Criminal Justice Assignment

University of Pennsylvania Law Review FOUNDED 1852 Formerly American Law Register VOL. 113 NOVEMBER 1964 No. 1 TWO MODELS OF THE CRIMINAL PROCESS * HEBERT L. PAcKE t There are two more or less separable complexes of issues which need to be investigated as one approaches the central question of the limits of criminal law. One complex of issues concerns what may be called the ideology of the criminal law, such as questions about the nature and purposes of criminal punishment. This is generally recog- nized as relevant to what I have termed the central question.' There does not seem to be an equivalent recognition of the relevance of the other complex of issues, which concerns what may be called the processes of the criminal law. 2 The major premise of this Article is that the shape of the criminal process has an important bearing on * This Article is a sketch for a portion of a work in progress concerning the criteria that a rational lawmaker should consider in determining what kinds of conduct to treat as criminal. Legal thought has not had much to say on this question: little enough if by "legal thought!' we mean thought about law; less still if by it we mean thought by lawyers. The present Article is intended as a prolegomenon directing attention to a group of problems necessarily affecting the behavior content of the criminal law. Its appear- ance in this forum, given its nontechnical nature, can be explained only by giving a broad construction to the "we" in Holmes's aphorism that "what we need at this time is education in the obvious more than investigation of the obscure." It would be both premature and presumptuous to identify all those who have aided in the enterprise whose first fruits are presented here. The burden of my gratitude cannot, however, be evaded by the Dean and Faculty of the University of Pennsylvania Law School, whose generous hospitality during a sabbatical year pro- vided the ideal environment for pursuing the subject of these reflections. t Professor of Law, Stanford University. B.A. 1944, LL.B. 1949, Yale University. Member, New York Bar. I See, e.g., DavmiN, THE ENFORCEMENT OF MORALS (1959) ; HART, LAWv, LIBERTY, AND MORAIrrY (1963). 2 There has been a tendency among students of the criminal process to treat procedural issues as if their resolution had nothing to do with judgments about the substantive uses of the criminal law, which apparently are thought to be immutable. See, e.g., Barrett, Police Practices and the Law-From Arrest to Release or Charge, 50 CALIF. L. REv. 11, 20 (1960). 2 UNIVERSITY OF PENNSYLVANIA LAW REVIEW questions about the wise substantive use of the criminal sanction. Its minor premise is that important trends in the development of the criminal process that are now underway make the task of appraising the uses of the criminal sanction an especially timely one. We will start by considering the spectrum of choices that is at least in theory open in fixing the shape of the criminal process and by proposing a device for identifying and appraising the poles and the distance between them. The device, a pair of models, will then serve as a framework for considering the dynamism that appears to char- acterize present-day trends in the evolution of the criminal process. Finally, after a summation of the trends and an attempt to evaluate their continued potency, some tentative suggestions will be advanced about the relevance of the criminal process to the elaboration of criteria for the substantive invocation of the criminal sanction. I. VALUES AND THE CRIMINAL PROCESS A. Why Build Models? People who commit crimes appear to share the prevalent im- pression that punishment is an unpleasantness that is best avoided. They ordinarily take care to avoid being caught. If arrested, they ordinarily deny their guilt and otherwise try not to cooperate with the police. If brought to trial, they do whatever their resources permit to resist being convicted. And, even after they have been convicted and sent to prison, their efforts to secure their freedom do not cease. It is a struggle from start to finish. This struggle is often referred to as the criminal process, a compendious term that stands for all the complexes of activity that operate to bring the substantive law of crime to bear (or to avoid bringing it to bear) on persons who are sus- pected of having committed crimes. It can be described, but only partially and inadequately, by referring to the rules of law that govern the apprehension, screening, and trial of persons suspected of crime. It consists at least as importantly of patterns of official activity that correspond only in the roughest kind of way to the prescriptions of procedural rules. As a result of recent emphasis on empirical research into the administration of criminal justice, we are just beginning to be aware how very rough the correspondence is.3 3 See, e.g., Goldstein, Police Discretion Not To Invoke the Criminal Process: Low-Visibility Decisions in the Administration of Criminal Justice, 69 YALE L.J. 543 (1960) ; LaFave, Detention for Investigation by the Police: An Analysis of Current Practices, 1962 WASH. U.L.Q. 331. Both articles are based to some extent on ma- terial gathered in the American Bar Foundation Survey of the Administration of Criminal Justice in the United States. [Vo1.113:1 TWO MODELS OF THE CRIMINAL PROCESS At the same time, and perhaps in part as a result of this new accretion of knowledge, some of our lawmaking institutions-and particularly the Supreme Court of the United States-have begun to add measurably to the prescriptions of law that are meant to govern the operation of the criminal process. This accretion has become, in the last few years, exponential in extent and velocity. We are faced with an interesting paradox: the more we learn about the Is of the criminal process, the more we are instructed about its Ought and the greater the gulf between Is and Ought appears to become. We learn that very few people get adequate legal representation in the criminal process; we are simultaneously told that the Constitution requires people to be afforded adequate legal representation in the criminal process. 4 We learn that coercion is often used to extract confessions from suspected criminals; we are then told that convictions based on coerced confessions may not be permitted to stand. 5 We discover that the police in gathering evidence often use methods that violate the norms of privacy protected by the fourth amendment; we are told that evidence so obtained must be excluded from the criminal trial.' But these prescriptions about how the process ought to operate do not automatically become part of the patterns of official behavior in the criminal process. Is and Ought share an increasingly uneasy co- existence. Doubts are stirred about the kind of criminal process we want to have. The kind of criminal process we have is an important determinant of the kind of behavior content that the criminal law ought rationally to comprise. Logically, the substantive question may appear to be anterior: decide what kinds of conduct one wants to reach through the criminal process, and then decide what kind of process is best calculated to deal with those kinds of conduct. It has not worked that way. On the whole, the process has been at least as much a Given as the content of the criminal law. But it is far from being a Given in any rigid sense.

The shape of the criminal process affects the substance of the criminal law in two general ways. First, one would want to know, before adding a new category of behavior to the list of crimes and therefore placing an additional burden on the process, whether it is easy or hard to employ the criminal process. The more expeditious the process, the greater the number of people with whom it can deal and, therefore, the greater the variety and, hence, the amount of anti- 4Gideon v. Wainvright, 372 U.S. 335 (1963). 5E.g., Haynes v. Washington, 373 U.S. 503 (1963); Escobedo v. Illinois, 378 U.S. 478 (1964). 6 Mapp v. Ohio, 367 U.S. 643 (1961). 1964] 4 UNIVERSITY OF PENNSYLVANIA LAW REVIEW social conduct that can be confided in whole or in part to the criminal law for inhibition. On the other hand, the harder the process is to use, the smaller the number of people who can be handled by it at any given level of resources devoted to staffing and operating it. The harder it is to put a suspected criminal in jail, the fewer the number of cases that can be handled in a year by a given number of policemen, prosecutors, defense lawyers, judges and jurymen, probation officers, etc. A second and subtler relationship exists between the char- acteristic functioning of the process and the kinds of conduct with which it can efficiently deal. Perhaps the clearest example, but by no means the only one, is in the area of what have been referred to as victimless crimes, i.e., offenses that do not result in anyone's feeling that he has been injured so as to impel him to bring the offense to the attention of the authorities. The offense of fornication is an example. In a jurisdiction where it is illegal for two persons not married to each other to have sexual intercourse, there is a substantial enforcement problem (or would be, if the law were taken seriously) because people who voluntarily have sexual intercourse do not often feel that they have been victimized and therefore do not often complain to the police. Consensual transactions in gambling and narcotics present the same problem, somewhat exacerbated by the fact that we take these forms of conduct rather more seriously than fornication from the standpoint of the criminal law. To the difficulties of ap- prehending a criminal when it is known that he has committed a crime are added the difficulties of knowing that a crime has been committed. In this sense the victimless crime always presents a greater problem to the criminal process than does the crime with an ascertainable victim. But this problem may be minimized if the criminal process has at its disposal measures that are designed to enhance the probability that the commission of such offenses will become known. If suspects may be entrapped into committing offenses, if the police may arrest and search a suspect without evidence that he has committed an offense, if wiretaps and other forms of electronic surveillance are permitted, it becomes easier to detect the commission of offenses of this sort. But if these measures are prohibited and if the prohibitions are observed in practice, it becomes more difficult, and eventually there may come a point at which the capacity of the criminal process to deal with victim- less offenses becomes so attentuated that a failure of enforcement occurs. In both of these ways, the characteristics of the criminal process bear a relationship to the central question of what the criminal law is good for. Both a general assessment of whether that process is a high-speed or a low-speed instrument of social control and a series of [Vol.l13:1 TWO MODELS OF THE CRIMINAL PROCESS specific assessments of its fitness for the handling of particular kinds of antisocial behavior are called for if we are to have a basis for elab- orating the criteria that ought to affect the invocation of the criminal sanction. How can we provide ourselves with an estimate of the criminal process that pays due regard to its static and dynamic ele- ments? There are, to be sure, aspects of the criminal process that vary only inconsequentially from place to place and from time to time. But its dynamism is clear-clearer today, perhaps, than ever before. We need to have an idea of the potentialities for change in the system and the probable direction that change is taking and may be expected to take in the future. We need to detach ourselves from the welter of more or less connected details that make up an accurate description of the myriad ways in which the criminal process does operate or may be likely to operate in midtwentieth-century America so that we can begin to appraise the system as a whole in terms of its capacity to deal with the variety of substantive missions we confide to it. One way to do this kind of job is to abstract from reality, to build a model. In a sense that is what an examination of the constitu- tional and statutory provisions that govern the operation of the criminal process would produce. This, in effect, is the way analysis of the legal system has traditionally proceeded. The method has considerable utility as an index of current value choices; but it produces a model that will not tell us very much about some important problems that the system encounters and that will only fortuitously tell us anything useful about how the system actually operates. On the other hand, the kind of model that might emerge from an attempt to cut loose from the law on the books and to describe, as accurately as possible, what actually goes on in the real-life world of the criminal process would so sub- ordinate the inquiry to the tyranny of the actual that the existence of competing value choices would be obscured. The kind of criminal process we have depends importantly on certain value choices that are reflected, explicitly or implicitly, in its habitual functioning. The kind of model we need is one that permits us to recognize explicitly the value choices that underlie the details of the criminal process. In a word, what we need is a normative model, or rather two models, to let us perceive the normative antinomy that runs deep in the life of the criminal law. These models may not be labelled Good and Bad, and I hope they will not be taken in that sense. Rather, they represent an attempt to abstract two separate value systems that compete for attention in the operation of the criminal process. Neither is pre- sented as either corresponding to reality or as representing what the criminal process ought to be. The two models merely afford a con- 19641 6 UNIVERSITY OF PENNSYLVANIA LAW REVIEW venient way to talk about the operation of a process whose day-to-day functioning involves a constant series of minute adjustments between the competing demands of two value systems and whose normative future likewise involves a series of resolutions, of greater or lesser magnitude, of the tensions between mutually exclusive claims. I call these two models the Due Process Model and the Crime Control Model. In the next section I shall sketch their animating presuppositions, and in succeeding sections I shall present the two models as they apply to a selection of representative problems that arise at successive stages of the criminal process. As we examine in succession this sampling of stage and substage of the criminal process on which the models operate, we will move from the description of the model stages to two further inquiries: first, where on a spectrum be- tween the extremes represented by the two models do our present practices seem approximately to fall; second, what appears to be the direction and thrust of current and foreseeable trends along each such spectrum? There is a risk in an enterprise of this sort that is latent in any attempt to polarize. It is, simply, that values are too various to be pinned down to yes or no answers. When we polarize, we distort. The models are, in a sense, distortions. The attempt here is only to clarify the terms of discussion by isolating the assumptions that under- lie competing policy claims and examining the conclusions to which those claims, if fully accepted, would lead. This Article does not make value choices, but only describes what are thought to be their consequences. B. Values Underlying the Models In this section we shall develop two competing systems of values, the tension between which accounts for the intense activity now observable in the development of the criminal process. The models we are about to examine attempt to give operational content to these conflicting schemes of values. Like the values underlying them, the models are polarities. Just as the models are not to be taken as describing real-world situations, so the values that underlie them are not to be regarded as expressing the values held by any one person. The values are presented here as an aid to analysis, not as a program for action. 1. Some Common Ground One qualification needs to be made to the assertion of polarity in the two models. While it would be possible to construct models that exist in an institutional vacuum, it would not serve our purposes to [Vo1.113:1 TWO MODELS OF THE CRIMINAL PROCESS do so. We are not postulating a criminal process that operates in any kind of society at all, but rather one that operates within the framework of contemporary American society. This leaves plenty of room for polarization, but it does require the observance of some limits. A model of the criminal process that left out of account relatively stable and enduring features of the American legal system would not have much relevance to our central task. For convenience, these elements of stability and continuity can be roughly equated with minimal agreed limits expressed in the Constitution of the United States and, more importantly, with unarticulated assumptions that can be perceived to underlie those limits. Of course, it is true that the Constitution is constantly appealed to by proponents and opponents of many measures that affect the criminal process. And only the naive would deny that there are few conclusive positions that can be reached by appeal to the Constitution. Yet assumptions do exist about the criminal process that are widely shared and that may be viewed as common ground for the operation of any model of the criminal process. Our first task is to clarify these assumptions. First, there is the assumption implicit in the ex post facto clause of the Constitution 7 that the function of defining conduct that may be treated as criminal is separate from and anterior to the process of identifying and dealing with persons as criminals. How wide or narrow the definition of criminal conduct must be is an important question of policy that yields highly variant results depending on the values held by those making the relevant decisions.' But that there must be a means of definition that is in some sense separate from and anterior to the operation of the process is clear. If that were not so, our efforts to deal with the phenomenon of organized crime would appear ludicrous indeed (which is not to say that we have by any means exhausted the potentialities for dealing with that problem within the limits of this basic assumption). A related assumption that limits the area of controversy is that the criminal process ordinarily ought to be invoked by those charged with the responsibility for doing so when it appears that a crime has been committed and that there is a reasonable prospect of apprehend- ing and convicting its perpetrator. Although the area of police and prosecutorial discretion not to invoke the criminal process is demon- strably broad, it is common ground that these officials have no general dispensing power. If the legislature has decided that certain conduct is to be treated as criminal, the decision-makers at every level of the 7 U.S. CONST. art. 1, § 9. 8 See Note, The Void-for-Vaqueness Doctrine it the Supreme Court, 109 U. PA. L. REv. 67 (1960). 1964] 8 UNIVERSITY OF ,PENNSYLVANIA LAW REVIEW criminal process are expected to accept that basic decision as a premise for action. The controversial nature of the occasional case in which that is not the role played by the relevant decision-makers only serves to highlight the strength with which the premise holds. 9 This assump- tion may be viewed as the other side of the ex post facto coin. Just as conduct that is not proscribed as criminal may not be dealt with in the criminal process, so must conduct that has been denominated as criminal be so treated by the participants in the process. 1 Next, there is the assumption that there are limits to the powers of government to investigate and apprehend persons suspected of committing crimes. I do not refer to the controversy (settled recently, at least in broad outline) as to whether the fourth amendment's pro- hibitions against unreasonable searches and seizures applies to the states with equal force as to the federal government." 1 Rather, I refer to the general assumption that there is a degree of scrutiny and a degree of control that have to be exercised with respect to the activities of law enforcement officers, that the security and privacy of the indi- vidual may not be invaded at will. It is possible to imagine a society in which not even lip service is paid to this assumption. Nazi Germany approached but never quite reached this position. But no one in our society would maintain that every individual may be taken into custody at any time and held without any limitation of time during the process of investigating his possible commission of crimes, or that there should be no form of redress for violation of at least some standards for official investigative conduct. Although this assumption may not appear to have much in the way of positive content, its ab- sence would render moot some of our most hotly controverted prob- lems. If there were not general agreement that there must be some limits on police power to detain and investigate, the very contro- versial provisions of the Uniform Arrest Act,' permitting the police to detain for questioning for a short period even though they do not have grounds for making an arrest, would be a magnanimous con- cession by the all-powerful state rather than, as it is now perceived, a substantial expansion of police power. Finally, there is a complex of assumptions embraced within terms like "the adversary system," "procedural due process," ".notice and an opportunity to be heard," "day in court," and the like. Common to them all is the notion that the alleged criminal is not merely an object to be acted upon, but an independent entity in the process who may, 9 E.g., Poe v. Ullman, 367 U.S. 497 (1961). 1O HALL, GENERAL PRINCIPLES OF CRIMINAL LAW 382-87 (2d ed. 1960). 1l Mapp v. Ohio, 367 U.S. 643 (1961) ; Ker v. California, 374 U.S. 23 (1963). 12See pp. 28-29 infra. [Vol.l13 :I TWO MODELS OF THE CRIMINAL PROCESS if he so desires, force the operators of the process to demonstrate to an independent authority (judge and jury) that he is guilty of the charges against him. It is a minimal asumption. It speaks in terms of "may," not "must." It permits but does not require the accused, acting by himself or through his own agent, to play an active role in the process; by virtue of that fact, the process becomes or has the capacity to become a contest between, if not equals, at least independent actors. Now, as we shall see, much of the space between the two models is occupied by stronger or weaker notions of how this contest is to be arranged, how often it is to be played, and by what rules. The Crime Control Model tends to deemphasize this adversary aspect of the process; the Due Process Model tends to make it central. The common ground, and it is an important one, is that the process has, for everyone subjected to it, at least the potentiality of becoming to some extent an adversary struggle. So much for common ground. There is a good deal of it, even on the narrowest view. Its existence should not be overlooked because it is, by definition, what permits partial resolutions of the tension be- tween the two models to take place. The rhetoric of the criminal process consists largely of claims that disputed territory is "really" common ground; that, for example, the premise of an adversary system "necessarily" embraces the appointment of counsel for everyone accused of crime, or conversely, that the obligation to pursue persons sus- pected of committing crimes "necessarily" embraces interrogation of suspects without the intervention of counsel. We may smile indul- gently at such claims; they are rhetoric and no more. But the form in which they are made suggests an important truth: that there is a common ground of value assumption about the criminal process that makes continued discourse about its problems possible. 