is a good example, since we have no reliable measure of how coercive a given interrogation tactic is. But instead of picking a rule and imposing it nationwide, the Supreme Court could do either of two things: select a menu of permissible conduct rules, or (better still) create a penalty default — a default rule that is unattractive by design, so that politicians have an incentive to pick an appropriate conduct rule themselves. This concept is common in contract law and theory, where it is used to prompt parties to share valuable information,233 but its utility extends farther. Legislators know more about relevant policy alternatives than courts do. The goal of constitutional law should be to prompt elected officials to put that information to good use, not to have Supreme Court Justices choose among the policy alternatives themselves. What if legislators adopt inadequate conduct rules — legal fig leaves that pretend to regulate but actually leave police and prosecutors free to do as they please? At some point, any system of defaults must draw lines between permissible and impermissible regulatory approaches. But that truth does not negate the value of constitutional defaults, unless legislators always aim to regulate as little as possible. History proves that hypothesis false. Legislatures, not courts, initiated serious regulation of police interrogation.234 The same story holds throughout criminal procedure, unless one supposes that history began in 1961. Odds are, constitutional defaults would trigger a range of legislative responses. Brandeis’s laboratories of 235New State Ice Co. v. Liebmann, 285 U.S. 262, 311 (1932) (Brandeis, J., dissenting) (“It is one of the happy incidents of the federal system that a single courageous State may, if its citizens choose, serve as a laboratory; and try novel social and economic experiments without risk to the rest of the country.”). 236For the best discussion, see Debra Livingston, The Unfulfilled Promise of Citizen Review, 1 OHIO ST. J. CRIM. L. 653 (2004). See also Walker, supra note 101, at 27-28 (noting examples of citizen review boards publicizing complaint information in multiple languages in order to reach immigrant communities). 237See supra note 103 and accompanying text. 238For a good, brief history of the law and politics of lending discrimination, see Robert G. Schwemm, Introduction to Mortgage Lending Discrimination Law, 28 J. MARSHALL L. REV. 317 (1995). For a study showing a substantial increase in both loans and loan applications from minority neighborhoods in New York following legislation mandating detailed disclosure of lending patterns, see Richard D. Marsico, Shedding Some Light on Lending: The Effect of Expanded Disclosure Laws on Home Mortgage Marketing, Lending, and Discrimination in the New York Metropolitan Area, 27 FORDHAM URB. L.J. 481 (1999). For a broader explanation and defense of federal laws that combat lending discrimination, see Michael S. Barr, Credit Where it Counts: The Community Reinvestment Act and its Critics, 80 NYU L. REV. 513 (2005) 239Bradley Karkkainen touts mandatory disclosure as a “new paradigm” for environmental regulation. See Bradley C. Karkkainen, Information as Environmental Regulation: TRI and Performance Benchmarking, Precursor to a New Paradigm?, 89 GEO. L.J. 257 (2001). 62 democracy235 would work in criminal justice as they do elsewhere. That happy result will more likely materialize if claims of misconduct are more salient, so voters can see and respond to them. Citizen review boards that publicize allegations of police brutality are an obvious example of such visibility- promoting mechanisms.236 Contemporary anti-racial profiling statutes are another. Most such laws require nothing of the police save that they keep records of the details of traffic stops,237 which sounds like a minor piece of bureaucratic housekeeping. It isn’t. If and when the data expose discrimination, hearings will be held. At the least, stories will be written, and those stories will put pressure on politicians to respond. That may sound naive, but a similar approach has worked in other contexts. Banks forced to disclose lending demographics discriminate less.238 Polluters forced to disclose emissions pollute less.239 In these settings, disclosure requirements not only 240See NYPD STOPS AND FRISKS, supra note 78, at 88-135. 241See Walker, supra note 101, at 41-46 (discussing data collection on traffic stops and the use of court-appointed monitors to ensure that accurate data was collected). 242Governments can default on their debts, just as private corporations can. For a discussion of the most famous recent instance in the United States, see Frederick Tung, After Orange County: Reforming California Municipal Bankruptcy Law, 53 HASTINGS L.J. 885 (2002). But the greater risk of failure is of a different sort: that government institutions will do their jobs badly — take bribes, abuse those they serve, and the like. In the private sector, those kinds of failure risk institutional collapse. Not so in the public sector, unless institutional injunctions are on the table. 243See GERALD N. ROSENBERG, THE HOLLOW HOPE: CAN COURTS BRING ABOUT SOCIAL CHANGE? (1991). Rosenberg’s answer to that question was a resounding “no.” 63 act as a deterrent themselves, they also reinforce political regulation. Police departments and district attorneys’ offices might respond similarly. As they already have: When officers in the New York City Police Department were told to fill out forms describing the details of street stops, they did so — and the forms provided substantial evidence of police discrimination.240 When the Justice Department has obtained injunctions against local police departments under section 14141, it has routinely required the keeping of better records. The literature suggests the results have been positive.241 As section 14141 litigation suggests, recordkeeping and citizen review probably require another underused regulatory tool: institutional injunctions. Injunctions appear to be the ultimate in command-and-control regulation — hardly the way to make constitutional law more flexible and political market-friendly. Actually, injunctions are analogous to a kind of regulation that plays a large role in private markets: bankruptcy. Private firms that fail face reorganization; managers lose their jobs. No comparable market discipline applies to public institutions.242 Injunctions that put underperforming police departments or district attorneys’ offices into receivership are a substitute. Thanks in large part to Gerald Rosenberg,243 the idea that judicially crafted institutional injunctions have been tried and found wanting now approaches the status of conventional wisdom. School desegregation litigation produced decrees that were both intrusive and ineffective; the decrees in turn produced a political backlash that may have left black school children worse off than if courts had stayed out of the 244For the best version of this argument, see JOHN C. JEFFRIES, JR., JUSTICE LEWIS F. POWELL, JR. 282-331 (1994). For a spirited defense of the Supreme Court’s retreat from school desegregation decrees, see Bradley W. Joondeph, A Second Redemption?, 56 WASH. & LEE L. REV. 169 (1999). 245For good versions of the standard critique, see Neal Devins, I Love You, Big Brother, 87 CALIF. L. REV. 1283 (1999); Michael W. McConnell, Why Hold Elections? Using Consent Decrees to Insulate Policies From Political Change, 1987 U. CHI. LEGAL FORUM 295. 246See generally Charles Sabel & William H. Simon, Destabilization Rights: How Public Law Litigation Succeeds, 117 HARV. L. REV. 1015 (2004). For an argument that injunction litigation has changed less than Sabel and Simon imagine, see Margo Schlanger, Civil Rights Injunctions Over Time: A Case Study of Jail and Prison Court Orders, 81 NYU L. REV. ___ (forthcoming May 2006). 247For a particularly good telling of this sad story, see James E. Ryan, Schools, Race, and Money, 109 YALE L.J. 249, 275-84 (1999). 248For generally sympathetic accounts of this phenomenon, see FEELEY & RUBIN, supra note 172, at 59-62, 373-78; Margo Schlanger, Beyond the Hero Judge: Institutional 64 picture.244 Prison litigation led to complaints that overreaching judges were running America’s penal system.245 At first blush, these examples seem to suggest that institutional injunctions are at least dangerous, and at most perverse. There are three responses. First, the past is sometimes a poor predictor. Federal judges have the benefit of their predecessors’ experience; like all systems, judicial systems can learn from past failures. As Charles Sabel and Bill Simon demonstrate, substantial evidence of that learning process already exists: newer, more flexible and experimentalist styles of public-law injunction litigation have arisen to take the place of the older, command-and-control style that characterized the school and prison cases of the 1970s.246 Even if injunctions were applied to the same institutions as before, the results would be better today. Second, most criminal justice institutions differ in key ways from schools and prisons. White flight fatally undermined school desegregation decrees.247 Injunctions requiring less trigger-happy policing, more careful prosecutorial screening, or more accurate adjudication are hardly likely to produce mass exodus from the relevant jurisdictions. Prison cases were often collusive: prison officials cooperated with plaintiffs to try to get more resources from the state governments that funded them.248 Reform Litigation as Litigation, 97 MICH. L. REV. 1994, 2010-13 (1999). For more critical accounts, see Devins, supra note 246; McConnell, supra note 246. 