2. Crime Control Values The value system that underlies the Crime Control Model is based on the proposition that the repression of criminal conduct is by far the most important function to be performed by the criminal process. The failure of law enforcement to bring criminal conduct under tight control is viewed as leading to the breakdown of public order and thence to the disappearance of an important condition of human free- dom. If the laws go unenforced, which is to say, if it is perceived that there is a high percentage of failure to apprehend and convict in the criminal process, a general disregard for legal controls tends to develop. The law-abiding citizen then becomes the victim of all sorts of un- justifiable invasions of his interests. His security of person and prop- 1964] 10 UNIVERSITY OF PENNSYLVANIA LAW REVIEW erty is sharply diminished and, therefore, so is his liberty to function as a member of society. The claim ultimately is that the criminal process is a positive guarantor of social freedom.1 3 In order to achieve this high purpose, the Crime Control Model requires that primary attention be paid to the efficiency with which the criminal process operates to screen suspects, determine guilt, and secure appropriate dispositions of persons convicted of crime. Efficiency of operation is not, of course, a criterion that can be applied in a vacuum. By "efficiency" we mean the system's capacity to apprehend, try, convict, and dispose of a high proportion of criminal offenders whose offenses become known. In a society in which only the grossest forms of antisocial behavior were made criminal and in which the crime rate was exceedingly low, the criminal process might require many more man-hours of police, prosecutorial, and judicial time per case than ours does, and still operate with tolerable efficiency. On the other hand, a society that was prepared to increase substantially the resources devoted to the suppression of crime might cope with a rising crime rate without sacrifice of efficiency while continuing to maintain an elaborate and time-consuming set of criminal processes. However, neither of these hypotheses corresponds with social reality in this country. We use the criminal sanction to cover an increasingly wide spectrum of behavior thought to be antisocial, and the amount of crime is very large indeed. At the same time, while precise measures are not available, it does not appear that we are disposed in the public sector of the economy to increase very drastically the quantity, much less the quality, of the resources devoted to the suppression of criminal activity through the operation of the criminal process. These factors have an important bearing on the criteria of efficiency and, therefore, on the nature of the Crime Control Model. The model, in order to operate successfully, must produce a high rate of apprehension and conviction and must do so in a context where the magnitudes being dealt with are very large, and the resources for dealing with them are very limited. There must then be a premium on speed and finality. Speed, in turn, depends on informality and on uniformity; finality depends on minimizing the occasions for challenge. The process must not be cluttered with ceremonious rituals that do not advance the progress of a case. Facts can be established more quickly through interrogation in a police station than through the formal process of examination and cross-examination in a court; it follows that extrajudicial processes should be preferred to judicial processes, informal to formal operations. Informality is not enough; there must 13 For a representative statement see Barrett, supra note 2, at 11-16. [V0l.l13:1 TWO MODELS OF THE CRIMINAL PROCESS also be uniformity. Routine stereotyped procedures are essential if large numbers are being handled. The model that will operate successfully on these presuppositions must be an administrative, almost a managerial, model. The image that comes to mind is an assembly line or a conveyor belt down which moves an endless stream of cases, never stopping, carrying the cases to workers who stand at fixed sta- tions and who perform on each case as it comes by the same small but essential operation that brings it one step closer to being a finished product, or, to exchange the metaphor for the reality, a closed file. The criminal process, on this model, is seen as a screening process in which each successive stage-prearrest investigation, arrest, post- arrest investigation, preparation for trial, trial or entry of plea, con- viction, and disposition-involves a series of routinized operations whose success is gauged primarily by their tendency to pass the case along to a successful conclusion. What is a successful conclusion? One that throws off at an early stage those cases in which it appears unlikely that the person apprehended is an offender and then secures, as expeditiously as possible, the conviction of the rest with a minimum of occasions for challenge, let alone postaudit. By the application of administrative expertness, primarily that of the police and prosecutors, an early determination of probable innocence or guilt emerges. The probably innocent are screened out. The probably guilty are passed quickly through the remaining stages of the process. The key to the opera- tion of the model as to those who are not screened out is what I shall call a presumption of guilt. The concept requires some explanation, since it may appear startling to assert that what appears to be the precise converse of our generally accepted ideology of a presumption of innocence can be an essential element of a model that does corre- spond in some regards to the real-life operation of the criminal process. The presumption of guilt allows the Crime Control Model to deal efficiently with large numbers. The supposition is that the screening processes operated by police and prosecutors are reliable indicators of probable guilt. Once a man has been investigated with- out being found to be probably innocent, or, to put it differently, once a determination has been made that there is enough evidence of guilt so that he should be held for further action rather than released from the process, then all subsequent activity directed toward him is based on the view that he is probably guilty. The precise point at which this occurs will vary from case to case; in many cases it will occur as soon as the suspect is arrested or even before, if the evidence of probable guilt that has come to the attention of the authorities is 12 UNIVERSITY OF PENNSYLVANIA LAW REVIEW sufficiently strong. But in any case, the presumption of guilt will begin to operate well before the "suspect" becomes a "defendant." The presumption of guilt is not, of course, a thing. Nor is it even a rule of law in the usual sense. It simply exemplifies a complex of attitudes, a mood. If there is confidence in the reliability of in- formal administrative factfinding activities that take place in the early stages of the criminal process, the remaining stages of the process can be relatively perfunctory without any loss in operating efficiency. The presumption of guilt, as it operates in the Crime Control Model, is the expression of that confidence. It would be a mistake to think of the presumption of guilt as the opposite of the presumption of innocence that we are so used to thinking of as the polestar of the criminal process and which, as we shall see, occupies an important position in the Due Process Model. The presumption of innocence is not its opposite; it is irrelevant to the presumption of guilt; the two concepts embody different rather than opposite ideas. The difference can perhaps be epitomized by an example. A murderer, for reasons best known to himself, chooses to shoot his victim in plain view of a large number of people. When the police arrive, he hands them his gun and says: "I did it, and I'm glad." His account of what happened is corroborated by several eye- witnesses. He is placed under arrest and led off to jail. Under these circumstances, which may seem extreme but which in fact char- acterize with rough accuracy the factfinding situation in a large proportion of criminal cases, it would be plainly absurd to maintain that more probably than not the suspect did not commit the killing. But that is not what the presumption of innocence means. It means that until there has been an adjudication of guilt by an authority legally competent to make such an adjudication, the suspect is to be treated, for reasons that have nothing whatever to do with the probable outcome of the case, as if his guilt is an open question. The presumption of innocence is a direction to officials how they are to proceed, not a prediction of outcome. The presumption of guilt, however, is basically a prediction of outcome. The presumption of innocence is really a direction to the authorities to ignore the pre- sumption of guilt in their treatment of the suspect. It tells them, in effect, to close their eyes to what will frequently seem to be factual probabilities. The reasons why it tells them that are among the animating presuppositions of the Due Process Model, and we will come to them shortly. It is enough to note at this point that the presumption of guilt is descriptive and factual; the presumption of innocence is normative and legal. The pure Crime Control Model finds [Vo1.113:1 TWO MODELS OF THE CRIMINAL PROCESS unacceptable the presumption of innocence although, as we shall see, its real-life emanations are brought into uneasy compromise with the dictates of this dominant ideological position. For this model the presumption of guilt assures the dominant goal of repressing crime through highly summary processes without any great loss of efficiency (as previously defined), for in the run of cases, the preliminary screen- ing processes operated by the police and the prosecuting officials con- tain adequate guarantees of reliable factfinding. Indeed, the position is a stronger one. It is that subsequent processes, particularly of a formal adjudicatory nature, are unlikely to produce as reliable fact- finding as the expert administrative process that precedes them. The criminal process thus must put special weight on the quality of adminis- trative factfinding. It becomes important, then, to place as few re- strictions as possible on the character of the administrative factfinding processes and to limit restrictions to those that enhance reliability, excluding those designed for other purposes. As we shall see, the desire to avoid restrictions on administrative factfinding is a consistent theme in the development of the Crime Control Model. For this model the early administrative factfinding stages are centrally vital. The complementary proposition is that the subsequent stages are relatively unimportant and should be truncated as much as possible. This, too, produces tensions with presently dominant ideology. The pure Crime Control Model has very little use for many conspicuous features of the adjudicative process and in real life works a number of ingenious compromises with it. Even in the pure model, however, there have to be devices for dealing with the suspect after the preliminary screening process has resulted in a determination of probable guilt. The focal device, as we shall see, is the plea of guilty; through its use adjudicative factfinding is reduced to a mini- mum. It might be said of the Crime Control Model that, reduced to its barest essentials and when operating at its most successful pitch, it consists of two elements: (a) an administrative factfinding process leading to exoneration of the suspect, or to (b) the entry of a plea of guilty. 3. Due Process Values If the Crime Control Model resembles an assembly line, the Due Process Model looks very much like an obstacle course. Each of its successive stages is designed to present formidable impediments to carrying the accused any further along in the process. Its ideology is not the converse of that underlying the Crime Control Model. It does not deny the social desirability of repressing crime, although its critics have been known to claim so. Its ideology is composed of a 14 UNIVERSITY OF PENNSYLVANIA LAW REVIEW complex of ideas, some of them based on judgments about the efficacy of crime control devices. The ideology of due process is far more deeply impressed on the formal structure of the law than is the ideology of crime control; yet, an accurate tracing of the strands of which it is made is strangely difficult. 4 What follows is only an attempt at an approximation. The Due Process Model encounters its rival on the Crime Con- trol Model's own ground in respect to the reliability of factfinding processes. The Crime Control Model, as we have suggested in a preliminary way, places heavy reliance on the ability of investigative and prosecutorial officers, acting in an informal setting in which their distinctive skills are given full sway, to elicit and reconstruct a tolerably accurate account of what actually took place in an alleged criminal event. The Due Process Model rejects this premise and substitutes for it a view of informal, nonadjudicative factfinding that stresses the possibility of error: people are notoriously poor observers of dis- turbing events-the more emotion-arousing the context, the greater the possibility that recollection will be incorrect; confessions and admis- sions by persons in police custody may be induced by physical or psychological coercion, so that the police end up hearing what the suspect thinks they want to hear rather than the truth; witnesses may be animated by a bias or interest that no one would trouble to discover except one specially charged with protecting the interests of the accused-which the police are not. Considerations of this kind all lead to the rejection of informal factfinding processes as definitive of factual guilt and to the insistence on formal, adjudicative, adversary factfinding processes in which the factual case against the accused is publicly heard by an impartial tribunal and is evaluated only after the accused has had a full opportunity to discredit the case against him. Even then the distrust of factfinding processes that animates the Due Process Model is not dissipated. The possibilities of human error being what they are, further scrutiny is necessary, or at least must be available, lest in the heat of battle facts have been overlooked or suppressed. How far this subsequent scrutiny must be available is hotly controverted today; in the pure Due Process Model the answer would be: at least as long as there is an allegation of factual error that has not received an adjudicative hearing in a factfinding context. The demand for finality is thus very low in the Due Process Model. This strand of due process ideology is not enough to sustain the model. If all that were at issue between the two models was a series of 14 For a perceptive account dealing with a wider spectrum of problems than those posed by the criminal process, see Kadish, Methodology and Criteria in Due Process Adjudication--A Survey and Criticism, 66 YALE L.J. 319 (1957). [Vol.l13:1 TWO MODELS OF THE CRIMINAL PROCESS questions about the reliability of factfinding processes, we would have but one model of the criminal process, the nature of whose con- stituent elements would pose questions of fact, not of value. Even if the discussion is confined for the moment to the question of reliability, it is apparent that more is at stake than simply an evaluation of what kinds of factfinding processes, alone or in combination, are likely to produce the most nearly reliable results. The stumbling-block is this: how much reliability is compatible with efficiency? Granted that informal factfinding will make some mistakes that will be remedied if backed up by adjudicative factfinding, the desirability of providing this backup is not affirmed or negated by factual demonstrations or pre- dictions that the increase in reliability will be x percent or x plus n percent. It still remains to ask how much weight is to be given to the competing demands of reliability (a high degree of probability in each case that factual guilt has been accurately determined) and efficiency (a process that deals expeditiously with the large numbers of cases that it ingests). Just as the Crime Control Model is more optimistic about the unlikelihood of error in a significant number of cases, it is also more lenient in establishing a tolerable level of error. The Due Process Model insists on the prevention and elimination of mistakes to the extent possible; the Crime Control Model accepts the probability of mistakes up to the level at which they interfere with the goal of repressing crime, either because too many guilty people are escaping or, more subtly, because general awareness of the unreliability of the process leads to a decrease in the deterrent efficacy of the criminal law. On this view reliability and efficiency are not polar opposites but rather complementary characteristics. The system is reliable because efficient; reliability becomes a matter of independent concern only when it becomes so attenuated as to impair efficiency. All of this the Due Process Model rejects. If efficiency suggests shortcuts around re- liability, those demands must be rejected. The aim of the process is at least as much to protect the factually innocent as it is to convict the factually guilty. It somewhat resembles quality control in industrial technology: tolerable deviation from standard varies with the impor- tance of conformity to standard in the destined use of the product. The Due Process Model resembles a factory that has to devote a sub- stantial part of its input to quality control. This necessarily reduces quantitative output. This is only the beginning of the ideological difference between the two models. The Due Process Model could disclaim any attempt to provide enhanced reliability for the factfinding process and still produce a set of institutions and processes that would differ sharply 1964] 16 UNIVERSITY OF PENNSYLVANIA LAW REVIEW from those posited by the demands of the Crime Control Model. In- deed, it may not be too great an oversimplification to assert that in point of historical development the doctrinal pressures that have emanated from the demands of the Due Process Model have tended to evolve from an original matrix of concern with the maximization of reliability into something quite different and more far-reaching. 5 This complex of values can be symbolized although not adequately described by the concept of the primacy of the individual and the complementary concept of limitation on official power. The combination of stigma and loss of liberty that is embodied in the end result of the criminal process is viewed as being the heaviest deprivation that government can inflict on the individual. Further- more, the processes that culminate in these highly afflictive sanctions are in themselves coercive, restricting, and demeaning. Power is always subject to abuse, sometimes subtle, other times, as in the criminal process, open and ugly. Precisely because of its potency in subjecting the individual to the coercive power of the state, the crim- inal process must, on this model, be subjected to controls and safe- guards that prevent it from operating with maximal efficiency. Accord- ing to this ideology, maximal efficiency means maximal tyranny. And, while no one would assert that minimal efficiency means minimal tyranny, the proponents of the Due Process Model would accept with considerable equanimity a substantial diminution in the efficiency with which the criminal process operates in the interest of preventing offi- cial oppression of the individual. The most modest-seeming but potentially far-reaching mechanism by which the Due Process Model implements these antiauthoritarian values is the doctrine of legal guilt. According to this doctrine, an individual is not to be held guilty of crime merely on a showing that in all probability, based upon reliable evidence, he did factually what he is said to have done. Instead, he is to be held guilty if and only if these factual determinations are made in procedurally regular fashion and by authorities acting within competences duly allocated to them. Furthermore, he is not to be held guilty, even though the factual determination is or might be adverse to him, if various rules designed to safeguard the integrity of the process are not given effect: the tribunal that convicts him must have the power to deal with his kind of case ("jurisdiction") and must be geographically ap- 15 It is instructive to compare, for example, the emphasis on diminished reliability in early coerced confession cases like Brown v. Mississippi, 297 U.S. 278 (1936), with the subsequent development of a rationale that stresses the assertedly limited roles assigned to the state and the accused in an adversary system, e.g., Rogers v. Richmond, 365 U.S. 534 (1961). [Vo1.113 :I TWO MODELS OF THE CRIMINAL PROCESS propriate ("venue"); too long a time must not have elapsed since the offense was committed ("statute of limitations"); he must not have been previously convicted or acquitted of the same or a substantially similar offense ("double jeopardy"); he must not fall within a cate- gory of persons, such as children or the insane, who are legally im- mune to conviction ("criminal responsibility"); and so on. None of these requirements has anything to do with the factual question of whether he did or did not engage in the conduct that is charged as the offense against him; yet favorable answers to any of them will mean that he is legally innocent. Wherever the competence to make adequate factual determinations lies, it is apparent that only a tribunal that is aware of these guilt-defeating doctrines and is willing to apply them can be viewed as competent to make determinations of legal guilt. The police and the prosecutors are ruled out by lack of capacity in the first instance and by lack of assurance of willingness in the second. Only an impartial tribunal can be trusted to make determinations of legal as opposed to factual guilt. In this concept of legal guilt lies part of the explanation for the ap- parently quixotic presumption of innocence of which we spoke earlier. A man who after police investigation is charged with having com- mitted a crime can hardly be said to be presumptively innocent, if what we mean is factual innocence. But if any of a myriad of legal doctrines may be appropriately invoked to exculpate this particular accused, it is apparent that as a matter of prediction it cannot be said with any confidence that more probably than not he will be found guilty. Beyond the question of predictability this model posits a func- tional reason for observing the presumption of innocence: by forcing the state to prove its case against the accused in an adjudicative con- text, the presumption of innocence serves to force into play all the qualifying and disabling doctrines that limit the use of the criminal sanction against the individual, thereby enhancing his opportunity to secure a favorable outcome. In this sense the presumption of innocence may be seen to operate as a kind of self-fulfilling prophecy. By open- ing up a procedural situation that permits the successful assertion of defenses that have nothing to do with factual guilt, it vindicates the proposition that the factually guilty may nonetheless be legally innocent and should therefore be given a chance to qualify for that kind of treatment.

The possibility of legal innocence is expanded enormously when the criminal process is viewed as the appropriate forum for correcting its own abuses. This notion may well account for a greater amount of the distance between the two models than any other. In theory the 1964] 18 UNIVERSITY OF PENNSYLVANIA LAW REVIEW Crime Control Model can tolerate rules that forbid illegal arrests, un- reasonable searches, coercive interrogations, and the like if their en- forcement is left primarily to managerial sanctions internally imposed. What it cannot tolerate is the vindication of those rules in the criminal process itself through the exclusion of evidence illegally obtained or through the reversal of convictions in cases where the criminal process has breached the rules laid down for its observance. The availability of these corrective devices fatally impairs the efficiency of the process. The Due Process Model, while it may in the first instance be addressed to the maintenance of reliable factfinding techniques, comes eventually to incorporate prophylactic and deterrent rules that result in the release of the factually guilty even in cases in which blotting out the illegality would still leave an adjudicative factfinder convinced of the accused's guilt. 16 Another strand in the complex of attitudes that underlies the Due Process Model is the idea-itself a shorthand statement for a complex of attitudes-of equality. This notion has only recently emerged as an explicit basis for pressing the demands o the Due Process Model, but it appears to represent, at least in its potential, a most powerful norm for influencing official conduct. Stated most starkly, the ideal of equality holds that "there can be no equal justice where the kind of trial a man gets depends on the amount of money he has." 17 The factual predicate underlying this assertion is that there are gross inequalities in the financial means of criminal defendants as a class, that in an adversary system of criminal justice, an effective defense is largely a function of the resources that can be mustered on behalf of the accused, and that a very large proportion of criminal defendants are, operationally speaking, "indigent" 18 in terms of their ability to finance an effective defense. This factual premise has been strongly reinforced by recent studies that in turn have been both a cause and an effect of an increasing emphasis upon norms for the criminal process based on the premise. The norms derived from the premise do not take the form of an insistence upon governmental responsibility to provide literally equal 16 This tendency, seen most starkly in the exclusionary rule for illegally seized evidence, Mapp v. Ohio, 367 U.S. 643 (1961), is also involved in the rejection of the "special circumstances" approach to testing the deprivation of counsel, Gideon v. Wainwright, 372 U.S. 335 (1963), and in the apparently similar trend in confession cases, Mallory v. United States, 354 U.S. 449 (1957) ; Escobedo v. Illinois, 378 U.S. 478 (1964).17 Griffin v. Illinois, 351 U.S. 12, 19 (1956). 18 The vacuity of the concept of indigence is exposed in ATT'y GEN. CoMm. Ox PovaTY AND THE ADMINISTRATION OF FEDERAL CRIMINAL JUSTICE, REPORT 7-8, 40-41 (1963) [hereinafter cited as AiT'y GEN. REP.). [Vo1.113:1 TWO MODELS OF THE CRIMINAL PROCESS opportunities for all criminal defendants to challenge the process. Rather, they take as their point of departure the notion that the crim- inal process, initiated as it is by government and containing as it does the likelihood of severe deprivations at the hands of government, im- poses some kind of public obligation to ensure that financial inability does not destroy the capacity of an accused to assert what may be meritorious challenges to the processes being invoked against him. 9 The demands made by a norm of this kind are likely by its very nature to be quite sweeping. Although its imperatives may be initially limited to determining whether in a particular case the accused was injured or prejudiced by his relative inability to make an appro- priate challenge, the norm of equality very quickly moves to another level on which the demand is that the process in general be adapted to minimize discriminations rather than that a mere series of post hoc determinations of discrimination be made or makeable. .It should be observed that the impact of the equality norm will vary greatly depending upon the point in time at which it is intro- duced into a model of the criminal process. If one were starting from scratch to decide how the process ought to work, the norm of equality would have nothing very important to say on such questions as, for example, whether an accused should have the effective assistance of counsel in deciding whether to enter a plea of guilty. One could decide, on quite independent considerations, that it is or is not a good thing to afford that facility to the generality of persons accused of crime. But the impact of the equality norm becomes far greater when it is brought to bear on a process whose contours have already been shaped. If our model of the criminal process affords defendants who are in a financial position to consult a lawyer before entering a plea the right to do so, then the equality norm exerts powerful pressure to provide such an opportunity to all defendants and to regard the failure to do so as a malfunctioning of the process from whose consequences the accused is entitled to be relieved. In a sense that has been the role of the equality norm in affecting the real-world criminal process. It has made its appearance on the scene comparatively late"' and has therefore encountered a situation in which, in terms of the system as it operates, the relative financial inability of most persons accused of crime sharply distinguishes their treatment froin the small minority of the financially capable. For that reason its impact has already been substantial and may be expected to be even more so in the future. 19 E.g., id. at 8-11. 20 Griffin v. Illinois, 351 U.S. 12 (1956), is generally regarded as being the first decision of the Supreme Court explicitly and exclusively grounded on the equality norm. 20 UNIVERSITY OF PENNSYLVANIA LAW REVIEW There is a final strand of thought in the Due Process Model whose presence is often ignored but which needs to be candidly faced if thought on the subject is not to be obscured. That is a mood of skepticism about the morality and the utility of the criminal sanction, taken either as a whole or in some of its applications. The subject is a large and complicated one, comprehending as it does much of the intellectual history of our times."' To put the matter in parvo, one cannot improve upon the statement by Professor Paul Bator: [I]n summary we are told that the criminal law's notion of just condemnation and punishment is a cruel hypocrisy visited by a smug society on the psychologically and econom- ically crippled; that its premise of a morally autonomous will with at least some measure of choice whether to comply with the values expressed in a penal code is unscientfic and out- moded; that its reliance on punishment as an educational and deterrent agent is misplaced, particularly in the case of the very members of society most likely to engage in criminal conduct; and that its failure to provide for individualized and humane rehabiliation of offenders is inhuman and wasteful. 22 This skepticism, which may be fairly said to be widespread among the most influential and articulate of contemporary leaders of informed opinion, leads to an attitude toward the processes of the criminal law which, to quote Mr. Bator again, engenders a peculiar receptivity toward claims of injustice which arise within the traditional structure of the system itself; funda- mental disagreement and unease about the very bases of the criminal law has, inevitably, created acute pressure at least to expand and liberalize those of its processes and doctrines which serve to make more tentative its judgments or limit its power. 3 In short, doubts about the ends for which power is being exercised create pressure to limit the discretion with which that power is exercised. The point need not be pressed to the extreme of doubts about or rejection of the premises upon which the criminal sanction in gen- eral rests. Unease may be stirred simply by reflection on the variety of uses to which the criminal sanction is put and by judgment that an increasingly large proportion of those uses may represent an unwise 21 A portion of the work in progress of which this paper is a part is concerned with the impact of modern skeptical doubts on the ideology of the criminal law. 22 Bator, Finality in Criminal Law and Federal Habeas Corpus for State Prison- ers, 76 HARv. L. Rsv. 441, 442 (1963). 23Id. at 442-43. [Vol.l13:1 TWO MODELS OF THE CRIMINAL PROCESS invocation of so extreme a sanction.' 4 It would be an interesting irony if doubts about the utility of certain uses of the criminal sanction prove to contribute to a restrictive trend in the criminal process that in the end requires a choice among uses and finally an abandonment of some of the very uses that stirred the original doubts. There are two kinds of problems that need to be dealt with in any model of the criminal process. One is what the rules shall be. The other is how the rules shall be implemented. The second is at least as important as the first. As we shall see time and again in our detailed development of the models, the distinctive difference between the two models is not only in the rules of conduct that they lay down, but also in the sanctions that are to be invoked when a claim is pre- sented that the rules have been breached and, no less importantly, in the timing that is permitted or required for the invocation of those sanctions. As I have already suggested, the Due Process Model locates at least some of the sanctions for breach of the operative rules in the criminal process itself. The relation between these two aspects of the process-the rules and the sanctions for their breach-is a purely formal one unless there is some mechanism for bringing them into play with each other. The hinge between them in the Due Process Model is the availability of legal counsel. This has a double aspect: many of the rules that the model requires are couched in terms of the availability of counsel to do various things at various stages of the process-this is the conventionally recognized aspect; beyond it, there is a pervasive assumption as to the necessity for counsel in order to invoke sanctions for breach of any of the rules. The more freely available these sanctions are, the more important is the role of counsel in seeing to it that the sanctions are appropriately invoked. If the process is seen as a series of occasions for checking its own operation, the role of counsel is a much more nearly central one than is the case in a process that is seen as primarily concerned with expeditious deter- mination of factual guilt. And if equality of operation is a governing norm, the availability of counsel to some is seen as requiring it for all. Of all the controverted aspects of the criminal process, the right to counsel, including the role of government in its provision, is the most dependent on what one's model of the process looks like, and the least susceptible of resolution unless one has confronted the antinomies of the two models. I do not mean to suggest that questions about the right to counsel disappear if one adopts a model of the process that conforms more or 24See pp. 66-68 infra. 22 UNIVERSITY OF PENNSYLVANIA LAW REVIEW less closely to the Crime Control Model, but only that such questions become absolutely central if one's model moves very far down the spectrum of possibilities toward the pure Due Process Model. The reason for this centrality is to be found in the shared assumption underlying both models that the process is an adversary one in which the initiative in invoking relevant rules rests primarily on the parties concerned, the state and the accused. One could construct models that placed central responsibility on adjudicative agents such as com- mitting magistrates and trial judges. And there are, as we shall see, marginal but nonetheless important adjustments in the role of the adjudicative agents that enter into the models with which we -are concerned. 25 For present purposes it is enough to say that these ad- justments are marginal, that the animating presuppositions that under- lie both models in the context of the American criminal system rele- gate the adjudicative agents to a relatively passive role and therefore place central importance on the role of counsel. One last introductory note. What assumptions do we make about the sources of authority to shape the real-world operations of the criminal process? What agencies of government have the power to pick and choose between their competing demands? Once again, the limiting features of the American context come into play. Ours is not a system of legislative supremacy. The distinctively American institution of judicial review exercises a limiting and, ultimately, a shaping influence on the criminal process. Because the Crime Control Model is basically an affirmative model, emphasizing at every turn the existence and exercise of official power, its validating authority is ultimately legislative (although proximately administrative). Because the Due Process Model is basically a negative model, asserting limits on the nature of official power and on the modes of its exercise, its validating authority is judicial and requires an appeal to supra- legislative law, to the law of the Constitution. To the extent that tensions between the two models are resolved by deference to the Due Process Model, the authoritative force at work is the judicial power, working in the distinctively judicial mode of invoking the sanction of nullity. That is at once the strength and the weakness of the Due Process Model: its strength because in our system the appeal to the Constitution provides the last and the overriding word; its weakness because saying no in specific cases is an exercise in futility unless there is a general willingness on the part of the officials who operate the process to apply negative prescriptions across the board. It is no accident that statements reinforcing the Due Process Model come from 25 As, for example, in the role of judge in the plea of guilty. See pp. 46-51 infra. [Vol.l13:1 TWO MODELS OF THE CRIMINAL PROCESS the courts while at the same time facts denying it are established by the police and prosecutors.