249The Supreme Court of the 1970s and 1980s certainly thought so. See Los Angeles v. Lyons, 461 U.S. 95 (1983); O’Shea v. Littleton, 414 U.S. 488 (1974). 65 Save for suits seeking better funded public defenders, collusive criminal justice litigation seems hard to imagine. The third response has to do with the identity of the plaintiffs. Private parties — school children, prisoners, and their families — brought the suits that produced school desegregation and prison conditions decrees. Private litigation may be the right choice in the criminal justice context as well. But a more modest option is available: give state and federal governments the relevant cause of action; section 14141 is the obvious model. One might plausibly fear that governments would not enforce the law vigorously enough — and if experience shows that concern is justified, more conventional private-attorney-general injunction litigation could be allowed. But the fear may be misplaced. After all, Congress enacted section 14141 at a time when the conventional academic wisdom held that the politics of crime permitted no legislation favorable to defendants. The same political forces that made that legislation possible may produce aggressive enforcement by state and federal officials. Perhaps, as exclusionary rules and other reversal remedies recede, 14141- style litigation will expand to take their place. Even if governments were the sole plaintiffs, injunctions would be intrusive.249 But current law is worse. Conduct rules intrude on the liberty of everyone to whom they apply. Apprendi doctrine imposes a particular sentencing process on all jurisdictions that wish to use sentencing guidelines, just as the Fourth Amendment law of street stops requires that all such stops conform to the Court’s preferred template. Injunctions are selectively intrusive. They zero in on particular institutions that are producing bad outcomes and impose heavy legal regulation on them — and on them alone. To put the point another way, injunctions amount to defendant-specific lawmaking. That is a promising approach when identifying bad defendants is easier than defining bad behavior. Police departments that injure and kill too many of their citizens, prosecutors’ offices that prosecute many black drug defendants but few white ones, courts that regularly convict (or public defenders’ offices that regularly plead out) defendants later exonerated by DNA — all are examples of institutions that need an extra dose of regulation, and probably new 66 management. As to them, intrusiveness is a virtue, not a vice. And the threat of intrusion is a useful political incentive. Which is, after all, the point of the exercise. The goal of the constitutional rules that regulate police violence is to eliminate needless violence without needlessly risking police injury and death. The law of trial procedure aims to minimize punishment of innocents without making it too hard to punish the guilty. If we had a working body of constitutional law aimed at police and prosecutorial discrimination, its goal would be to ensure that both punishment and legal protection are allocated fairly across demographic divides. In all these areas, the current approach to constitutional regulation — appellate courts define forbidden conduct, victims sue or move to suppress evidence, and institutions respond to the signals that lawsuits and suppression hearings send — fails to get the right information into the right hands quickly and at reasonable cost. So the regulation accomplishes little. Greater use of outcome measures, default rules, publicity mechanisms, and institutional injunctions would help. But these regulatory tools can be properly used only if trial judges have the flexibility to use them. Conduct rules are top-down, a tool made for a centralized decisionmaker. The measures discussed above are more bottom-up. They will work only if the Supreme Court cedes some of its power to the courts that lie beneath it on the judicial pyramid. III. A REFORM PROGRAM The following sections sketch a program based on the principles laid out in Parts I and II. This is only a sketch, and a thin one at that. Books could be written, and have been, on each of the topics covered below: policing, adjudication and crime definition (usually treated separately), punishment, and federalism. One might plausibly conclude that little can be gained from a list of proposals unsupported by any developed argument or theory. That conclusion may be right, but it may be wrong. One unfortunate consequence of the sheer mass of current law and literature on these subjects is that lawyers, judges, and scholars rarely talk about any two of them — much less all four — in tandem. These are separate subjects, each with its own body of rules and theoretical underpinnings and policy debates. But the subjects interact. Changes in one area tend to produce changes in others. That is the basic lesson of the past four decades of constitutional criminal procedure. Constitutional change produced political change; the net effect does not appear to have been a good one. Seeing the 250See Kerr, Constitutional Myths, supra note 69; Orin S. Kerr, Digital Evidence and the New Criminal Procedure, 105 COLUM. L. REV. 279 (2005). 251See William J. Stuntz, The Virtues and Vices of the Exclusionary Rule, 20 HARV. J. L. & PUB. POL’Y 443, 451-55 (1997). 252Most of the innovation has centered around community policing. For much the best history of that phenomenon, in the context of broader trends in policing over the past half-century, see David A. Sklansky, Police and Democracy, 103 MICH. L. REV. 1699, 1778- 1814 (2005). Sklansky sees community policing as a mixed blessing. For a more positive account, with a focus on ties between police and churches in high-crime communities, see Tracey L. Meares, Praying for Community Policing, 90 CALIF. L. REV. 1593 (2002). 67 system whole, considering change in all its parts together, may be the key to wise reform. A. Policing the Police The chief focus of the constitutional law of policing is protecting the privacy of search targets; the chief remedy for violations is suppression of illegally obtained evidence. That system is perverse on several independent grounds. Privacy protection amounts to a legal Robin-Hood-in-reverse, since the rich get a good deal more protection than the poor. Perhaps because privacy tilts toward the middle and upper classes, legislatures will protect it adequately (and then some) without the need for elaborate constitutional rules, as Orin Kerr’s work on technology and privacy legislation demonstrates.250 And the exclusionary rule — a good tool for protecting privacy, but useless for any other purpose — distorts criminal litigation, drawing time and energy away from litigation’s central task: determining whether defendants committed the crimes charged.251 For these reasons, the best thing to do with the massive body of Fourth Amendment privacy regulation, together with the equally massive body of law on the scope and limits of the exclusionary rule, is to wipe it off the books. Let states experiment with different regulatory regimes. Some will protect privacy more than others, just as some offer more environmental protection or better public schools than others. Voters will decide what tradeoffs they prefer. One could stop there. The law of policing might work reasonably well — better than the current system — without any constitutional regulation. Policing has seen a great deal of productive innovation over the past generation;252 withdrawing 253For most of the twentieth century, New York City had a lower homicide rate than the nation as a whole. Beginning in the late 1950s and for four decades thereafter, New York’s rate was vastly higher. ERIC H. MONKKONEN, MURDER IN NEW YORK CITY 9 tbl. 1.1 (2001). And New York has lower crime rates than a number of other major cities. See, e.g., Levitt, Understanding Why Crime Fell, supra note 148, at 168 tbl. 4. 254Nationally, the ratio of reported index crimes to local police officers is roughly 17 to 1. FEDERAL BUREAU OF INVESTIGATION, UNIFORM CRIME REPORTS: CRIME IN THE UNITED STATES — 2003, at 70 tbl. 1 (showing 11,816,781 index crimes in 2003); 2003 SOURCEBOOK, supra note 5, at 42 tbl. 1.27 (showing 686,667 officers working for local police departments and sheriff’s offices). Most cities, particularly in the Sunbelt, have worse ratios, as the following figures show. Atlanta (30 to 1), Houston (27 to 1), Jackson, Mississippi (38 to 1), Memphis (35 to 1), Miami (33 to 1), Phoenix (39 to 1), San Antonio (45 to 1), San Diego (26 to 1). Northern cities do better but most still fall below the national average, as these figures illustrate: Baltimore (15 to 1), Detroit (22 to 1), Cleveland (18 to 1), San Francisco (20 to 1). New York City, which has a wealthier local tax base than other large cities, has a much more favorable crimes-to-cops ratio of 8 to 1. For the raw data, see CRIME IN THE UNITED STATES — 2003, supra, at 131-78 tbl. 8; id. at 375-447 tbl. 78. 68 judge-driven legal regulation might produce even more. It would almost certainly produce better funded local police forces. Policing is redistributive — given the demographics of crime, the poor receive the lion’s share of the benefits while wealthier taxpayers foot the bill. Since 1960, crime rates have been much higher in cities than elsewhere.253 City taxpayers can easily move to the suburbs and thereby lower their tax bills. So urban police forces are seriously, sometimes massively underfunded.254 That state of affairs will change only if state legislators and Congress spend more on local police, which will happen only if constitutional law regulates policing less. Constitutional retreat might produce regulatory advance. Nevertheless, constitutional law probably should regulate, modestly, three other kinds of police misconduct: violence, discrimination, and corruption. Unlike privacy-based regulation, all three regulatory goals are broadly consistent with a focus on guaranteeing the rule of law. Excessive force amounts to extralegal punishment. Discrimination applies one law to the majority and a different, harsher law to the minority. Corruption puts law itself up for sale. The manner of regulation should be flexible — a set of penalty defaults that apply only if no alternative regulation is in place. In most cases, citizen review boards and similar institutions 255Another word for “publicity,” at least in this context, is “transparency.” See Erik Luna, Transparent Policing, 85 IOWA L. REV. 1107 (2000). 256On the pros and cons of using money damages to deter police misconduct, see, for example, JEROME H. SKOLNICK & JAMES J. FYFE, ABOVE THE LAW: POLICE AND THE EXCESSIVE USE OF FORCE 198-205 (1993). As Daryl Levinson has explained, damages are inevitably a poor deterrent, because government agencies do not automatically maximize revenues. Levinson, supra note 225. Consequently, damages liability may actually lead to more police misconduct, by “buying off” opposition to the police practices that produce the liability. See id. at 367-73. 257For an extended argument, see William J. Stuntz, Miranda’s Mistake, 99 MICH. L. REV. 975, 992-96 (2001). 258See, e.g., Bureau of Justice Statistics, U.S. Dep’t of Justice, Homicide Trends in the U.S.: Trends by Race, available at http://www.ojp.usdoj.gov/bjs/homicide/race.htm (showing that 86% of white homicide victims were killed by whites, while 94% of black victims were killed by blacks). 69 should suffice. Publicity255 is a better regulatory tool than money damages256 or evidentiary suppression, because it triggers political accountability. Three problems deserve specific mention. The first involves a small but important subset of the regulation of police violence: interrogation of suspects. Here, conduct rules and evidentiary suppression are probably the best alternative: output measures won’t work (what would you measure?), which makes injunctions less useful in this context than in others. But Miranda is the wrong conduct rule. It protects sophisticated suspects — those savvy enough to invoke their rights as soon as the famous warnings are read — too much, and protects vulnerable suspects too little.257 That is a bad recipe in an age of sophisticated terrorist networks. On the other hand, Miranda might be a serviceable default rule, leaving states free to experiment with different means of preventing coerced confessions. The second problem concerns the relationship between two different forms of police discrimination. Local police can discriminate against either black defendants or black victims. Both are serious wrongs, worthy of serious constitutional regulation. Yet because most crime is intraracial,258 any effort to attack one of these wrongs risks aggravating the other. Protecting black victims seems to require prosecuting more black defendants, while reducing the number of black defendants apparently means less protection for black victims. But there is a way to 259See supra notes 186-87 and accompanying text. 260This proposition has long been the conventional wisdom in the literature on police violence. See, e.g., William A. Geller & Hans Toch, Understanding and Controlling Police Abuse of Force, in POLICE VIOLENCE 292, 297-303 (William A. Geller & Hans Toch eds., 1996). 261Cf. Dorf & Sabel, supra note 14, at 345-54 (advocating “rolling best-practices rules” by regulatory agencies). 262Cf. id. at 351 (suggesting environmental regulation in which “polluters are pushed from the bottom of the heap rather than pulled toward the top”). 70 walk this regulatory tightrope, because the two kinds of discrimination tend to arise in different classes of cases. When crime victims exist — think criminal violence and theft — the law should focus on discrimination against black victims. A mostly white political establishment will not immunize white defendants: white crime victims wouldn’t stand for it. Far more likely that white and black criminals who victimize blacks will get a free pass. When crimes are victimless — think drugs — the focus should be on discrimination against black suspects and defendants. The different racial compositions of the drug prisoner and drug user populations suggest that black suspects and defendants are punished more often and more severely than their white counterparts.259 Since the crimes that lead to victim-based discrimination differ from the crimes that produce defendant-based discrimination, the law can regulate both kinds of discrimination without working at cross-purposes. The proper regulatory tools are the same in both instances: require better data collection, and put the most discriminatory departments under new management. Data collection is the key.260 If adequate records of police stops and uses of force are kept, it should be possible to identify large-scale deviations from industry norms, and to target the offending police departments with injunctions like the ones the Justice Department has employed in section 14141 cases. This sounds like, but isn’t, a constitutional “best practices” requirement.261 Such a requirement would be mistaken, because it risks preempting more effective political regulation. The goal should be more modest: to identify and punish worst practices.262 Departments that show a consistent track record of brutality or discrimination, like Frank Rizzo’s Philadelphia police or Daryl Gates’s LAPD, should be put under new management. Politics should handle day-to-day regulation. Given the publicity that attends civilian 263Scholars disagree about the relative importance of politics and law in reducing police violence. Compare, e.g., PAUL CHEVIGNY, EDGE OF THE KNIFE: POLICE VIOLENCE IN THE AMERICAS 7 (1995) (“Since the sixties, . . . the use of deadly force by the police in the United States has declined precipitously; it is clear that the new management practices of the departments have brought about the decrease in shootings”) with, e.g., Abraham N. Tennenbaum, The Influence of the Garner Decision on Police Use of Deadly Force, 85 J. CRIM. L. & CRIMINOLOGY 241 (1994) (finding 16% decline in police shootings attributable to 1985 Supreme Court decision regulating police use of deadly force). Even if constitutional law has played a substantial role, its role may have been essentially political, with litigation serving as a means by which police violence could be better publicized. And damages litigation is, to say the least, an inefficient publicity mechanism. 264That was the central finding of the Mollen Commission: corruption in the NYPD was not an isolated phenomenon but rather a consequence of a defective organizational culture. See COMMISSION TO INVESTIGATE ALLEGATIONS OF POLICE CORRUPTION, ANATOMY OF FAILURE: A PATH FOR SUCCESS (1994). See also Armacost, supra note 101, at 501-02, 521-22 (noting links between corruption and police violence). 71 review boards and given the sunlight that better recordkeeping would cast on police violence and discrimination, politics should work reasonably well in all but the worst cases.263 The third problem is the constitutional status of police corruption — to date, not a subject of constitutional concern. That is a serious omission, made more so by the fact that political checks are hard to apply. Cops on the take do not announce themselves; the public hears about only a tiny fraction of cases. Local prosecutors depend on local police forces, which makes them poor candidates for the job of rooting out bad cops. Federal officials might do better on that score, if they had the right remedial tools. The need is for a remedy that is both civil and institutional, a remedy that can attack the organizational and managerial diseases that produce corrupt cops.264 Institutional injunctions are the obvious answer. In this instance, the remedy seems clearer than the right. As the law stands now, a police department that shows a pattern of corruption does not violate anyone’s constitutional rights. But that unfortunate bottom line is a function of a larger mistake in the law of constitutional criminal procedure: its exclusive focus on action rather than inaction. Discrimination against a class of crime victims — failing to prosecute their victimizers — robs a portion of the citizenry of the law’s protection. So does a corrupt police force that declines