II. THE MODELS IN OPERATION We turn now to some details of the Crime Control and Due Process Models. This is an effort, first, to convey a sense of the extraordinary complexity of the criminal process, no matter which model one visualizes as corresponding more closely to reality; even the Crime Control Model is a formidable consumer of human re- sources. Again, we shall try to document the existence throughout the process of recurrent themes that divide the two models, posing an essentially limited number of basic choices for shaping its real-life structure. Finally, this assortment of instances will serve to document the minor premise of the Article: that the present real-world criminal process tends by and large to resemble the Crime Control Model but that the current trend is pushing it a significant distance down the spectrum toward the Due Process Model. There are various ways of dividing the criminal process for purposes of description and analysis. We shall view it as consisting of three major stages or periods: the period from arrest through the decision to charge the suspect with a crime; the period from the decision to charge through the determination of guilt; and the stage of review and correction of errors that have occurred during the earlier periods. From the first period I have chosen two problems to illustrate the contrasting requirements of the two models: arrests for investigation, and detention and interrogation after a "lawful" arrest. From the second period I have again selected two problems: pretrial detention and the plea of guiltyT. From the third stage I have selected problems of direct appeal and collateral attack that raise issues of equality of access to the courts, of the special problems of criminal justice in a federal system, and of retroactivity in the application of changed norms in the criminal process. Finally, there is a postscript on the pervasive and strategically crucial problem of access to counsel. The themes here dealt with could be developed through applica- tion of the model technique to many other problems of the criminal process, including electronic surveillance, discovery in criminal cases, the presentation of the insanity defense, and the institution of the jury trial. As to each of these, and others as well, it can be asserted with some confidence that the antinomies of the two models show up in the same or similar form as those chosen for discussion, and also that an examination of the situation and the trend would produce 24 UNIVERSITY OF PENNSYLVANIA LAW REVIEW conclusions in line with those reached on the basis of the examples presented here. A. From Arrest to Charge 1. Arrests for Investigation The act of taking a person into physical custody is ordi- narily spoken of as an arrest. The term "arrest" carries with it important legal consequences, so that great controversy attends the question whether certain forms of physical restraint, such as stopping a person on the street for questioning, or taking him to the station house for a brief period of questioning without then and there intending to prefer any particular charge against him, is "really" an arrest. Since our discussion of the competing norms will refer to a number of different kinds of restraining conduct that may or may not be thought of as desirable depending on the dictates of the particular model, there is no reason not to refer indifferently to all of them as arrests, with the understanding that the term is used in the sense of physical description, not of operational legal norm. Two crucial issues arise at this stage of the process: (1) on what basis are the police entitled to make an arrest, and (2) what conse- quences, if any, will flow from their making an "illegal" arrest? These are the issues that divide our two polar models, and that may in addition be thought of as representing a paradigm of the kind of division that will occur over numerous other issues that arise in the process of investigation and apprehension. The Crime Control Model. Of course the police should be entitled to arrest a person when they have probable cause to think that he has committed a particular criminal offense, but it would be absurd to sug- gest that an arrest is permissible only in that situation. The slight in- vasion of personal freedom and privacy involved in stopping a person on the street to ask him questions, or even taking him to the station house for a period of questioning and other investigation, is justified in a wide variety of situations that only by the exercise of hypocrisy could be described as involving "probable cause." To give only a few examples: (1) people who are known to the police as previous offenders should be subject to arrest at any time for the limited purpose of deter- mining whether they have been engaging in antisocial activities, espe- cially when a crime has taken place of the sort they have committed in the past and it is known that it was physically possible for them to have committed it; (2) anyone who behaves in a way that arouses sus- picion that he may be up to no good should be subject to arrest for [Vol.l13:1 TWO MODELS OF THE CRIMINAL PROCESS investigation; it may turn out that he has committed an offense, but more importantly, the very fact of stopping him for questioning, either on the street or at the station house, may prevent the commission of a crime; (3) those who make a living out of criminal activity should be made to realize that their presence in the community is unwanted if they persist in their criminal occupations; periodic checks of their activity, whether or not they involve an arrest, will help to bring that attitude home to them. In short, the power of the police to arrest people for the purpose of investigation and prevention is one that must exist if the police are to do their job properly; the only question is whether arrests for in- vestigation and prevention should be made hypocritically and deviously, or openly and avowedly. It causes disrespect for law when there are great deviations between what the law on the books authorizes the police to do and what everyone knows they actually do. The police have no reason to abuse this power by arresting and holding law-abiding people. The innocent have nothing to fear. It is enough of a check on police discretion to let the dictates of police efficiency determine under what circumstances and for how long a person may be stopped and held for investigation. But if laws are thought to be required limiting police discretion to make an arrest, they should either provide very liberal outer limits so as to accommodate all possible cases or, preferably, should require nothing more explicit than behavior that is reasonable under all the circumstances. The question of appropriate sanctions for breach of whatever rules are devised to limit police arrest powers is, as a practical matter, at least as important to the ends of crime control as is the nature of the substantive rules themselves. The most appropriate sanction is discipline of the offending policemen by those best qualified to judge whether his conduct has lived up to professional police standards-his superiors in the police department. Discipline by his superiors may make him a better policeman; in cases where that seems improbable, he should be dismissed from the force. Civil remedies for the arrested person, administered in the ordinary courts, are also a possibility, al- though they are less likely to serve the end of educating the erring police officer. The one kind of sanction that should be completely inadmissible is the kind that takes place in the criminal process itself: dismissal of prosecution, suppression of evidence, etc. That kind of sanction for police misconduct simply gives the criminal a windfall without affecting the conduct of the erring police. This is particularly true in the light of the fact that any set of rules for the governance of police conduct is apt to be quite technical, leading to a certain 26 UNIVERSITY OF PENNSYLVANIA LAW REVIEW number of good faith mistakes. 26 The policeman who made the mistake may never know or be only dimly aware that his conduct re- sulted in a criminal's going free. His own conduct is much more likely to be changed by measures that affect him personally and that do not have the fortuitous effect of conferring benefit on the criminal and thus reducing the effectiveness of law enforcement. There are, generally speaking, two kinds of devices for giving the police adequate scope in making arrests for investigation. The first is what might be called the direct method: explicitly providing broad powers to stop and question persons, irrespective of whether they are reasonably suspected of having committed a particular crime. The second is the indirect method: framing broad enough definitions of criminal conduct to give the police the power to arrest on the orthodox "probable cause" basis a wide variety of people who are engaged in suspicious conduct. Vagrancy laws, disorderly conduct laws, and laws making it a crime not to give an account of oneself in response to police interrogation are all examples. It is not too important which of these methods is used; often a combination of the two will produce the desired result. The Due Process Model. It is a basic right of free men-basic in the sense that his other rights depend upon it-not to be subject to physical restraint except for good cause. The only measurable stand- ard of cause is the time-honored prescription that no one may be arrested except upon a determination-preferably made independently by a magistrate in deciding whether to issue a warrant, but in situations of necessity by a police officer acting upon probative data subject to subsequent judicial scrutiny-that a crime has probably been com- mitted, and that he is the person who probably committed it. Any less stringent standard opens the door to the probability of grave abuse, as repeated investigations of police practices have shown. A society that covertly tolerates indiscriminate arrest is hypocritical; but one that approves its legality is well on the way to becoming totali- tarian in nature. It is far from demonstrable that broad powers of arrest for investigation are necessary to the efficient operation of the police. The argument that they are necessary is open to a serious charge of inconsistency since it is also argued that changes in the law are necessary to bring it into conformity with prevalent though un- acknowledged practices. If arrests for investigation are actually now 26 See the famous mot of Mr. Justice Cardozo in People v. Defore, 242 N.Y. 13, 21, 150 N.E. 585, 587 (1926) : "the criminal is to go free because the constable has blundered." [Vo1.113:1 TWO MODELS OF THE CRIMINAL PROCESS tolerated on a wide scale, it makes no sense to assert that legalizing them is necessary to protect efficiency from being impaired. In the end, however, arguments about what is required by efficiency are wide of the mark. A totally efficient system of crime control would be a totally repressive one since it would require a total suspension of rights of privacy. We have to be prepared to pay a price for a regime that fosters personal privacy and champions the dignity and inviolability of the individual. That price inevitably involves some sacrifice in efficiency; consequently, an appeal to efficiency alone is never sufficient to justify any encroachment on the area of human freedom. It must be shown that efficient law enforcement will be so heavily impaired by failure to adopt the proposed measure that the minimal conditions of public order necessary to provide the environment in which indi- viduals can be allowed to enjoy the fruits of personal freedom will in themselves cease to exist or be gravely impaired. No one has seri- ously suggested that we are at or near that point. The practical consequence of enlarging police authority to detain individuals for questioning is not likely to be that all classes of the population will thereupon be subjected to interference. If that were the consequence, the practice would carry its own limiting features because the popular outcry would be so great that these measures could not long be resorted to. The danger, rather, is that they will be applied in a discriminatory fashion to precisely those elements in the population-the poor, the ignorant, the illiterate, the unpopular-who are least able to draw attention to their plight and to whose sufferings the vast majority of the population are the least responsive. Respect for law, never high among minority groups, would plunge to a new low if what the police are now thought to do sub rosa became an officially sanctioned practice. The need, then, is not to legalize practices that are presently illegal but widespread. Rather, it is to reaffirm their illegality and at the same time to take steps to reduce their incidence. This brings us to the question of sanctions for illegal arrests. To the extent possible, these sanctions should be located within the criminal process itself; since it is the efficiency of that process that they seek so mistakenly to promote, the process should penalize, and thus label as inefficient, arrests that are based on any standard less rigorous than probable cause. Of course and as a minimal requirement, any evidence that is obtained directly or indirectly on the basis of an illegal arrest should be suppressed. Beyond that, any criminal prosecution commenced on the basis of an illegal arrest should be dismissed, preferably with prejudice, but at the least with the consequence that the entire process, 28 UNIVERSITY OF PENNSYLVANIA LAW REVIEW if it is to be re-invoked, must be started over again from scratch, and all records, working papers, and the like prepared in the course of the first illegal proceeding impounded and destroyed. Many illegal arrests do not result in criminal prosecution (one of their undoubted vices) and are therefore not amenable to sanctions imposed in the criminal process itself. A variety of devices should be marshalled to provide effective sanctions against arrests for investi- gation. The ordinary tort action against the policeman has been demonstrated to be of very limited utility. It should be supplemented by provision for a statutory action against the governmental unit employing the offending policeman, with a high enough minimum re- covery to make suit worthwhile. Since an important public service is performed by attorneys who bring suits against errant police officers, there should also be provision for allowing adequate attorney's fees in cases where the action is successful. Measures of this kind will give governmental units a stake in proper police activity and an incentive to discourage illegal activity that they do not now have. Direct disciplinary measures against the offending police officer are also desirable, but it is unrealistic to expect these to be initiated by a departmental authority. Outside scrutiny is needed, both to insure that the law is being impartially enforced and, perhaps even more im- portantly, to reassure the general public that the police are not a law unto themselves. To this end there is need to set up civilian (i.e., nonpolice) boards to which complaints about illegal police activity may be directed and which can at least initiate, if not conduct, dis- ciplinary proceedings in cases where preliminary investigation shows that the complaint may be meritorious. The Situation and the Trend. The legal power of the police to stop persons on the street, search them for weapons, and require them to answer questions about their identity and business is ambiguous. 7 There is no doubt that the police, like any other person, may approach others and ask them questions. The question is what the police may do if the person refuses to stop or to answer questions. Strangely enough, there is no authoritative holding on whether the police may constitutionally be authorized to restrain a person who refuses to stop and answer questions when there is no probable cause to arrest him. The proposed Uniform Arrest Act would give the police the power to "stop any person abroad who he has reasonable ground to suspect is committing, has committed or is about to commit a crime, 27 The confused state of the law has been demonstrated by many commentators. E.g., Remington, The Law Relating to "On the Street" Detentio n, Questioning and Frisking of Suspected Persons and Police Arrest Priileges inGeneral, 51 3. Gnns. L., C & P.S. 386 (1960). [Vol.l13 :I TWO MODELS OF THE CRIMINAL PROCESS and may demand of him his name, address, business abroad and whither he is going." ' It goes on to provide that any person so ques- tioned who fails to give a satisfactory account of himself may be "de- tained" for further questioning and investigation, the period of such "detention" not to exceed two hours. The Act has been adopted by only three states and has apparently been construed out of existence in one of them. 29 It is silent on the question of remedies available to a person wrongfully detained under its provisions. "There is no doubt that it is common police practice to stop and question suspects as to whom there are no sufficient grounds for arrest." 3' The knowledgeable commentator who expressed that opinion goes on to say that its truth is not easy to document. It is also clear that many persons so stopped who do not give a satisfactory ac- count of themselves are taken to the station house for further investi- gation. In the District of Columbia, 3 where the police had an explicit category of arrest denominated as a "taking into custody without prob- able cause," a recent study showed that in a one-year period there were 3,743 such arrests. This represented less than one percent of the total arrests in the District-415,925--during the period. But since such arrests are apparently made only for investigation of felonies, the more relevant figure is the percentage of arrests for felonies that fall into the category of arrests for investigation. Here the per- centage is twenty-eight, or about one in every four felony arrests. Most such arrests were made on the street, but a substantial proportion were made in the suspects' homes. The present sanctions for illegal arrest are essentially as follows: exclusion in a criminal prosecution of evidence obtained by a search conducted incident to an illegal arrest; 32 tort action against the offending police officer; 33 complaint to the police department which, with rare exceptions, disposes of the complaint through departmental channels and without "civilian" scrutiny."' Recent Supreme Court decisions make it quite clear that if the de- tention of suspects in the absence of probable cause must be justified by the constitutional standards that prevail for a technical "arrest," 28 The text of the act is set forth in Warner, The Uniform Arrest Act, 28 VA. L. REV. 315, 320-21 (1942). 29 De Salvatore v. State, 52 Del. 550, 163 A.2d 244 (1960). 30 Remington, mupra note 27, at 389. 31 WASn., D.C., COMMISSIONERS' CoMM. ON POLIcE ARRESTS FOR INVESTIGATION, REPORT AND RECOcMENDATIONS 9 (1962). 32Mapp v. Ohio, 367 U.S. 643 (1961). 33See Foote, Tort Remedies for Police Violations of Individual Rights, 39 MINN. L. REv. 493 (1955). 34 See Note, The Administration of Complaints by Civilians Against the Police, 77 H v. L. REV. 499 (1964). 