to enforce the law against criminals who pay bribes. Both are the antithesis of the rule of law. Both are proper objects of 72 constitutional scrutiny. B. Defining Crimes and Adjudicating Guilt The vast bodies of constitutional law that regulate criminal trials are self- defeating. Elaborate trial procedures have produced not better trials, but bigger criminal codes and more guilty pleas. Worse, the current law of trial procedure has stifled innovation. An age that has seen enormous experimentation in sentencing procedure still uses the trial process that James Madison decreed 215 years ago. The central problem with that process concerns the small quantity and low quality of the information it produces. The juries and judges who decide criminal cases do not know enough to decide those cases well. The trial judges and appellate courts that review criminal convictions do not know which institutions err frequently and which ones get it right most often. Prosecutors do not know when they have charged unfairly (in the view of anyone but themselves). Finally, state legislators do not know which crimes are charged in which cases, knowledge that is essential to wise substantive lawmaking. Each of these information deficits is easily redressed; four reforms would largely solve this interlocking set of problems. (1) Mandate adequate funding for indigent criminal defense. Raising funding for defense counsel would encourage states to increase prosecutors’ budgets as well, and better funded prosecutors and defense lawyers would give judges and juries better information on which to decide criminal cases. (2) In place of current procedure-heavy review, appellate courts should test outcomes for accuracy — using DNA or other reliable forensic evidence where possible — and keep good records of where and how mistakes happen. Institutions that regularly convict innocents should be enjoined to follow more stringent procedures. (3) Trial judges (perhaps appellate judges too) should be given the power to acquit for any reason, and encouraged to use that power aggressively. Over time, judicial acquittals would establish local market signals that rein in overzealous prosecution. (4) Last but not least, courts should require that charged crimes be systematically enforced — in every criminal case, the government should be forced to prove that in some number of factually similar cases other defendants have been convicted of the same crimes. In order to meet that requirement, states would be forced to keep records of prosecutions, fact patterns, and crimes charged — useful information for lawmakers designing criminal codes. Take these four proposals in turn. The attempt to regulate defense attorney conduct directly, through ineffective assistance doctrine, has failed. Ensuring a 265For a good argument for this surprisingly controversial proposition, see WILLIAM T. PIZZI, TRIALS WITHOUT TRUTH (1999). 266For a good discussion of the evolving science and its effects on litigation, see Julian Adams, Nuclear and Mitochondrial DNA in the Courtroom, 13 J.L. & POL’Y 59 (2005). 73 reasonable level of attorney quality is probably beyond the law’s grasp. Ensuring an adequate quantity of representation, though, is an achievable goal — and raising quantity tends to raise quality as well. The great drawback of judicially ordered funding requirements is that they are judicially ordered. But mandating adequate funding need not involve judges setting budget lines. Appellate courts could instead use a penalty default: Say, in all jurisdictions that set up expert commissions to recommend appropriate funding for indigent criminal defense and then follow those recommendations, ineffective assistance doctrine will not apply. Elsewhere, ineffective assistance standards will be ratcheted up sharply. If this default rule applied, state legislators would have an incentive to establish sensible processes for fixing defense budgets, and room to experiment with different funding patterns — more money for defense lawyers in some jurisdictions, more money for investigators or defense crime labs in others. Adequate funding for criminal defense would have large side benefits. Prosecutors’ offices are likewise underfunded. States would be compelled, as a practical matter, to increase their budgets in order not to lose too many criminal cases. More state money will tend to produce more state supervision and higher professional standards. Over time, higher budgets would bring down the guilty plea rate from current astronomically high levels. Guilty pleas would cease to be a budgetary necessity and would become a means of settling easy cases, which is their proper role. Similarly, the proper role for criminal trials is to produce accurate results.265 For all the effort that goes into regulating the trial process, the law does little to ensure that the goal is met. Instead of reviewing process, courts should test convictions for accuracy, using technology where possible266 and tougher evidentiary review elsewhere. That testing process should generate useful information: which district attorneys regularly prosecute innocents and which trial courts regularly convict them. Offending institutions should be required to improve performance. In time, as more jurisdictions experiment with different processes — different 267See Fong Foo v. United States, 369 U.S. 141 (1962). 74 combinations of jury and judge, greater use of non-adversarial experts to resolve some issues — we would learn which procedures produce the lowest error rates. Trial judges may already acquit for any reason in bench trials,267 but the law discourages exercise of that power. The discouragement should cease, and the power should apply to all trials. Broader judicial power over verdicts would give legislators an incentive to enact better procedural rules — procedures that judges think fair are more likely to produce the convictions that prosecutors and legislators want. The market mechanism would work, to some degree, even in a system configured as ours is now. But it would work better if another requirement were added: either the trial judge or the head of the prosecutor’s office must be appointed rather than elected. That way, in every case both an elected official who reflects majoritarian preferences and an appointed official who represents professional legal norms must find that criminal conviction is fair, or the defendant must be acquitted. The last item — requiring systematic enforcement — is the most important. For all crimes with a sentence of incarceration, prosecutors should be required to show that some number of other defendants in factually similar cases within the same state have been convicted of the same crime. (In federal cases, the relevant unit should be circuits rather than states.) The proper number is not clear. To begin with, it should be small: perhaps a dozen for most offenses. Over time, as databases grow, baselines could be adjusted upward as appropriate, with variation across crimes and places. Obviously, the requirement cannot apply to crimes or fact patterns that rarely occur. But most crimes and fact patterns occur frequently. Where that is so, it seems reasonable to require some evidence that the government treats similar cases similarly. This requirement would have two key side benefits. First, it would reduce prosecutors’ plea bargaining power. Prosecutorial threats are credible only if those threats can be carried out; this rule would limit prosecutors’ ability to threaten unusual criminal liability in order to extract guilty pleas. That should in turn make plea bargains more accurate. Second, the rule would force states to keep centralized records, a database with information on which defendants are prosecuted for which crimes, with what results. State recordkeeping might prompt more state supervision over local prosecutions — a good thing on several grounds. The massive run-up in state prison populations is, at least in part, due to interjurisdictional competition: a race to see 268See Robert L. Misner, Recasting Prosecutorial Discretion, 86 J. CRIM. L. & CRIMINOLOGY 717 (1996). 269See Doron Teichman, The Market for Criminal Justice: Federalism, Crime Control, and Jurisdictional Competition, 103 MICH. L. REV. 1831 (2005). 270See Barkow, Administering Crime, supra note 16; Barkow, Federalism and Sentencing, supra note 138. 