1964] 30 UNIVERSITY OF PENNSYLVANIA LAW REVIEW such detention would be held to be a violation of the fourth amend- ment. 5 And it seems quite likely that detention, whether on the street or at the station house, would be considered an arrest (or to use the language of the fotirth amendment, a "seizure") in the consti- tutional sense 36 If this judgment is right, it would follow that the provisions of the Uniform Arrest Act, referred to earlier, would now be held unconstitutional by the Supreme Court, even though there might have been substantial doubt about that proposition as recently as 1941, when the act was first proposed. It is one thing for a court to declare a procedure unconstitutional. It is quite another to translate that declaration into operative fact. By and large, all that courts have available is the sanction of nullity. They can reverse criminal convictions based on evidence obtained through an unlawful arrest. They can (and perhaps will) go farther and reverse such convictions or nullify at an earlier stage of the process whenever it is shown that an illegal arrest has taken place. But that is still only a retail operation, and the problem is a wholesale one. One indication of further trend may be that the Commissioners of the District of Columbia, acting in response to the study already de- scribed, have directed the police department to discontinue the practice of making arrests for investigation. 7 That action, not in itself perhaps of major significance, may provide a microcosmic example of a process that appears to be underway on many fronts in the administration of criminal justice in this country. The legal norm ostensibly says one thing. There is some suspicion that actual practice is quite different. However, it is only when investigation of the actual practice, spurred by developing legal norms, shows the magnitude of the discrepancy that the tension between norm and practice is partially resolved by reform. The increase in visibility itself contributes to the evolution and refinement of the norm, which in turn stimulates and keeps alive an interest in the development of actual practice. How long such a cycle of reform may sustain itself is not clear. It does seem, however, that such a cycle is now underway in the area of arrests for investigation and that it is slowly but perceptibly pushing practice in the direction of the Due Process Model. 35 See, e.g., Henry v. United States, 361 U.S. 98 (1959); Johnson v. United States, 333 U.S. 10, 15-16 (1948). 86The question of when a technical "arrest" takes place was left open in Henry v. United States, 361 U.S. 98 (1959), and Rios v. United States, 364 U.S. 253 (1960). Cf. United States v. Bonnano, 180 F. Supp. 71 (S.D.N.Y. 1960). But see United States v. Bufalino, 285 F.2d 408, 420 n.3 (2d Cir. 1960). 37 But not without some backing and filling. See Kamisar, O1 the Tactics of Police-Prosecutiom Oriented Critics of the Courts, 49 CoamELL L.Q. 436, 444-45 (1964). [Vol.l13:1 TWO MODELS OF THE CRIMINAL PROCESS 2. Detention and Interrogation After a "Lawful" Arrest Once a person suspected of crime has been taken into custody, someone has to decide whether a prosecution should be formally initiated and, if so, what specific offense or offenses should be charged. Typically, this is a job for the prosecutor; yet, at this point he will ordinarily not even know that the suspect is in custody. Furthermore, there is a question whether the charge against the suspect ought not to be evaluated by some impartial authority in order to determine whether he should be held for judicial action. Since there is also a question whether the suspect should be held in custody until his guilt is ad- judicated or released pending that determination, an occasion needs to be afforded for making that decision. Both of these decisions- whether or not to hold for the institution of charges and, if so, whether or not to release pending further steps in the process-are, in theory at least, made at a preliminary hearing before a judicial officer who, when he sits in this capacity, is known as a magistrate. This occasion provides typically a terminus ad quem for the initial investigatory and apprehending phase of the criminal process, a mechanism for turning the criminal investigation into a criminal prosecution. However, prior to the independent magisterial hearing, it must be determined to what extent the accused may be required to cooperate in the postarrest investigation: (1) May the police hold the accused in- definitely or must they bring him before a magistrate at some par- ticular time? (2) If the latter, what sanctions, if any, should be im- posed for failure to comply with the requisite time limits? (3) May the suspect be interrogated by the police during this time and, if so, under what limitations? (4) If the accused admits his guilt during this period, what restrictions, if any, are there on the use that may be made of this evidence at his trial? (5) Should the accused be entitled to the assistance of counsel during the time between his arrest and the preliminary hearing and, if so, under what conditions and with what consequences for failure by the police to adhere to those conditions? The Crime Control Model. The police cannot be expected to solve crimes by independent investigation alone. The best source of information is usually the suspect himself. Without his cooperation many crimes could not be solved at all. The police must have a reasonable opportunity to interrogate the suspect in private before he has a chance to fabricate a story or to decide that he will not cooperate. The psychologically optimal time for getting this kind of cooperation from the suspect is immediately after his arrest, before he has had a chance to rally his forces. Any kind of outside interference is likely to 32 UNIVERSITY OF PENNSYLVANIA LAW REVIEW diminish the prospect that the suspect will cooperate in the interroga- tion; therefore, he should not be entitled to summon his family or friends, and most importantly, he should not be entitled to consult a lawyer. The first thing that a lawyer will tell him is to say nothing to the police. Once he obtains that kind of reinforcement, the chances of getting any useful information from him sink to zero. Of course, the police should not be entitled to hold the suspect indefinitely for interrogation nor would they want to do so. The point of diminishing return in interrogation is reached fairly soon, and, anyhow, the police do not have extensive enough resources to be able to go on interrogating indefinitely. But no hard and fast rule can be laid down about how long the police should be permitted to inter- rogate the suspect before bringing him before a magistrate. The gravity of the crime, its complexity, the amount of criminal sophistica- tion that the suspect appears to have, all of these are relevant factors in determining how long he should be held. The standard ought to be the length of time under all the circumstances during which it is reasonable to suppose that legitimate techniques of interrogation may be expected to produce useful information, or that extrinsic investiga- tion may be expected to produce convincing proof either of the sus- pect's innocence or of his guilt. The suspect should not be held incommunicado under normal circumstances. His family is entitled to know where he is; but they should not be entitled to talk with him, since that may impair the effectiveness of the interrogation. Occasionally, it may be justifiable not to notify them at all, as where a confederate is still at large and does not know that his partner in crime has been apprehended. The point of all these illustrations, however, is that hard and fast rules cannot be laid down if police efficiency is not to be impaired. It follows that good faith mistakes in applying these rules in any given case should not be penalized. If the police err by holding a suspect too long, he has no complaint, because by hypothesis they have some basis for belief that he has committed a crime. The public has a com- plaint to the extent that police resources are thereby demonstrated to be used inefficiently; but the redress for that is intradepartmental dis- cipline in flagrant cases and a general program of internal administra- tive management that minimizes such occasions. Any trustworthy statement obtained from a suspect during a period of police interrogation should of course be admissible into evidence against him. Criminal investigation is a search for truth, and anything that aids the search should be encouraged. There is, to be sure, a danger that occasionally police will not live up to [Vol.l13:1 TWO MODELS OF THE CRIMINAL PROCESS professional standards and will use coercive measures to elicit a con- fession from a suspect. That is not to be condoned, but at the same time we should keep in mind that the evil of a coerced confession is that it may result in the conviction of an innocent man. There is no way of laying down hard and fast rules about what kinds of police conduct are coercive. It is a factual question in each case whether the accused's confession is unreliable. A defendant against whom a confession is introduced into evidence should have to convince the jury that the circumstances under which it was elicited were so coercive that more probably than not the confession was untrue. In reaching a determination on that issue, the trier of fact should of course be entitled to consider the other evidence in the case and, if it points toward guilt and tends to corroborate the confession, should be entitled to take that into account in determining whether, more likely than not, the confession was untrue. To say this is not to say that the unlawful use of force by the police on an accused is ever to be condoned. It is simply to say that its use is not in itself determinative of the reliability of a con- fession and should therefore not in itself be conclusive against the admissibility of a confession. The sanctions available for mistreating a person in custody are ample, if vigorously pursued, to ensure that this kind of conduct will be found only in rare instances. It is through the raising of professional standards by internal administrative methods, rather than through the happenstance outcome of a criminal prosecution, that improper police conduct is being eliminated. It follows a fortiori from what has been said that factors less probably coercive than the use of force, like an overly long period of detention unaccompanied by physical abuse, should not count con- clusively against the admissibility of a confession. The Due Process Model. A valid decision to arrest must be based on probable cause to believe that the suspect has committed a crime. To put it another way, the police should not arrest unless on the basis of the information at that time in their hands a case exists that, subject to the vicissitudes of the litigation process, seems likely to result in a conviction. It follows that if proper arrest standards have been employed, there is no necessity to obtain additional evidence from the mouth of the defendant. He is to be arrested so that he may be held to answer the case against him, not so that a case against him that does not exist at the time of his arrest can be developed. Once a suspect has been arrested, he should be brought before a magistrate without unnecessary delay, which is to say, as soon as it is physically possible to do so, once the preliminary formalities of 34 UNIVERSITY OF PENNSYLVANIA LAW REVIEW recording his arrest have been completed. Anyone who is held in arrest has the right to test the legality of his arrest, i.e., whether there is probable cause to hold him, in a judicial proceeding. As a prac- tical matter that right is diluted through delay unless the accused is promptly brought before a magistrate. Since a suspect is entitled to be at liberty pending the judicial determination of his guilt or inno- cence, there must be as promptly as possible after arrest a proceeding in which the conditions of his release, as for example on bail, are determined. This right, too, is diluted by delay unless the suspect is promptly brought before a magistrate. And the suspect is entitled to the assistance of counsel, a right that he most acutely needs to enjoy as soon as he is arrested. As a practical matter he is unlikely to re- ceive that right unless he is promptly advised of it. Once again, his prompt production before an impartial judicial officer is necessary if his right is not to be diluted by delay. It is never proper for the police to hold a suspect for the purpose of interrogation or investigation. Of course, some interval of time must always elapse between his arrest and his production before a magistrate, and it would be unrealistic to expect the police to maintain complete silence toward him during that period. However, there is a decisive difference between an interrogation conducted during the relatively brief span of time necessary to get the suspect before a magistrate and an interrogation whose length is measured by the time necessary to get him to confess. Any such interrogation should by that fact alone be held illegal. As soon as a suspect is arrested, he should be told by the police that he is under no obligation to answer questions, that he will suffer no detriment by refusing to answer questions, that he may answer questions in his own interest to clear himself of suspicion, but that anything he says may be used in evidence, and, above all, that he is entitled to see a lawyer if he wants to do so. If the suspect does make self-incriminatory statements while under arrest and before he is brought before a magistrate, their admissibility in evidence against him should be barred under any of the following conditions: (1) the failure of the police to apprise him of his rights, including his right to the assistance of a lawyer; (2) the fact that the confession was made during a period of detention that exceeded what was necessary to get him promptly before a magistrate; or (3) that the confession was made under other coercive circumstances, such as the use of force against him. Any confession made under these circumstances should be regarded as "involuntary" and should be excluded at the trial. Any further evidence secured on the basis of [V0l.l13 :I TWO MODELS OF THE CRIMINAL PROCESS an involuntary confession should likewise be excluded at the trial to deprive the police of any incentive to obtain such a confession. The rationale of exclusion is not that the confession is untrust- worthy, but that it is at odds with the postulates of an accusatory system of criminal justice in which it is up to the state to make its case against a defendant without forcing him to cooperate in the process. It follows, then, that the existence of other evidence of guilt has no bearing on the admissibility of the confession or on the necessity for reversing a conviction based in part on such a confession. It also follows that the procedure for determining the admissibility of a con- fession must be such as to avoid any possibility of prejudice to the defendant through the process of determining admissibility. Spe- cifically, in a jury trial the issue of the admissibility of a confession should be litigated on a record made before the judge and out of the hearing of the jury so that the trial judge has the clear and undivided responsibility for deciding whether the jury should hear the confession, and so that a reviewing court can have an unambiguous basis for de- ciding whether the trial judge reached the proper conclusion. The Situation and the Trend. The power of the police to inter- rogate a suspect between his arrest and his production before a magis- trate is generally recognized; but there is a strong and apparently accelerating judicial trend toward limiting the duration and the circum- stances of such interrogation. The Supreme Court has laid down increasingly strict standards for determining when a confession is "involuntary" and therefore inadmissible in evidence against the accused. 38 It has become clear that the criterion of voluntariness is not the trustworthiness of the confession, but rather its compatibility with the asserted postulates of an accusatorial system in which the case against the accused must be established "by evidence independently and freely secured" and which precludes the state "by coercion [to] prove its charge against an accused out of his own mouth." 9 Although the criterion has ostensibly been applied on a case-by-case basis, there appears to be a trend toward general and automatic standards for determining whether the circumstances of the interrogation were such as to be "inherently coercive." This trend has been most dramatically manifested in federal criminal prosecutions where the Supreme Court, in an exercise of its supervisory power over the administration of fed- eral criminal justice, has laid down a rule rejecting confessions that are secured during a period of detention that exceeds what is required 38 See, e.g., Haynes v. Washington, 373 U.S. 503 (1963), and Gallegos v. Colo- rado, 370 U.S. 49 (1963), neither of which involved particularly gross forms of "coercion!' S39Rogers v. Richmond, 365 U.S. 534, 541 (1961). 19641 36 UNIVERSITY OF PENNSYLVANIA LAW REVIEW to bring the accused before a magistrate "without unnecessary delay." 40 The states have so far been left free of this requirement and have, in general, not applied it to their criminal prosecutions. However, the length of detention has been a factor conspicuously stressed by the Supreme Court in reversing state criminal convictions based in part on a confession, and there have been intimations that the federal rule or something like it may eventually be applied to the states. 41 Apart from specific intimations in confession cases, there has been a general tendency to extend restrictions originally applied to federal prosecu- tions to state prosecutions as well. 4' There has been a concomitant tendency to move from rules that require case-by-case determination of prejudice to the accused to rules setting forth general standards of police and prosecutorial conduct. 43 These developments may well presage the ultimate extension to state criminal prosecutions of a rule outlawing the admissibility of confessions secured during a period of detention whose length and purpose is determined by the dictates of efficiency in securing confessions rather than by the unavoidable lapse of time between arrest and production before a magistrate. It has become a firmly established principle that convictions based in part on involuntary confessions (so deemed by increasingly rigorous standards) must be reversed regardless of the strength of other evi- dence of guilt in the case. And, in a most recent development, the states have been told that they must establish procedures for litigating the admissibility of confessions that do not give juries a clandestine opportunity to hold either the substance of a confession or the fact that one was made against the defendant. 44 Running through most of the problems that arise during this preliminary phase of the criminal process is the pervasive theme of access to counsel. It has a twofold relationship with the other prob- lems posed for resolution by our competing models of the process, acting as both their cause and effect. The importance of counsel is either enhanced or diminished depending on the view one takes of the rules that ought to govern arrest, search, and interrogation. On the other hand, if one starts with a position on the utility or disutility 40 McNabb v. United States, 318 U.S. 332 (1943); Mallory v. United States, 354 U.S. 449 (1957). 41 Perhaps directly, by giving that rule constitutional status, cf. Wong Sun v. United States, 371 U.S. 471 (1963), or indirectly, through a requirement of access to counsel immediately after arrest, cf. Escobedo v. Illinois, 378 U.S. 478 (1964). 42 Compare Mapp v. Ohio, 367 U.S. 343 (1961), with Wolf v. Colorado, 338 U.S. 25 (1949). Compare Gideon v. Wainwright, 372 U.S. 335 (1963), with Betts v. Brady, 316 U.S. 455 (1942). Compare Malloy v. Hogan, 378 U.S. 1 (1964), with Adamson v. California, 332 U.S. 46 (1947). 43 Compare Rochin v. California, 342 U.S. 165 (1952), and Irvine v. California, 347 U.S. 128 (1954), with Mapp v. Ohio, 367 U.S. 643 (1961). 44Jackson v. Denno, 378 U.S. 368 (1964). [Vol.l13:1 TWO MODELS OF THE CRIMINAL PROCESS of counsel at this stage of the process, that view is itself likely to be determinative of many of the other rules. And, as a practical matter, the availability of counsel is bound to have important consequences for the effectiveness with which the applicability of the governing rules, whatever they may be, is challenged. A concrete instance of this ubiquitous problem is the question whether the admissibility of a confession should be conditioned on access to counsel during the period between arrest and production before a magistrate. So far, the Supreme Court has not placed this general limitation on the admissibility of confessions. The point at which access to counsel becomes an absolute limitation on interrogation has been somewhat later-roughly the point at which the suspect, upon being formally charged with a crime by indictment or information, becomes "the accused." However, there are signs that this point may be pushed back to an earlier stage in the criminal process. In a number of cases in which confessions have been held involuntary, some stress has been laid upon the "special circumstance"-among others-that the suspect was denied an opportunity to consult a lawyer or was not informed of his right to remain silent. More significantly, the Court has just held, by a bare majority, that a "principal suspect" who is induced to confess by an interrogation conducted despite his express requests to see his lawyer and despite his lawyer's attempts to see him is entitled to have his confession excluded from evidence. 45 While the court spoke of the "particular circumstances" of the case, this decision casts considerable doubt on the continued vitality of cases decided only a few years ago in which confessions secured under similar conditions were upheld. 46 It seems very unlikely that this view, if it continues to be adhered to by a majority of the Court, can be confined to explicit requests for legal assistance. The ignorant, inexperienced defendant, for whom the Court has shown special solicitude in the past, no less than the sophisticate who knows his rights and tries to insist upon them, will presumably be afforded equivalent protection, which may ultimately mean that no confession will stand unless given by a defendant who, having been fully advised by the police of his right to remain silent and to consult a lawyer, nonetheless "freely" chooses to speak. Needless to say, any such rule as that would be only inches away from one that ruled out of evidence any statement made by an accused to the police before his production before a magistrate. 45Escobedo v. Illinois, 378 U.S. 478 (1964). 46 See Crooker v. California, 357 U.S. 433 (1958) ; Cicenia v. Lagay, 357 U.S. 504 (1958). 1964] 38 UNIVERSITY OF PENNSYLVANIA LAW REVIEW It may be worthwhile calling attention to features of this develop- ment that are typical of what seems to be a general tendency in the evolution of the criminal process: (1) a tendency to apply to the states the somewhat more rigorous rules first laid down for federal prosecutions; (2) a tendency to move from case-by-case adjudication to the laying down of broad rules of administration; (3) a tendency to narrow the area of police discretion. What all of this seems to portend is a criminal process that is being forced increasingly to take on the contours of the Due Process Model, at least in terms of the norms that ostensibly govern it. B. From Charge to Guilt Determination 1. Pretrial Detention Some interval of time must always elapse between the decision to hold a person for trial adjudication and the adjudication itself. What is to be done with the person who is charged with a crime but not yet convicted of it? The answer has important consequences for the shape of the subsequent proceedings; indeed, it may determine whether there will be any subsequent proceedings, since a decision that the defendant will remain in custody once he has been charged may itself induce him to plead guilty, thereby short-circuiting the part of the process concerned with guilt determination and moving directly to the question of ultimate disposition. In our system the question, baldly put, is bail or jail: will the defendant be able to provide the required financial assurance that he will appear for trial or will he, for lack of such provision, be kept in custody until the case against him has been prepared for trial and he is brought before the court either to stand trial or to plead guilty? The issues that divide the two models run much deeper, posing as they do questions that are begged by existing institutional arrangements. What reasons justify keeping a defendant in custody before his guilt has been formally adjudicated? Is the only relevant consideration the likelihood that he will not appear for trial? If so, are financial de- terrents the only, or the most appropriate, means of assuring presence at trial? If not, what other considerations are relevant? The possi- bility that the defendant may tamper with the evidence, as by intimi- dating prospective witnesses? The possibility that he may commit other offenses while he is at large? The degree of probability that he is guilty of the offense charged against him? Questions of this order, which are blurred in the day-to-day administration of the criminal law, may be clarified by examining them in the context of our polarized models. [Vo1.113:1 TWO MODELS OF THE CRIMINAL PROCESS The Crime Control Model. The vast majority of persons charged with crime are factually guilty. An arrest that results in a charge being placed has behind it a double assurance of reliability: the judgment of the police officer who made the arrest is supported by that of the prosecutor who has decided that there is enough evidence to hold the defendant for trial. For all practical purposes, the defendant is a criminal. just because the assembly line cannot move fast enough for him to be immediately disposed of is no reason for him to go free. If he does go free, he may not appear for trial, a risk that is heightened when he has a strong consciousness of guilt and a lively expectation of probable punishment. If he does not appear voluntarily, we will have to devote some of our limited resources to tracking him down and bringing him in. That may be tolerable when it occurs sporadically and on a small scale; but if large numbers of people are turned loose before trial, the chances are that the problem will get out of hand, and we will be faced with a vicious circle; the more people fail to appear, the more people will be encouraged not to appear, and the whole system will collapse. Failure to appear is not the only risk involved in pretrial liberty. The prospect that known criminals will commit further crimes while at large awaiting trial is in itself an adequate reason for not fostering pretrial liberty as the norm. The resulting danger to property and to human life is inexcusable because it is so easily avoidable. Even for first offenders and others who do not present any obvious danger of repetition while awaiting trial, there are good reasons why pretrial liberty should not be available as a matter of right. Courts are inclined to be lenient with first and other minor offenders. Their prosecutions are apt to be dismissed in a large pro- portion of cases because it is not worthwhile to use the limited re- sources available to law enforcement agencies to prosecute them. If they are not dismissed, they may nonetheless be put on probation or fined or given suspended sentences, all dispositions which fall short of exercising any significant effect on their future conduct. For many such persons, a short period spent in jail awaiting trial is not only a useful reminder that crime does not pay, but also the only such re- minder they are likely to receive. Other considerations apart, it is likely that a significantly higher percentage of defendants who now plead guilty would elect to stand trial if they could be at liberty pending trial. People who know that they are guilty tend to accept their punishment if, in order to gamble on the off-chance of an acquittal, they have to spend weeks or months in jail awaiting trial. But if they are released pending trial, the in- 1964] 40 UNIVERSITY OF PENNSYLVANIA LAW REVIEW centive to plead guilty is greatly reduced. The inevitable delays of the process, as well as those that are not so inevitable but can be induced by carelessness or bad faith, would then work in favor of the defendant rather than, as is the situation when he is in custody, against him. It is unlikely that there would be a significant rise in the percentage of defendants eventually found not guilty because these people are, by hypothesis, probably guilty. There would be some rise, partly attributable to the disappearance of witnesses through delays in bringing the case to trial, partly attributable to the fact that some de- fendants whom we know are guilty will be exculpated through human error-by judges, jurors, or prosecutors. But the danger is only secondarily that some few defendants will be exculpated who otherwise would not. The main risk is that the increased consumption of time required to litigate cases that do not really need to be litigated would put an intolerable strain on what is already an overburdened process. That consideration alone argues against a policy that makes pretrial liberty the norm. These arguments against automatic pretrial liberty are not neces- sarily arguments in favor of the present bail system, under which there is a nominal right to pretrial liberty, which is in practice not a right at all because of the power of the committing magistrate to set bail in an amount which the defendant is unlikely to be able to afford. However, that system works well enough in practice, even if its supposed premise is at odds with the postulates underlying this model. Preferably the courts, with the expert help of the police and the prosecutors, should select those people who, for whatever reason, ought not to be at liberty pending formal adjudication of guilt, and see to it that they are not.

There are, it is true, injustices in the bail system that are not required by the demands of the Crime Control Model. There is no reason, for example, why defendants who are ultimately convicted and sentenced to prison terms should not have time spent in pretrial custody credited against their postconviction prison terms. And there may be many instances in which police efficiency would be pro- moted by not cluttering station houses and detention centers with minor offenders. For these, the use of summons instead of arrest or release after arrest without the posting of bail may be desirable in the interest of cutting down the use of police for convoy duty through the pretrial pipeline. However, if pretrial detention is to be mitigated or avoided for some people, it ought to be done explicitly for the purpose of promoting the efficiency of the crime control process rather than because of any abstract "right" to pretrial liberty. In cases of serious [Vol.l13:1 TWO MODELS OF THE CRIMINAL PROCESS crime, at the least, the confinement of the defendant before adjudication of guilt serves the ends of the process and should be regarded as the norm. If the present system of requiring bail for some reason or other stopped producing a high rate of pretrial confinement, it would have to be replaced by one that did. The Due Process Model. A person accused of crime is not a criminal. The sharpest distinction must be observed between the status of the defendant and of the person who has been duly convicted of committing a crime. Perhaps the most important and certainly the most obvious operational distinction between the two lies in the issue of physical restraint. Pending the formal adjudication of guilt by the only authority with the institutional competence to decree it- a court-the status of the accused cannot be assimilated to that of the convicted in this most important respect. There are practical as well as ideological reasons why this should be so. An accused who is confined pending trial is greatly impeded in the preparation of his defense. He needs to be able to confer on a free and unrestricted basis with his attorney, something notoriously hard to do in custody. He may be in the best position to interview and track down witnesses in his own behalf, which he cannot do if in jail. His earning capacity is cut off; he may lose his job; his family may suffer extreme economic hardship-all in advance of any determination of guilt. Further- more, the economic and other deprivations sustained as a result of pretrial confinement all act as coercive measures that inhibit the accused's will to resist. He is rendered more likely to waive the various safeguards to unjust conviction which the system provides by agreeing to plead guilty. When this happens on a large scale, the adversary system as a whole suffers because its vitality depends on effective challenge. A person accused of crime is entitled to freedom except to the extent necessary to serve the legitimate ends of a legal system. The only legitimate end that is threatened by an absolute right to be free pending trial is the assurance that a defendant will not subvert the orderly processes of criminal justice by deliberately absenting himself at the time and place appointed for trial. If persons accused of crime could with impunity fail to appear, the premise of cooperation on which a system of pretrial liberty depends could not in practice be realized. Hence, it is important that the right to pretrial liberty be exercised in a way that does not jeopardize the process as a whole. The historic way in which the right to pretrial liberty has been vindicated is through the institution of bail. It has been thought that the requirement of a financial deterrent to flight would adequately pro- 42 UNIVERSITY OF PENNSYLVANIA LAW REVIEW tect the viability of the system while ensuring that the defendant could in fact enjoy liberty before trial. This has been manifested through a constitutional guarantee of an absolute right to bail in noncapital cases, expressed in the form that "excessive bail shall not be required." ' But the constitutional guarantee should properly be understood not merely as a guarantee that "reasonable" bail will be set, but rather as a guarantee that the defendant will be released pending trial on the basis of bail or whatever other device or combination of devices will insure his presence at trial without defeating his right to be free on grounds that have nothing to do with the assurance of his presence. Bail is simply an instance, not the exclusive means, of assuring presence at trial. If the institution of bail does not adequately promote the de- sired combination of goals, then the spirit of the constitutional guar- antee requires that alternative means be developed. Such alternatives include nonfinancial deterrents to flight, such as criminal penalties for nonappearance, the use of summons rather than arrest (with its at- tendant physical custody) as an initial process in criminal cases, re- lease of arrested defendants on their own recognizance or in the custody of some responsible person, use of cash bail rather than bail bonds, and the like.

Where bail is used, it must be set with regard to the circumstances of the individual case rather than on a mechanical basis. Thus, the nature of the offense charged is only one of several elements to be taken into account in making the bail decision, and if bail is set mechanically on the basis of a schedule for certain offenses, that may in itself be an effective denial of the defendant's right. Essentially, bail-setting must be a factfinding process in which the financial re- sources of the defendant, his roots in the community, the nature and circumstances of the offense charged, and other relevant factors are all taken into account in arriving at an individualized decision as to the minimum level of bail required to assure a reasonable probability of his appearance for trial. It is, of course, completely inadmissible to set bail at a figure that the defendant is thought to be unable to meet. Speedy appellate review must be available to correct errors of that sort, which is yet another reason why the bail decision must initially be made on the basis of a record that others can subsequently appraise. To the extent that adequate investigative and other fact- finding resources are not brought to bear, the defendant should be entitled to go free on nominal bail or no bail. The period of custody should in no event exceed the minimum required after arrest to ascertain the relevant facts about the suspect's situation. Normally, 47 U.S. CONST. amend. VIII. [Vol.l13:1 TWO MODELS OF THE CRIMINAL PROCESS that should be done by the time the committing magistrate has made the decision to hold the arrested person for subsequent proceedings. There remains, however, a large class of persons for whom any bail at all is "excessive bail." They are the people loosely referred to as "indigents." Studies of the operation of the bail system have demonstrated that even at the very lowest levels of bail-say five hundred dollars-where the bail bond premium may be only twenty- five or fifty dollars, there is a very substantial percentage of persons who do not succeed in making bail and are therefore held in custody pending trial." It may be that the decision not to seek bail in many of these cases is a voluntary one: a man who knows that he is factually guilty may simply decide that it isn't worth his while to spend money on a bail bond premium. However, many people who are eventually adjudged guilty do post bond and are released pending trial. Their consciousness of guilt may be just as great as the poor man's, but they avail themselves of their right to be free pending a formal adjudication of guilt. It discriminates unfairly against the poor to deny them the same right simply because for them the marginal utility of the bail money is higher than it is for the rich. At any rate, it is clear that if all persons in custody were informed of their right to be free on the basis of nonfinancial conditions if they so elect, many of those who presently spend days, weeks, or months in custody awaiting trial would avail themselves of these devices. It seems to follow that a system that conditions pretrial freedom on financial ability is dis- criminatory. Indeed, given the malfunctioning of the present system, viewed from the standpoint of the financially disadvantaged, it may well be that the bail system should be ruled out for rich and poor alike. One need not pursue the argument to that extreme, however, to recognize that a system that conditions pretrial release exclusively or even predominantly on the provision of financial assurances of presence at trial is a seriously defective one. Other asserted bases for pretrial detention are either entirely with- out merit or present special problems that need to be handled on a more discriminating basis than a general rule permitting detention before guilt-determination. It is antithetical to our conceptions of justice to permit pretrial detention to be used as a kind of informal punishment in advance of (or instead of) a formal determination of guilt and sentence. To speak of the possibility that the accused may commit further crimes if left at large begs the question, since it has not yet been determined that he has committed any crime at all. Many of the limitations on substantive criminal enactments safeguard us 48 E.g., ATr'Y GEN. RE'. 