75 who can be most punitive. Because state rather than local taxpayers pay for prison beds, local prosecutors tend to “spend” those beds more readily than they should.268 That feeds another tendency: to overpunish in order to scare criminals across county lines, or to fend off the neighboring district attorney trying to do the same.269 State supervision would curb these tendencies, as states’ experience with guidelines sentencing shows. As Rachel Barkow has explained, the most successful state sentencing commissions — the ones that have won legislative approval most consistently — have done so by gathering useful data about the budgetary consequences of different sentencing rules. When state legislators understand the prices of different rules, they are less quick to opt for whatever rule is most severe.270 Centralized data on prosecutions should shape substantive criminal law similarly. And state-level data would make it easier for state legislators to address enforcement disparities, particularly racial ones. The proposed rule would likewise reduce legislators’ incentive to enact crimes that will only rarely be enforced — at the least, it would avoid strategic use of those prohibitions in actual criminal prosecutions. If the crime were newly enacted or if prosecutors wished to enforce a previously unenforced crime, the defendant’s conviction could be stayed for a brief period, until prosecutors could win enough convictions in factually similar cases to satisfy the rule. It has long been clear that some means of addressing, and redressing, overcriminalization is essential if the system is to treat defendants fairly. The Supreme Court is loath to gin up a constitutionalized common law of crimes, and rightly so. Those two propositions have produced impasse. Protecting the rule of law by requiring a measure of systematic enforcement would break the impasse. C. Punishment Four major problems plague criminal sentencing today. The first is its severity. America’s inmate population tops two million, roughly seven times the 271See 2003 SOURCEBOOK, supra note 5, at 478 tbl. 6.1; CAHALAN, supra note 41, at 29 tbl.3-2. 272Whitman reports that average time served for French prisoners stood at 8 months in 1999; in 1975 the figure had been a little over 4 months. WHITMAN, supra note 1, at 70. Average time served in American state prisons was 34 months in 1999. 2002 SOURCEBOOK, supra note 109, at 506 tbl. 6.38. The analogous figure was 28 months in 1990, id., and also in 1960, CAHALAN, supra note 41, at 52 tbl. 3-23. 273On the racial breakdown of the inmate population, see 2003 SOURCEBOOK, supra note 5, at 505 tbl. 6.34. On the breakdown of the general population, see 2004-2005 STATISTICAL ABSTRACT, supra note 10, at 14 No. 13. 274See Steiker & Steiker, supra note 61. 275See generally id., passim (cataloguing the Supreme Court’s rejection of substantive limits on capital punishment). But cf. Roper v. Simmons, 125 S.Ct. 1183 (2005) (invalidating death penalty for offenders who were under 18 at the time of their crimes). 76 comparable figure in 1970, nearly five times the 1980 figure.271 Average sentences are vastly higher than in other Western countries, and significantly higher than in previous periods in this country.272 Second comes racial disparity: black inmates outnumber white ones by a large margin, though blacks comprise only 13% of the general population and non-Hispanic whites more than two-thirds.273 Third is disparity of other sorts, the different treatment similar cases receive and the similar treatment different ones receive, depending on which prosecutor’s desk houses the relevant file. Fourth is excessive prosecutorial power. Over the course of the past few decades, prosecutors have replaced judges as the system’s key sentencing decisionmakers, exercising their power chiefly through plea bargaining. That prosecutorial power is unchecked by law and, given its invisibility, barely checked by politics. So far, the Supreme Court has considered addressing these problems through conventional trial-type regulation of the sentencing process, or through open-ended review of substantive sentencing rules. The Eighth Amendment law of capital punishment follows the procedural model,274 as do the Sixth Amendment rules spawned by Apprendi. The substantive model has mostly been rejected, in capital275 276See Ewing v. California, 538 U.S. 11 (2003); Harmelin v. Michigan, 501 U.S. 957 (1991). 277And as Frank Bowman has explained, there is a strong connection between “more” and “harsher.” See supra note 139. 278434 U.S. 357, 358 (1978). 279Id. at 362-65. 77 and noncapital cases alike.276 Combining procedural regulation and substantive deference causes the same problems for sentencing as for criminal trials. Procedural regulation encourages more and harsher sentencing laws277 and more plea bargains, both of which reinforce prosecutors’ power instead of checking it. Some way to regulate substance must be found, but the Court is rightly unwilling to cast itself as the moral arbiter of criminal sentences. There is a way out of this box: deregulate process and apply the same rule-of- law limits to sentences as to criminal trials. As to process, neither Apprendi nor Eighth Amendment law is likely to produce better sentences than the processes that legislators define. More likely the reverse. Legislative innovation should be encouraged, not shut down. As to substance, the arguments are essentially the same as in the preceding section. For all sentences over some minimum level — say, three or six months — prosecutors should be required to show that sentences at least as severe have been imposed some minimum number of times for the same crime on similar facts. Those limits would make sentences less harsh and disparities smaller. Best of all, such limits would also reduce prosecutors’ power in plea negotiations. The defendant in Bordenkircher v. Hayes had two convictions for low- level felonies and faced a third small-time felony charge for “uttering a forged instrument” worth eighty-eight dollars.278 The prosecutor offered a deal: plead to the forged check and take a five year prison term. If the defendant refused, he would face a life sentence under Kentucky’s three-strikes law. The defendant did refuse, whereupon the prosecutor carried out his threat — with the Supreme Court’s approval.279 Bordenkircher encouraged the wave of three-strikes statutes that followed in the 1980s and 1990s, and it encouraged prosecutors to use those statutes and others like them as threats to extract pleas — not as punishments to be imposed on defendants who deserve them. It seems unlikely that Glen Bagby, the man who sent Paul Hayes to prison for the rest of his life, could have shown that other 280See Stephanos Bibas, Plea Bargaining Outside the Shadow of Trial, 117 HARV. L. REV. 2463, 2509-10 & nn.194-98 (2004). 281For an excellent discussion of both doctrinal and conceptual issues, see Carol S. Steiker, Punishment and Procedure: Punishment Theory and the Criminal-Civil Procedural Divide, 85 GEO. L.J. 775 (1997). 282See United States v. Salerno, 481 U.S. 739 (1987). 283See Kansas v. Hendricks, 521 U.S. 346 (1997). 78 Kentucky prosecutors had imposed life sentences on other felons with records like Hayes’s, since most prosecutors would think life an unjust sentence for someone like Hayes. Indeed, Bagby himself evidently thought so — hence his willingness to take a plea with a five-year sentence attached. Notice the incentives Bordenkircher creates. For prosecutors, the message is: threaten everything in your arsenal in order to get the plea bargain you want. For defendants, the message is simpler: take the deal, or else. (The incentive to plead applies to innocent and guilty defendants alike. Indeed, it may apply more strongly to innocents, who are more risk averse than their guilty counterparts.280) The most important incentive applies to legislators, who have every reason to vote for harsh penalties, secure in the knowledge that prosecutors need not and will not enforce them as written. Short of judges writing sentencing codes themselves, the only way to break this cycle is to require that the laws be enforced regularly or not at all. A similar requirement could and probably should apply to sanctions that, legally speaking, don’t count as “punishment” under current law.281 Pretrial detention is a classic example.282 Civil commitment for serial sex offenders is another.283 In both settings, the consequences of civil incarceration are indistinguishable from those of ordinary criminal punishment. That makes a strong case for abolishing constitutional distinctions between “punitive” incarceration and (allegedly) non- punitive kinds. If the same rules applied, judges could detain or civilly commit defendants only if the government showed that roughly similar facts and similar criminal histories led to similar results in other cases. One might go farther and offer judges the right to reduce sentences as they see fit, to hear arguments for lenity and respond to those arguments as the judges think 284I have argued for such a rule in the past. See Stuntz, Pathological Politics, supra note 82, at 594-96. 285See 125 S.Ct. 738, 756 (2005) (Breyer, J., opinion of the Court in part). 286Reitz, Sentencing Conundrum, supra note 58, at 1114-18. Interestingly, both mandatory rule-based sentencing and purely discretionary sentencing have produced harsh sentencing results, while a more mixed system with presumptive guidelines but not mandatory rules has produced more moderation. See id. at 1106 tbl. 1. 287One recent study examines three recent law enforcement initiatives and their effects on local homicide rates: Richmond’s Project Exile, Boston’s Project Ceasefire, and New York’s Compstat system of tracking local violent crimes. The study finds that Project Exile substantially outperformed the other two. Richard Rosenfeld, Robert Fornango, & Eric Baumer, Did Ceasefire, Compstat, and Exile Reduce Homicide?, 4 CRIMINOLOGY & PUB. POL’Y 419 (2005). The key feature of Project Exile was the sentences that were given to offenders convicted of federal gun crimes, which were both harsh and easily predictable. See Daniel C. Richman, “Project Exile” and the Allocation of Federal Law Enforcement Authority, 43 ARIZ. L. REV. 369, 379 (2001). 79 appropriate.284 After United States v. Booker, that appears to be roughly the rule in federal cases.285 Perhaps it should be the rule everywhere. But that proposition is not so obviously true as to justify a constitutional requirement. Kevin Reitz has noted that Pennsylvania — the state with sentencing laws that most resemble federal law after Booker — has seen its inmate population grow faster than the national average.286 Booker-style discretionary guidelines might produce not lenity but its opposite. And there is evidence that determinate sentencing produces larger deterrent gains than the indeterminate kind.287 Further study may show otherwise. But as long as the point is in doubt, legislatures should be free to choose. The massive racial disparity in America’s inmate population is a harder nut to crack. The changes in federal sentencing discussed below would help, as would budget increases for local police and local district attorneys. The African American portion of the prison population has grown as the proportion of indigent defendants has grown. Both tendencies are partly the consequence of docket pressure, which pushes police and prosecutors away from middle-class drug crimes and toward offenders in poor city neighborhoods. More generous budgets for law enforcement agencies should mean less docket pressure. Mandating adequate funding for public defenders’ offices, plus giving legislators more room to govern trial procedure, 288Sklansky, Cocaine and Race, supra note 139. 289481 U.S. 279 (1987). On McCleskey’s consequences for victim discrimination claims, see Kennedy, supra note 142. 290517 U.S. 456 (1996). For the best discussions of Armstrong’s meaning and impact, see Pamela S. Karlan, Race, Rights, and Remedies in Criminal Adjudication, 96 MICH. L. REV. 2001, 2023-29 (1998); Richard H. McAdams, Race and Selective Prosecution: Discovering the Pitfalls of Armstrong, 73 CHI.-KENT L. REV. 605 (1998) 80 should produce more generous budgets. Still, those things may not suffice. Less constrained budgets and less severe federal sentencing rules might reduce the system’s racial tilts, but only modestly. If that turns out to be the case, courts should undertake serious equal protection review of sentencing rules, testing the sentences white suburbanites and black city-dwellers get for similar crimes. Similar, but not identical — crack sentences should be brought into line with cocaine powder sentences, as David Sklansky argued a dozen years ago.288 Punishment for crimes that victimize blacks and crimes that victimize whites should likewise be aligned, once relevant differences in fact patterns are accounted for. Both McCleskey v. Kemp, which all but bars claims of victim discrimination,289 and United States v. Armstrong, which does the same for claims of defendant discrimination,290 should be overturned. Over time, strong antidiscrimination rules together with rule-of-law protection would moderate sentencing levels. Legislators could still craft symbolic sentencing rules — but those rules would be purely symbolic, mere expressions of outrage with no practical consequence. With respect to more consequential rules, this approach gives legislators information they too often lack about the price of substantive rules. Barkow’s work shows that legislators choose lenity when the price of severity is made clear. Banning discrimination and requiring systematic enforcement would make the price of severity a great deal clearer than it is now. D. Federalism America’s criminal justice system has a federalism problem, but the problem is not the one most readers will suspect. The usual concern is that an overlarge federal government will crowd out state and local agencies. Whatever the dangers 291Cf. Daryl J. Levinson, Empire-Building Government in Constitutional Law, 118 HARV. L. REV. 915 (2005) (arguing that Leviathan is not much of a danger generally). 292For the figures, see supra notes 18-22 and accompanying text. 81 of a federal Leviathan in other settings,291 no such danger exists here. Eleven thousand FBI agents will not soon displace 700,000 local police officers.292 The small size of the federal law enforcement bureaucracy creates a different problem. Because federal criminal law has so few consequences, Congress makes too much of it. Senators and Representatives use criminal law and sentencing doctrine to send messages, not to define prohibited conduct and its deserts. So Congress criminalizes too much and sentences too harshly. It also devotes far too much legislative energy to making law for the occasional federal prosecution, instead of regulating — and appropriating funds for — the local officials who do the real work of catching and punishing criminals. These tendencies are self-reinforcing. As Congress gives federal agents and prosecutors more room to pick and choose among potential targets, any one federal criminal prohibition or sentencing rule matters less. The less the rules matter, the more eager Congress is to add to them. So both the federal criminal code and the federal law of sentencing metastasize. That affects more than federal cases. Local district attorneys can threaten to send drug or gun crime defendants to the nearest United States Attorney’s office. If plea bargaining works like other markets, those threats affect the bargains local prosecutors and defense attorneys strike. Defendants agree to harsher sentences in state court for fear of what might happen to them in federal court. Federal law thus acts as an unfunded mandate for the states, raising state sentencing levels without paying for the increase. The doctrines that purport to protect federalism — chiefly, the law of federal criminal jurisdiction, enforced through jurisdictional elements that attach to individual crimes — make the situation worse. Federalism-based doctrines in criminal law cut across crimes, not between them: instead of assigning bribery to federal officials and arson to the locals, federal law covers some bribery and some arson, leaving local police and prosecutors the rest — with a fuzzy and constantly changing line between the two. A large fraction of federal criminal litigation is devoted to issues like whether robbery victims or torched buildings were sufficiently 293On the torched buildings, see Jones v. United States, 529 U.S. 848 (2000). On the robberies, see Craig R. Bradley, Federalism and the Federal Criminal Law, 55 HASTINGS L.J. 573, 592-98 (2004). 294As Dave Barry is wont to say, I am not making this up. See United States v. Terry, 257 F.3d 366, 369-70 (4th Cir. 2001) (finding federal jurisdiction because burned church ran a not-for-profit daycare center); id. at 373 (King, J., concurring in the judgment) (agreeing, in part because church ordered Sunday School materials from out of state); United States v. Rayborn, 312 F.3d 229, 234-35 (6th Cir. 2002) (finding federal jurisdiction because the burned church broadcast its services on the radio, because some churchgoers may have crossed state lines on Sunday mornings, and — best of all — because the church owned a recreational vehicle); cf. United States v. Lamont, 330 F.3d 1249, 1255 (9th Cir. 2003) (concluding that most church arsons would not give rise to federal jurisdictions but arsons of “mega-churches” might). 82 “commercial” to support federal charges.293 This generates lines of cases devoted to such questions as whether arsons of churches that order Sunday School materials from out of state are within the scope of federal authority.294 That kind of judicially mandated federalism obscures accountability and wastes the time of litigants and courts alike. The pattern recurs throughout the federal criminal code. Voters cannot know whom to credit when the system functions well and whom to blame when it doesn’t. That encourages irresponsible legislation. Better to draw some plausible lines between crimes that should be exclusively federal and crimes that should be exclusively enforced by state and local officials. Courts are poorly positioned to draw those lines, and Congress has no incentive to do so itself. Some mechanism is needed to encourage Congress to make law where law will count, and not elsewhere. There are a number of possibilities. The simplest and probably the best is a broad rule of sentencing preemption: Let federal sentences apply if, and only if, federal criminal law is exclusive. If a given federal crime is regularly enforced by local prosecutors, let federal sentences be fixed by state law. Harsh federal sentences for drug crimes would disappear unless Congress were willing to take over responsibility for enforcing drug laws nationwide, in which case federal sentences would grow less harsh. Giving federal law larger consequences would produce more moderate federal legislation. If a given crime is not regularly enforced by local prosecutors or if Congress decides to act preemptively, federal jurisdiction should be judged crime by crime, not case by case. Either fraud falls within the scope of federal power or not; either 295See Super, supra note 157. The major theme of Super’s article is state and local governments’ sensitivity to the business cycle. Federal spending is sensitive to economic peaks and troughs too, but in the opposite direction: when the economy tanks, federal spending rises. The consequence, as Super notes, is that other