67, 135 Table IV. See generally FREED & WAn, BAI. IN THE UxrrED STATES: 1964, at 9-21 (1964). 44 UNIVERSITY OF PENNSYLVANIA LAW REVIEW against being punished for a mere propensity to commit crime. The logic of preventive detention would extend to persons newly released from prison; why not re-arrest them and lock them up because they may commit another crime? The problem of what to do with "dangerous" people who have not been convicted of committing crimes is a troublesome one. It far transcends the preventive detention of persons accused of crime. The solution, if there is one, must include setting standards for deter- mining who is "dangerous" and providing the minimal procedural due process safeguards of notice and a hearing for persons sought to be confined on this ground. Whatever the solution, it cannot bypass these basic due process requirements by permitting the indiscriminate preventive detention of people who are accused of crime. The problem can, in any event, be minimized by shortening the interval between charge and trial. In some cases there may be a possibility that the defendant if left at large will threaten witnesses, destroy evidence, or otherwise impede the preparation of the case against him. This is said to be par- ticularly likely in the case of organized criminals. The argument is a little difficult to understand. The higher the degree of organization involved, the less likely it would seem to be that the personal attention of the defendant would be required to promote obstructive tactics. To the extent that there is a threat of this kind, it can be dealt with in other ways: by giving witnesses police protection, by placing the accused under injunction backed by the contempt power, by providing criminal penalties for tampering with witnesses, and the like. The vice of detaining a defendant before he actually does anything bad is obvious: it penalizes him for a mere disposition, something totally un- provable, thus opening the way for the most widespread abuses. At the first concrete sign that the accused has engaged in obstructive activities, it is altogether proper to seek to confine him on the basis of proof that obstructive activities have taken place. But there is all the difference between so doing on the basis of proof after the fact and on the basis of suspicion before the fact. In summary, then, pretrial liberty should be the norm; the only exception that the criminal process as such should recognize ought to be the assurance of the accused's presence at trial; assuring his presence ought to be accomplished by measures other than detention; and de- tention should never be resorted to merely because the accused is unable to provide financial deterrents to nonappearance. The Situation and the Trend. The legal norm embraced in the eighth amendment which forbids the requirement of excessive bail [Vo1.113:1 TWO MODELS OF THE CRIMINAL PROCESS is construed to confer an absolute right to pretrial release on reasonable bail in noncapital cases. 49 Bail in turn is required to be set solely with respect to assuring defendant's presence at trial; other con- siderations are inadmissible. The federal standards are echoed by those prevailing in the states. While it has not been authoritatively so held, it appears that the federal constitutional provision on bail would be held applicable to the states." There is at the present time very little appellate control over bail-setting. The court of first in- stance has wide discretion; bail is ordinarily set pretty much on the basis of the offense charged; bail reductions are usually ordered only when the bail set in a particular case is thought to be out of line for the particular offense charged. There is no general right to be free if one is unable to provide bail; the accused's remedy is said to be to move for prompt trial. 5' In practice there is very little judicial control over bail practices. Bail is set mechanically." There is almost never any investigation of the circumstances of a particular accused or the likelihood that he will appear for trial. Bail or jail therefore becomes a question answered solely on the basis of a defendant's financial resources and his ability to obtain a professional bondsman to post bail for him. On the whole it is the alleged gangster or hardened criminal who is freed on bail and the first or sporadic offender who stays in jail. Furthermore, the system in practice permits and fosters the setting of bail in amounts that ensure that defendants will remain in jail; there is no question but that bail is widely used for purposes that are supposedly denied it by the legal norm, because of the unavailability of prompt remedial measures, the discretionary nature of the bail decision, and the lack of assistance of counsel in calling attention to infractions of the legal norms. It is a notable fact that the unbailed defendant is also to a large extent the unrepresented defendant. 3 The combination of factors alluded to makes the constitutional guarantee to a large extent nugatory in practice. If the legal norm is thought of as conforming in most respects to the Due Process Model, it is evident that the practical reality indicates a situation much closer to the Crime Control Model. There exists today a widespread, vocal, and increasingly in- fluential dissatisfaction with the operation of a system that places prime reliance on the use of financial devices for assuring presence 49 See Stack v. Boyle, 342 U.S. 1 (1951). 50 See In re Shuttlesvorth, 369 U.S. 35 (1962). 61 See United States v. Rumrich, 180 F.2d 575 (2d Cir. 1950). b2 FREE & WAID, op. cit. supra note 48, at 18-21. 53ATT'Y GEN. REP. 70-72. 46 UNIVERSITY OF PENNSYLVANIA LAW REVIEW at trial. 54 Unlike most of the other model trends we are considering, this one is not primarily judicial in character. There have been no path-setting decisions creating new norms, as in the areas of investi- gatory practices or right to counsel. Rather, the trend has been so far manifested by the work of governmental and extragovernmental groups, who have revealed in detail the present state of affairs and framed new norms responsive to the needs believed to emerge from factual revelations. Experiments are now underway designed to demonstrate the efficacy of other means of assuring presence at trial: improved factfinding mechanisms to determine whether an individual accused is a good risk for release without financial conditions; the use of cash deposits in an amount equal to what would otherwise be the bail bond premium, in order to reduce reliance on the professional bondsman; initiating criminal prosecutions by summons rather than arrest in minor cases, among others. A bill has been introduced in the Senate, supported by all members of the Subcommittee on Con- stitutional Rights of the Committee on the Judiciary, that would pre- clude the pretrial detention of an accused solely on the ground that he is financially unable to make bail. 5 The special problem of bail in civil rights cases is receiving considerable attention and may provoke judicial trends that will promote greater conformity with the Due Process Model generally. Where all this ferment will lead is difficult to say, but undeniably the ferment is there. It seems safe to predict that in the foreseeable future legislation and court rulings will advance rather than retard the trend toward conformity with the Due Process Model. There will be an indeterminate but significant increase in the percentage of criminal defendants who are at liberty pending trial, accompanied by an indeterminate but significant increase in the number of criminal defendants who do not plead guilty to the initial charge against them, but who either succeed in obtaining a more advantageous outcome through plea bargaining or who elect to put the prosecution to its proof in a trial. 2. The Plea of Guilty The vast majority of criminal prosecutions terminate with the entry of a plea of guilty. The plea rather than the trial is the dominant 50 This dissatisfaction was epitomized and brought to a focus in the National Conference on Bail and Criminal Justice held in Washington, D.C., on May 27-29, 1964, under the cosponsorship of the United States Department of Justice and the Vera Foundation, a private organization concerned with remedying deficiencies in the present bail system. FREED & WALD, BAIL IN THE UNITED STATES: 1964 (1964), was prepared as a working paper for this conference and contains extensive docu- mentation of the present situation and current trends, as well as a valuable bibliography on the bail problem. 55 S. 2838, 88th Cong., 2d Sess. (1964). [Vol.l13:1 TWO MODELS OF THE CRIMINAL PROCESS mode of guilt-determination. It is widely asserted that any sig- nificant increase in the number of criminal prosecutions going to trial would result in a breakdown of the criminal justice system. The institution of the guilty plea is itself affected by factors operating at earlier levels of the criminal process, notably by the availability vel non of pretrial liberty and of the assistance of counsel. It seems clear both as a matter of logical inference and of demonstrable fact that a de- fendant who is out on bail and who enjoys the services of a lawyer is less likely to plead guilty than is one who lacks one or both of these advantages. It is of course possible that in many cases the advantages referred to are banked in the form of an advantageous plea-bargain rather than of insistence on a trial on the merits. Nonetheless, it appears likely that there would be a substantial shift in the proportion of cases going to trial if factors at earlier stages of the process operated with uniform frequency in favor of the defendant. Thus, if there were a reduction of twenty-five percent in the number of cases in federal courts disposed of on guilty pleas, there would be roughly twice the number of trials that are now held. 6 What do our two models tell us about the guilty plea? From the host of relevant aspects of this institution we will briefly examine these: Under what circumstances, if any, should a plea of guilty be set aside as "involuntary"? To what extent should a defendant who pleads guilty have the assistance of counsel? What obligation, if any, does the judge who receives the plea have to satisfy himself as to the factual and legal guilt of the accused? The Crime Control Model. The purpose of the arraignment, at which the defendant is required to plead to the charge against him, is to dispose of as large a proportion of cases as possible without trial. It is in the interest of all-the prosecutor, the judge, and the de- fendant-to terminate without trial every case in which there is no genuine doubt as to the guilt of the defendant. If the earlier stages of the process have functioned as they should, only a very small pro- portion of cases should at this point remain in that category. There is also an irreducible minimum of cases where so much is at stake- either because of the gravity of the offense or the position of the defendant-that there is no reasonable possibility of compromise. The murderer and the bank president charged with income tax evasion have GApproximately 807o of criminal defendants disposed of in the federal courts from 1956 through 1962 entered pleas of guilty or nolo contendere. See Note, Guilty Plea Bargaining: Compromises by Prosecutors To Secure Guilty Pleas, 112 U. PA. L. REv. 865 & n.4 (1964). Assuming that the rest stood trial and that all those included in the hypothesized 25% reduction also were to stand trial, the present 20% who stand trial would grow to 40%. 48 UNIVERSITY OF PENNSYLVANIA LAW REVIEW this in common: they have little to lose by going down fighting. Aside from these two categories of case where the will to litigate is strong, all criminal cases ought normally to be terminated by plea. If this general criterion is accepted, the details follow without much trouble. There is a distinct social advantage to terminating criminal pro- cedings without trial whenever the defendant is willing to do so. Any number of subtle interacting factors may make a defendant willing; it would be an endless operation and essentially a self-defeating one, in terms of the objectives of the process, to inquire into the precise nature of these factors in any large number of cases. For example, the judge in accepting a plea of guilty may of course inquire whether any promises have been made to the defendant, since a promise of leniency by the police or the prosecutor is one which cannot be delivered and is evidence of poor prosecutorial performance. But if the judge does discover that an improper promise has been made, the proper course is not to reject the plea but rather to set the defendant right about the legal situation and then permit him to enter the plea if he so wants. In the overwhelming majority of cases, then, the function of an inquiry by the judge is to provide an assurance of regularity on the record, not to protect any special right of the defendant. It is also perfectly proper for the judge to make it clear that a defendant who pleads guilty can expect greater leniency in sentencing than one who insists on putting the state to the time and expense of a trial. The general run of criminal defendants are capable of making up their own minds as to whether they want to plead guilty. If a de- fendant has a lawyer and wants to consult him about the guilty plea, that is proper. But the state should be under no obligation to provide counsel for a defendant at arraignment. All that is required for a plea of guilty is that the defendant understand its nature and conse- quences in a general kind of way, and that he enter it of his own free will. The judge's duty is to ensure that these conditions are met. It would involve a needless duplication of resources to insist that a defense lawyer as well as a judge must participate in the entry of a guilty plea. The judge need not inquire into the factual circumstances under- lying the commission of the offense except to the extent that he deems it desirable in helping him to perform his sentencing function. Cases do not reach this stage of the criminal process unless there is sub- stantial evidence of guilt. Any requirement that the judge inquire into the issue of guilt before accepting a plea would impair the effi- ciency of the process and undermine the purpose of the plea of guilty by converting the arraignment into an abbreviated trial on the merits.

[Vol.l13 :I TWO MODELS OF THE CRIMINAL PROCESS A fortiori, there should be no inquiry into the availability of defenses that do not go to the issue of factual guilt. It is cause for congratula- tion, not alarm, when a defendant who is factually guilty is convicted and sentenced despite the existence of possible defenses that have nothing to do with the merits of the case. One of the great strengths of the guilty plea is that it serves to bypass issues that can only result in a weakening of effective criminal justice. If a defendant is con- scious of his own guilt and willing to accept his punishment, it does neither him nor the community any service to inquire into possible errors made at earlier stages of the process that might serve to enable him to escape his just deserts. The Due Process Model. The arraignment is the fulcrum of the entire criminal process. It is at this point that one of two things happens: either the possible errors and abuses at the earlier, largely unscrutinized stages of the process are exposed to judicial scrutiny, or they are forever submerged in a plea of guilty. The guilty plea is not only a device for expediting the handling of criminal cases; it is a kind of Iron Curtain that cuts off, almost always irrevocably, any disinterested scrutiny of the earlier stages of the process. Guilty pleas should therefore be disfavored. There may indeed be a serious ques- tion whether they should ever be permitted at all; but it is clear that they should be hedged about with safeguards designed both to cut down their incidence and to prevent them from being used in cases where possibly meritorious challenges to the process exist. No kinds of pressure, either by the prosecutor or the judge, should be brought to bear on a defendant to induce him to plead guilty. Plea-bargaining by a defendant who is adequately advised by counsel may, under careful supervision by the judge, be an acceptable way of terminating a criminal case. But the prosecutor, in order to avoid any possibility of coercive pressure, should never take the lead in proposing or suggesting a compromise plea. It is manifestly improper for a judge to use his sentencing discretion to coerce a guilty plea, either by threatening severe punishment in a particular case or by reserving lenient treatment, as for example probation, for defendants who plead guilty. A criminal defendant is entitled to have the charges against him tried in the manner prescribed by law, no matter how overwhelming the evidence of guilt may be thought to be. A criminal trial is not to be viewed as an undesirable burden, but rather as the logical and proper culmination of the process. It follows that it de- feats the ends of the system to penalize a defendant for insisting on a trial or to intimidate him by threatening him with unpleasant conse- quences if he does insist. 1964] 50 UNIVERSITY OF PENNSYLVANIA LAW REVIEW No one should be permitted to plead guilty without the assistance of counsel. If a criminal defendant cannot receive a fair trial without counsel, how much less likely is he to have enjoyed fair process if he has to resolve the highly technical and complex strategic problems involved in a guilty plea without expert assistance? It is doubtful whether waiver of counsel should ever be allowed at this stage. As a practical matter, there is unlikely to be such a waiver if the judge, on hearing that a defendant wishes to plead guilty, informs him that he is reluctant to accept a plea at this time, explains the advantages of con- sultation with counsel, and offers to appoint a lawyer immediately. Even if these restrictions are faithfully observed, it is probable that a high proportion of criminal defendants will choose to plead guilty. The question then arises what guarantees of procedural regu- larity the judge should endeavor to provide before, in effect, closing the door to further scrutiny by accepting the plea. He ought, in the first place, to require the prosecutor to summarize the evidence against the defendant, indicating what the testimony will be and by whom it will be provided. He should satisfy himself that there is probably sufficient evidence to sustain a conviction on the charge or charges against the defendant. He should also satisfy himself as to the admissibility under applicable rules of evidence of the testimony pro- posed to be elicited. Beyond satisfying himself about the evidence, the judge should also take this occasion to probe, with the assistance of the prosecutor and the defense counsel, the possible existence of abuses at earlier stages of the process, such as illegally obtained evidence, improper confessions, failure to provide counsel at an appropriate stage, length of pretrial detention, and the like, in order to determine their possible bearing on the adverse termination of the proceeding by acceptance of a plea of guilty. Only after he is satisfied that the record is clear in these two general respects-the establishment of guilt and the absence of abusive practices at earlier stages of the process-should the judge accept a plea of guilty. To the extent that these protective measures are not employed, the defendant should be entitled at a later time to move to set aside the plea of guilty and to stand trial.