levels of government tend to undermine the federal fiscal stimulus. Id. at 2607-15. To put the point more positively, when state and local government spending falls, federal spending can cushion the blow. 296For a good discussion of the relationship between local and federal officials in the wake of both the Clinton legislation and September 11, see Daniel Richman, The Right 83 bribery gives rise to federal jurisdiction or it doesn’t. That across-the-board approach would lead to clearer lines of responsibility, and (one hopes) to a more politically accountable criminal justice system. More broadly, attaching a bigger price tag to federal criminal legislation might encourage different kinds of legislation — fewer criminal liability rules and sentencing regulations, and more funding and regulation of local agencies. A more fiscal federalism would better fit the federal government’s two great advantages in criminal justice policymaking: its ability to spend when state and local budgets are tight,295 and its concomitant ability to bribe local agencies to raise standards. E. Imagining Constitutional Reform Could it happen? Yes. Will it happen? Probably not. America’s criminal justice system is seriously diseased but not in crisis, and radical constitutional change generally follows a crisis — as the Civil War produced the Reconstruction Amendments and the Great Depression spawned the Roosevelt revolution of the late 1930s and after. Notice that in both of those instances, politics was the engine of transformative legal change. That fact should not surprise. Judging is an inherently conservative enterprise. Appellate courts must follow precedent or explain why they decline to do so. Politicians can more easily chart new paths. Which is why politicians, not judges, are likely to be the key players in any large-scale reform of the constitutional law of criminal justice. Bill Clinton’s “100,000 cops” initiative is a good example. The system Clinton inhabited made such initiatives politically difficult, which is why they have been so rare. It took a political entrepreneur like him to make that program happen. Still, it did happen. And because it did, local police departments can now imagine — and lobby for — a different, more productive relationship with the federal government than the one they had before.296 That could Fight, BOSTON REVIEW, Dec. 2004-Jan. 2005, at 6-11. 297Louis D. Bilionis, Conservative Reformation, Popularization, and the Lessons of Reading Criminal Justice as Constitutional Law, 52 UCLA L. REV. 979, 1027-47 (2005). 84 lead to better funded and better regulated local police. Victims of police misconduct can see a new regulatory tool (really an old tool — institutional injunctions — applied to a new context), one that holds some promise of tackling police violence and discrimination. Appellate judges can see as well. In time, they might be willing to play a smaller role in regulating local police so that laws like section 14141 can play a larger one. If the revolution comes, that will be its pattern. Politicians will innovate, courts will respond by loosening the constitutional reins, which will lead to still more innovation and yet looser reins, and so on in a virtuous circle. There are signs that politicians and voters are doing their part: the innocence projects that have swept the nation, the racial profiling debate of the last half-dozen years, the state-level sentencing commissions that are starting to reduce swollen prison populations. In decades to come, these examples may look like leading indicators of a healthier constitutional order. Maybe. But two large reasons for pessimism remain. As Louis Bilionis points out, the past few years have seen those constitutional reins grow tighter, not looser. The era of “conservative reformation” (his apt phrase) of the law of criminal procedure is over; the Supreme Court is once again extending its power.297 Clinton’s successors have less room to govern policing and procedure, not more. The second reason goes more to the manner of constitutional regulation than to its scope. Mandatory conduct rules for policing and trial procedure, the targeting of individual government officials rather than government institutions, enforcing constitutional rules through criminal litigation and (occasionally) damages actions — these basic features of the current system are all politically perverse. Conduct rules distract attention from unjust outcomes. Extensive regulation of policing prompts legislators to shift budget dollars from preventing crime to punishing it. Constitutionalized trial procedures encourage ever larger criminal codes and ever harsher sentences, both of which prosecutors use to evade the procedures. Targeting individual bad actors lets bad institutions off the hook and forces good ones to behave defensively. Whether the constitutional tide ebbs or flows, whether conservatives or liberals are ascendant, these basic dynamics remain. There is yet no 298Dean Bilionis draws a different parallel between criminal procedure and the rest of constitutional law: he argues that the former is a leading indicator of the latter. See id. at 990-1027. 299But cf. Robert C. Post, Fashioning the Legal Constitution: Culture, Courts, and Law, 117 HARV. L. REV. 4 (2003) (taking a more hopeful view of the law-politics relationship). Post’s article, while insightful (as is all his work), seems to me more optimistic than experience warrants. 85 sign that they are changing. Until they change, constitutional law will do little good in this critically important arena. That is a great shame, for constitutional regulation could play a far more productive role than simply getting out of politicians’ way. Legal minimalism is better than bad constitutional regulation, but good regulation is better than both. The key is understanding what separates good regulation from bad, productive constitutional law from the unproductive kind: attention to political incentives. Regulating politics is constitutional law’s raison d’être. But politicians cannot be commanded as the law commands criminals and tortfeasors. Effective regulation must be more cooperative than that. Generally speaking, constitutional law bears the same relation to domestic politics that international law bears to international politics. In both settings, politicians hold the high cards; direct conflict between law and politics rarely ends in law’s victory. So law that encourages political evasion and backlash inevitably does more harm than good. The goal should be to encourage something better — an achievable goal, if only the nine men and women who sit at the top of our judicial pyramid would work with politicians instead of against them. That hasn’t happened yet. The criminal justice system seems less a cooperative enterprise than a battleground — or a boxing ring, with judges in one corner and politicians in the other, each warily eyeing the other, looking for a chance to land a jab here or block a punch there. In that respect at least, criminal procedure is much like constitutional law generally. Politics sometimes works and so does constitutional law, but the two rarely work together. Perhaps that similarity follows from another: both criminal procedure and the rest of constitutional law use the same legal forms. Constitutional regulation usually employs mandatory conduct rules aimed at individual government officials and enforced through ordinary litigation, just like the law of criminal procedure.298 The political results seem broadly similar.299 American constitutional history has seen few examples of major 300The classic example is a very important one: Brown v. Board of Education, 347 U.S. 483 (1954), and the civil rights movement that followed it. But that is a more complicated example than first appears, as Michael Klarman explains well. See MICHAEL J. KLARMAN, FROM JIM CROW TO CIVIL RIGHTS: THE SUPREME COURT AND THE STRUGGLE FOR RACIAL EQUALITY 321-442 (2004). 301Dred Scott v. Sandford, 60 U.S. (19 How.) 393 (1857). On the decision’s political effects, see DON E. FEHRENBACHER, THE DRED SCOTT CASE: ITS SIGNIFICANCE IN AMERICAN LAW AND POLITICS (1978). 302FONER, supra note 218, at 523. 303Beginning with their sweeping victory in 1874, Democrats held the House majority for sixteen of the next twenty years, SEAN DENNIS CASHMAN, AMERICA IN THE GUILDED AGE: FROM THE DEATH OF LINCOLN TO THE RISE OF THEODORE ROOSEVELT 247 (3d ed. 1993) — often by large margins, as Leonard White’s oddly titled book on the period notes. See LEONARD D. WHITE, THE REPUBLICAN ERA: 1869-1901, at 49-51 (1958). 304Democrats Samuel Tilden and Grover Cleveland won the popular vote in 1876, 1884, 1888, and 1892 — the first and last of those years by sizeable margins. FONER, supra note 218, at 576; H. WAYNE MORGAN, FROM HAYES TO MCKINLEY: NATIONAL PARTY POLITICS 1877-1896, at 232, 317, 436 (1969). Republican James A. Garfield won the popular vote in 1880 by the smallest margin in American history. KENNETH D. ACKERMAN, DARK HORSE: THE SURPRISE ELECTION AND POLITICAL MURDER OF PRESIDENT JAMES A. GARFIELD 220 & n.13 (2003). 305The Democrats’ 1874 landslide led to Tilden’s near-victory in 1876, when the Republicans won only by compromising on civil rights. For the classic version of the story, see C. VANN WOODWARD, REUNION AND REACTION: THE COMPROMISE OF 1877 AND THE END OF RECONSTRUCTION (2d ed. 1966). The Republican retreat on civil rights in turn led, 86 constitutional change prompting political change in the same direction.300 But history is littered with examples of constitutional change prompting political backlash. Consider the following short list. Dred Scott led to the first Republican Congress, and later to the presidency that emancipated the slaves.301 The Reconstruction Amendments and the civil rights legislation that accompanied them led to “a new era in national politics”:302 two decades of Democratic dominance in the House of Representatives,303 a Democratic plurality in four of the next five presidential elections,304 the abandonment of federal civil rights protection, and the rise of Jim Crow.305 Lochner-era cases invalidating economic regulation led to the with some curious twists, to the rise of Jim Crow. Again, Woodward tells this sad tale better than anyone else. C. VANN WOODWARD, THE STRANGE CAREER OF JIM CROW (1974). 306The reference, of course, is to Lochner v. New York, 198 U.S. 45 (1905). The year after Lochner was decided, Congress passed and Theodore Roosevelt signed the Pure Food and Drug Act of 1906, Pub. L. No. 59-384, 34 Stat. 768. Judicial conservatism seems to have radicalized Roosevelt, who later advocated judicial recall in order to prevent judges from thwarting progressive legislation. See PATRICIA O’TOOLE, WHEN TRUMPETS CALL: THEODORE ROOSEVELT AFTER THE WHITE HOUSE 147-49 (2005). 307See Schechter Poultry Corp. v. United States, 295 U.S. 495 (1935). Schechter Poultry was decided on May 27, 1935. Eight days later, Franklin Roosevelt announced his support of the Wagner Act, which he had previously opposed and which Congress promptly passed by a voice vote. Two weeks after that, Roosevelt proposed a sharply progressive tax law which Congress likewise quickly passed. See MICHAEL BARONE, OUR COUNTRY: THE SHAPING OF AMERICA FROM ROOSEVELT TO REAGAN 85-92 (1990). 308347 U.S. 483 (1954). On the link between Brown and the rise of massive resistance, see KLARMAN, supra note 301, at 326-34, 389-421. 309See generally TAYLOR BRANCH, PARTING THE WATERS: AMERICA IN THE KING YEARS, 1954-1963 (1988); DAVID J. GARROW, BEARING THE CROSS: MARTIN LUTHER KING AND THE SOUTHERN CHRISTIAN LEADERSHIP CONFERENCE (1986); DAVID J. GARROW, PROTEST AT SELMA: MARTIN LUTHER KING, JR. AND THE VOTING RIGHTS ACT OF 1965 (1978). Michael Klarman argues persuasively that the civil rights movement owed its success in part to the violence with which white Southerners responded to Brown — that the civil rights legislation of the 1960s was, in large part, the consequence of a backlash to a backlash. KLARMAN, supra note 301, at 421-42. 310410 U.S. 113 (1973). For two different slants on the history of the movement, see DALLAS A. BLANCHARD, THE ANTI-ABORTION MOVEMENT AND THE RISE OF THE RELIGIOUS RIGHT (1994); JAMES RISEN & JUDY L. THOMAS, WRATH OF ANGELS: THE AMERICAN ABORTION WARS (1998). The presidents elected with strong pro-life support are Ronald Reagan and the two George Bushes. The fourth president elected during this period, Bill 87 rise of progressivism, which produced a raft of regulatory legislation.306 Schechter Poultry led to the Second New Deal, more radical than the first.307 Brown v. Board of Education gave birth to massive resistance308 — it was Martin Luther King’s civil rights movement and not the Supreme Court’s civil rights decisions that produced the landmark federal legislation of the mid-1960s.309 Roe v. Wade spawned the pro-life movement, which has elected three of the last four presidents.310 The Massachusetts Clinton, won a large measure of pro-life support by promising to make abortion “safe, legal, and rare” — a mantra the leading Democratic presidential hopeful for 2008 has adopted for her own White House bid. See Raymond Hernandez & Patrick D. Healy, The Evolution of Hillary Clinton, N.Y. TIMES, July 13, 2005, at B1. 311798 N.E.2d 941 (2003). On Goodridge’s political effects, see Michael J. Klarman, Brown and Lawrence (and Goodridge), 104 MICH. L. REV. ___, __-__ (forthcoming 2006) (manuscript at 46-87). 88 Supreme Judicial Court’s decision in Goodridge v. Department of Public Health prompted a wave of state referenda condemning gay marriage.311 The past half-century of American criminal justice seems very much of a piece with that troubled history. The Supreme Court decided to regulate policing and procedure, and the politicians responded with a forty-year backlash of overcriminalization and overpunishment. Just when the backlash seemed a spent force, the Justices decided to begin a new round of regulation. Expect more of the same. IV. CONCLUSION It is tempting to judge constitutional law by its aspirations. The bodies of Fourth, Fifth, Sixth, Eighth, and Fourteenth Amendment law that regulate American criminal justice seek to prevent oppression and to promote democracy. Few would object to those goals. But however virtuous their ends, those bodies of law have produced bad, even disastrous consequences. It is long past time to put the good intentions aside, and focus on the consequences. The most important consequences are political. Constitutional law does not dominate the politics of crime, but it does shape political rhetoric and define the turf on which political opponents do battle. Right now, the shaping is poorly done. Political discourse in this area is too punitive, too racially divisive, and insufficiently attentive to the liberty and autonomy interests that constitutional law allegedly protects. “Allegedly” is the right word. Far from protecting those interests, current law encourages politicians to undermine them. The upshot is that criminal justice in America is, all too often, more criminal than just. We can and should change that. The key lies in using constitutional law to reinforce healthy representative politics: the wise dream John Hart Ely advanced a quarter century ago. The title of Ely’s book is especially apt in this context. Democracy rules our 312Martin Luther King, Jr., I Have a Dream, (Aug. 28, 1963) (transcript available at http://www.americanrhetoric.com/speeches/Ihaveadream.htm). 313Not surprisingly, since — as David Strauss has explained — we are governed by a common-law Constitution. See David A. Strauss, Common Law Constitutional Interpretation, 63 U. CHI. L. REV. 877 (1996); David A. Strauss, The Irrelevance of Constitutional Amendments, 114 HARV. L. REV. 1457 (2001). 314See Levinson, supra note 225; Daryl J. Levinson, Framing Transactions in Constitutional Law, 111 YALE L.J. 1311 (2002). 89 criminal justice system: with its many elected sheriffs, district attorneys, and judges, America probably has the most democratic criminal justice apparatus in the world. But distrust — especially, racial distrust — likewise rules that system. Current constitutional law reinforces that distrust, by steering politics onto punitive and racially destructive paths. If the steering were done better, we could have a politics of criminal justice that is about fighting crime while doing justice, not about maximizing law enforcers’ discretion while exploiting racial division. The criminal justice system might then realize another wise man’s wise dream: a world where criminal suspects and defendants are “not . . . judged by the color of their skin, but by the content of their character.”312 It bears emphasizing that when Dr. King spoke those words, he was talking about — and he was doing — politics, not law. Constitutional law can create the conditions for a racially just criminal justice system, but it cannot mandate that system. For that job, we need politicians. They need law that helps them do the jobs our oddly designed system gives them. To date, constitutional law has been more hindrance than help, mostly because of the mismatch between the job it has — regulating politics — and the tools it uses. That mismatch plagues more than criminal justice (though it is especially acute there); the problem extends to other constitutionally regulated fields as well. The tools come from the common law’s toolbox; they are the same ones generations of judges used to regulate the private-law obligations of individuals and corporations.313 As my colleague Daryl Levinson has explained, government officials have different reward structures than private litigants and different relationships with injured claimants.314 Regulatory approaches that work for the one group are unlikely to have the desired effect on the other. The need is for different regulatory tools — again, in criminal procedure and probably elsewhere. 90 It seems we have been looking through the wrong end of the telescope. Judges, lawyers, and scholars alike find it easy to spot the many ills that afflict American politics. Hence the temptation to see politics as a disease and constitutional law as its cure. But the metaphor is backward. It is law, not politics, that risks stifling experiment and innovation in America’s criminal justice system, and probably in other areas as well. As between more constitutionally sensitive politics and a more politically sensitive Constitution, the latter is much the more urgent need. As they administer their legal medicines, Supreme Court Justices would do well to remember that, and to heed a piece of ancient medical wisdom: Physician, heal thyself.