The Situation and the Trend. Although reliable data are hard to come by, it is highly probable that any general view of the guilty plea in this country at the present time would disclose practices that conform far more closely to the Crime Control than to the Due Process Model. Pressures from earlier phases of the process particularly on 57 There is an excellent exposition of the present situation as well as a detailed examination of judicial decisions contributing to the trend herein described in Note, 112 U. PA. L. Rxv. 865 (1964). [Vol.l13 :I TWO MODELS OF THE CRIMINAL PROCESS those defendants who are not at liberty after arrest and who do not have the assistance of counsel, plus pressures exerted at or about the time of arraignment, tend strongly to militate toward the entry of pleas of guilty. The assistance of counsel, to the extent that it is available at arraignment, is perfunctory in the majority of cases. Waiver is easily accomplished and widespread. And the role of the judge is a relatively passive one, with no generally effective pattern of inquiry into factual guilt or into the possibility of abuses at earlier stages of the process. There are some signs, however, that the plea of guilty may be receiving increasingly close scrutiny. If the prosecutor enters into a plea arrangement that depends on promises he is unable to fulfill, some courts have taken the view that the resulting plea should be set aside as "involuntary." "I There also seems to be a trend develop- ing toward stricter standards for appraising the defendant's under- standing of the nature and consequences of the plea. The full impact of Gideon v. Wainwright is yet to be determined, but it would appear that pleas of guilty entered without the assistance of counsel may prove vulnerable, quite apart from the probability that implementation of Gideon will insure the participation of counsel in a higher propor- tion of guilty pleas than has hitherto been the case. It seems unlikely that the practice of guilty plea bargaining will itself come under attack. Rather, the trend seems to be toward regulating and equalizing the conditions under which the bargaining takes place. Greater insistence upon the participation of defense coun- sel and a more active role for the trial judge will probably char- acterize the development of plea bargaining. To the extent that this turns out to be an accurate forecast, the institution of the guilty plea will hold its place as the fulcrum of the criminal process only at the cost of a greater input of resources and, therefore, a diminution in "efficiency" as measured by the dictates of the Crime Control Model. C. Review of Errors in the Guilt-Determination Process Although it is probably common ground today that any criminal process ought to provide some opportunity for correcting some errors that have occurred at earlier stages in the process, it need be kept in mind that the institution of appellate review in criminal cases is less than one hundred years old. If it would indeed "go against the grain, today, to make a matter as sensitive as a criminal conviction subject to unchecked determination by a single institution," '9 that fact is a 58 See Machibroda v. United States, 368 U.S. 487 (1962) ; Dillon v. United States, 307 F.2d 445, 449-50 (9th Cir. 1962) (dictum). 59 Bator, supra note 22, at 453. 1964] 52 UNIVERSITY OF PENNSYLVANIA LAW REVIEW striking testimonial to the dynamic character of the criminal process and to the consequent fluidity of what may usefully be regarded as the range of practicable alternatives. Even if there is general agreement that in some sense appellate review is a necessary feature of the process, the terms on which review should be available divide the two models: What kinds of issues should be reviewable, "legal" issues only or "factual" determinations as well? What financial barriers to review, if any, should be allowed? Should review be automatically available, or should some screening devices be used to prevent frivolous appeals from being taken? If errors are found, what standards should deter- mine the outcome of review-must the defendant show that the out- come would probably have been different but for the error, or should some errors count as conclusive, and if so, which ones? What is the permissible timing of review, i.e., to what extent is review to be limited to a continuation of the original process-"appeal"-or per- mitted also to take place in a fresh proceeding-"collateral attack"? It will be convenient to divide the subject of review in accordance with the distinction just made between appeal and collateral attack, since the distinction has become thoroughly ingrained in our thinking about the criminal process. There is another reason for observing it: the opportunity that it affords to depart from the assumption that we are dealing with a wholly unitary system of criminal justice and to examine in one hotly contested instance the impact of federalism on the shape of the criminal process. For this purpose we will make the artificial and somewhat misleading assumption that review by appeal concerns review within a unitary system while review by collateral attack concerns federal review of state criminal processes, for the limit- ing purpose of selecting a few representative problems whose solutions divide our two models. 1. Appeal The general role of an appeal system is, of course, strongly con- ditioned by assumptions about what has occurred in previous stages of the process. The Crime Control Model, as we have seen, places very heavy emphasis upon the plea of guilty as the central guilt- determining device. The comparatively few cases that it confides to a more formal adjudicative process are those in which there is thought to be some doubt about the factual guilt of the accused. Those doubts are supposed to be resolved by the trier of fact. Accordingly, the role of an appellate review system is highly marginal: it is available to correct those occasional slips in which the trier of fact either makes a plain error about factual guilt or makes some kind of procedural [Vol.l13:1 TWO MODELS OF THE CRIMINAL PROCESS mistake so gross as to cause with some high degree of probability a substantial diminution in the reliability of the guilt-determining process. In contrast the appellate stage in the Due Process Model is seen as having a much broader function. It operates, true, to correct errors in the assessment of factual guilt (at least when they have redounded to the detriment of the accused), but that is only the beginning of its function. It serves, more importantly, as the forum in which infringe- ments on the rights of the accused, as laid down in the model, that have accumulated at earlier stages of the process, are to be redressed and their repetition in later cases deterred. The appellate forum, seen as having distance from and independence of the police-prosecutor nexus into which the trial court is so often drawn, is both guardian and vindicator of the Due Process Model. While the appellate stage is seen in the Crime Control Model as being a remote and marginal appendage of the process as a whole, it is perceived in the Due Process Model as being qualitatively crucial and quantitatively significant. Im- portant differential consequences follow for the resolution in the two models of problems of access and scope. The Crime Control Model. Once a determination of guilt has been made, either by entry of a plea or by adjudication, the paramount objective of the criminal process should be to carry out the sentence of the court as speedily as possible. We must be able to say that people who violate the law will be swiftly and certainly subjected to punishment. This end will be undermined if the process permits, and hence invites, delays in the execution of sentences. Finality of guilt determination is therefore the most important point of departure for evaluating any system of review. To put the matter bluntly, appeals should be so effectively discouraged that the mere taking of an appeal should be in itself a fairly reliable indicator that the case contains sub- stantial possibility of error going to the issue of factual guilt. If appeal in criminal cases is available as a matter of right, restric- tions must be imposed to ensure that the right is exercised responsibly. Specifically, the costs of an appeal, filing fees, printing costs for the record and briefs, and most importantly, counsel fees, should not be waived or publicly defrayed unless the appeal is screened and deter- mined to be probably meritorious. This screening power should be lodged in the court that has made the determination of guilt, since it is the tribunal most likely to be familiar with the case and therefore most likely to be able to make an expeditious determination of probable merit. A decision not to permit an appeal by a person unable to finance his own appeal should be conclusive, subject only to review for gross abuse of discretion. While in theory the same procedure should prob- 1964] 54 UNIVERSITY OP PENNSYLVANIA LAW REVIEW ably apply to all appeals, such a limitation may be sufficient, since in practice the financial burden of an appeal not defrayed by the state may adequately impede frivolity. Bail pending an appeal should be allow- able only as a matter of grace and should be withheld where the issues raised on appeal do not appear to be substantial. No issue should be raisable on appeal that was not raised at an earlier stage of the process. No conviction should be reversed for the insufficiency of evidence unless the appellate tribunal finds that no reasonable trier of fact could have convicted on the evidence presented. Appeals against a verdict of acquittal should be available to the prose- cution to the same extent as appeals against a conviction are available to the defense. Errors not going to the sufficiency of the evidence to establish factual guilt-errors in the admission or exclusion of evidence, in the trial court's instructions, or in the conduct of the prosecutor or the trial judge-should not provide a basis for reversal of a convic- tion on appeal unless it is found that in the absence of the error or errors the result would probably have been different. Further, no errors should suffice for reversal if the appellate court concludes on a review of all the evidence that the factual guilt of the accused was adequately established. The Due Process Model. The initial forum in which abuses of official power should be corrected in the criminal process is the trial. However, they are not always corrected there, and, indeed, the trial process may itself be a fertile source of additional abuses. If they are not corrected, they do not come to public attention since the trial process is in the great run of cases only slightly more visible than the police and prosecutorial processes that precede it. The right of appeal is an important safeguard for the rights of the individual accused. Beyond that, it plays an essential role in the lawmaking process; it is only the existence of a steady flow of criminal cases on the appellate level that provides the raw material for the elaboration of those very rights. If the Due Process Model is to retain its dynamic character, there must be full and unrestricted access to the appellate phase of the process. There should be no limitations on the convicted defendant's right to appeal. Financial restrictions are as much out of place here as they are at other levels of the process. If the appellant cannot afford to pay a filing fee, it must be waived; if he cannot afford to buy a transcript, it must be given to him; if he cannot afford to hire a lawyer, he must be assigned one. This last point is especially important; whether re- versible errors justifying an appeal have occurred is uniquely a matter on which the convicted defendant needs the help of a lawyer; there is [Vo1.113 :I TWO MODELS OF THE CRIMINAL PROCESS no more technical aspect to the criminal process. No lawyer will advise an appeal where grounds for appeal are lacking, but only a lawyer can tell whether the grounds exist or not; for at this stage of the process, it is legal errors rather than factual guilt that are primarily at issue. And, while bail pending appeal raises different and more restrictive issues than does the question of liberty pending trial, it is important that the discretion to allow bail pending appeal not be manipulated coercively to discourage the pursuit of any appeal that has a semblance of merit. Where discretionary judgments of this kind are inevitable, they ought to be lodged with the appellate court rather than with the trial court. While it should not be possible to sit by silently and allow errors to go unchallenged at the trial level, appellants should not be held rigidly to a requirement that the errors of which they complain must have been challenged below. The appellate court should be entitled to notice any plain error prejudicial to the rights of an accused. No unitary standard for determining reversible error can be advanced; even cumu- lative and repetitive errors of an insubstantial kind should suffice for reversal. Of course, any error abridging basic rights of the defendant -rights to be free of illegal searches and seizures, not to be coerced into confessing, and not to be forced to incriminate himself-should be ground for reversal irrespective of the strength of the rest of the case. To say that is simply to repeat that the process itself must afford reme- dies for its abuse and deterrents against the misuse of official power. The appellate process should afford similar sanctions against abuses that occur for the first time at the trial level, such as prosecutorial mis- conduct, prejudicial publicity, and ineffective counsel. The reversal of a criminal conviction is a small price to pay for an affirmation of proper values and a deterrent example of what happens when those values are slighted. When an appellate court finds it necessary to castigate the conduct of the police, the prosecutor, or the trial court, but fails to reverse a conviction, it simply breeds disrespect for the very standards it is trying to affirm. The Situation and the Trend. Appeals are taken in only a small proportion of criminal cases. The appeal is the apex of the pyramid and represents a final screening from a group of cases whose number has been sharply reduced through previous screenings. Nonetheless, the appeal is important out of all proportion to numbers because it is the level of the criminal process at which the governing legal norms are made explicit. There appears to be a fairly constant relationship between the number of appeals taken in criminal cases and the propor- tion of convictions that are reversed. Unsuccessful appeals are by and 19641 56 UNIVERSITY OF PENNSYLVANIA LAW REVIEW large not very significant legally; it is the successful appeal, in which reasons of more or less general applicability are given for reversing a conviction, that is significant for the laying down of operative norms. Consequently, the larger the absolute number of appeals taken, the greater the number of reversals there probably tends to be, and the more complex, precise, and thickly textured are the inhibitory norms that tend to be established. In a very real sense, therefore, the ques- tion of access to the appellate process is strategically crucial to the struggle between the two models. The fewer appeals there are, the likelier it is that Crime Control norms will prevail; the more appeals there are, the likelier it is that Due Process norms will prevail. There is no level of the criminal process at which the triumph of the Due Process Model, at least in terms of asserted norms, has been more speedily and more completely established than on the ques- tion of access to the appellate process. Only a decade ago there were significant differences in legal norm between the treatment of appeals by those who could afford to finance them and those who could not. The factual disparities doubtless remain; but the legal norms have been drastically pushed toward the Due Process end of the spectrum. In a series of pathbreaking decisions, the Supreme Court has estab- lished, on the state and federal levels alike, that the situation of finan- cially incapable persons must be substantially improved. If a transcript of the trial proceedings is necessary for appellate purposes, the state must supply one to persons who cannot buy their own. 6" The screening of appeals as a prerequisite to relief from financial barriers has been sharply eased, perhaps virtually eliminated. 6 And the right to counsel on appeal has been assured. 6" Here as elsewhere, of course, the develop- ment of Due Process norms has preceded their translation into opera- tive fact, and the process of providing the resources necessary to make the developing norms generally operative has barely begun. 3 On issues of scope and standards of review, the trend seems un- mistakeably to be toward the norms posited in the Due Process Model. Errors that are denominated "constitutional" are more and more being viewed as grounds for automatic reversal without regard to the suf- ficiency of evidence of factual guilt. And even nonconstitutional errors are being treated as calling for reversal in the teeth of what have pre- viously been regarded as well-established rules requiring an assessment 60 Griffin v. Illinois, 351 U.S. 12 (1956). 61 See Draper v. Washington, 372 U.S. 487 (1963) ; Coppedge v. United States, 369 U.S. 438 (1962). 02 See Douglas v. California, 372 U.S. 353 (1963). 63See Ar'y GEN. REP. 90-115. [Vol.l13:1 TWO MODELS OF THE CRIMINAL PROCESS of the record as a whole. 4 It is not surprising that this should be so. The appellate process is the forum par excellence for assertion of the norms that make up the Due Process Model. What can be observed on the appellate level today is merely an affirmation of the increased emphasis on Due Process norms in the criminal process as a whole. 2. Collateral Attack Should the criminal process end with a determination of guilt that is subject to review on direct appeal? The question may arise whether or not an appeal is actually taken. It becomes particularly acute when the determination of guilt is made by a plea of guilty, since in that situation an appeal is as a practical matter highly unlikely. The issue is complicated by the coexistence of state and federal law and of state and federal forums for deciding questions that arise in criminal cases. Assume that a state's criminal process has resulted in a determina- tion of guilt, and the accused, now a state prisoner, asserts in a federal habeas corpus proceeding that he is being held in custody in violation of the Constitution and laws of the United States, alleging that rights established under the fourteenth amendment have been abridged at some stage of the state criminal process. It is obvious that one important dimension of the problem involves the delineation of what those fourteenth amendment rights are thought to be. We can say with rough accuracy that such tenets of the Due Process Model as have been translated into legally binding imperatives, insofar as they apply to state criminal prosecutions, have been grounded in the commands of the fourteenth amendment. The substantive ques- tions, then, on collateral attack, raise the whole range of dynamically evolving dictates of the Due Process Model that we have been exam- ining. The procedural issues to which we now turn are all aspects of a single basic problem: to what extent should federal collateral attack on state criminal convictions be permitted to invoke the expanding tenets of the Due Process Model to nullify state criminal convictions? Three representative issues will serve to illustrate the distance between the two models with respect to this crucial procedural problem: (1) If a federal fourteenth amendment claim has been asserted by the habeas corpus petitioner at any point in the state criminal process and has been considered and rejected on the merits by a state court, may the peti- tioner relitigate the issue in a federal habeas corpus proceeding? (2) If the federal habeas petitioner has failed to avail himself of an oppor- 64See, e.g., People v. Modesto, 59 Cal. 2d 722, 382 P.2d 33, 31 Cal. Rptr. 225 (1963). See generally WrririN, CALIFORNIA CRmIINAL PRocEauRE 733-43 (1963), for a documentation of the trend in one jurisdiction by a commentator generally unsympathetic to it. 19641 58 UNIVERSITY OF PENNSYLVANIA LAW REVIEW tunity to raise a fourteenth amendment issue in the state criminal process and is therefore barred by state procedural rules from now raising the issue, may he nonetheless secure a federal determination? (3) If the governing legal norms have been changed since a state prisoner was convicted so that contentions formerly rejected are now viewed as established rights, may he secure relief, i.e., are changes in the law "retroactive"? Affirmative answers to each of these questions tend to provide a broad scope for judicial vindication of the Due Process Model; negative answers tend to perpetuate the prevalence of the Crime Control Model, notwithstanding the fact that the norms now on the books tend toward the tenets of the Due Process Model. We need not spell out in detail the positions that the two models reach on these representative issues; they are implicit in what has been said earlier about the operations of the model at the appellate level. The Situation and the Trend. Until 1953 it was "at most doubt- ful" " whether federal constitutional claims raised and decided on the merits in state criminal cases could be re-examined by a federal court in a habeas corpus proceeding. The landmark decision in Brown v. Allen 66 held that they could. Since that time there has been an increasing number of petitions for habeas corpus filed by state prisoners. The Supreme Court has recently reaffirmed the doctrine of Brown v. Allen, and has laid down rules considerably more favorable to habeas petitioners than contemporary practice in the lower federal courts has provided for determining the circumstances under which habeas petitioners are entitled to have an evidentiary hearing on the under- lying merits of their fourteenth amendment claims." At the same time that the Court opened the door to determination of federal claims already heard in a state court, it appeared to close the door to a first determination in a federal court of a fourteenth amendment claim that state courts had refused to hear because of some state procedural default by the petitioner. 68 In subsequent years there was lively controversy over the circumstances in which a procedural default by a state prisoner should not be counted against him for habeas purposes. 69 The controversy has been at least temporarily resolved by a Supreme Court decision setting forth in the most sweep- 65 The phrase is Professor Bator's. Bator, supra note 22, at 463. 66 344 U.S. 443 (1953). 647 Townsend v. Sain, 372 U.S. 293 (1963). 68 See Brown v. Allen, 344 U.S. 443, 482 (1953). 69 Compare Hart, Foreword: The Time Chart of the Justices, the Supreme Court, 1958 Tern, 73 HARv. L. Rv. 84, 101-22 (1959), with Reitz, Federal Habeas Corpus: Im pact of an Abortive State Proceeding, 74 HARv. L. Rav. 1315 (1961).

[Vo1.113:1 TWO MODELS OF THE CRIMINAL PROCESS ing terms that a federal habeas court may, and in the ordinary case should, determine federal claims even though the petitioner has failed to avail himself of state remedies once, but no longer, open to himY So, for example, a state prisoner who contends that his conviction was based on a coerced confession but who failed to appeal from that conviction and was therefore barred from receiving a state determina- tion is no longer barred from seeking release on federal habeas. The combined effect of the doctrines of Brown v. Allen and Fay v. Noia gives the federal courts a broad supervisory power over the administration of state criminal justice. The amplitude of this spatial reach is matched in the temporal dimension by a doctrine, yet to be made explicit by the Supreme Court but apparently emerging from a number of its summary dispositions, that changes in fourteenth amendment doctrine are to be deemed retroactive. Thus, a state prisoner convicted on the basis of illegally obtained evidence before Mapp v. Ohio, 7' or without the provisions of counsel before Gideon v. Wainwright," or, presumably, on the basis of a confession secured during a period of interrogation without the aid of counsel before Escobedo v. Illinois, 73 or of comment on his failure to testify in his own behalf before Malloy v. Hogan 74 may now be entitled to release on habeas corpus. Of course, formidable obstacles stand in the way of any indi- vidual prisoner's success in pressing an application for habeas corpus. Working in most cases without the aid of counsel, he must convince a federal district judge, one of a notably unsentimental group of men, that there is arguable merit to his cause before he will even be given a hearing on his allegations. But with all the difficulties that collateral attack presents for the prisoner seeking to invoke it, it is undeniable that the remarkable broadening of its theoretical availability that has been taking place in recent years constitutes a powerful weapon for maintaining, capitalizing upon, and expanding the influence of the Due Process Model on the criminal processes of the state and the nation. D. Access to Counsel: A Postscript At every stage in the criminal process, as we have seen, our two models divide on the role to be played by counsel for the accused. In 70 Fay v. Noia, 372 U.S. 391 (1963). 71367 U.S. 643 (1961). 72372 U.S. 335 (1963). 73 378 U.S. 478 (1964). 74378 U.S. 1 (1964) (the fourteenth amendment makes the privilege against self-incrimination as embodied in the fifth amendment binding on the states). 60 UNIVERSITY OF PENNSYLVANIA LAW REVIEW the Crime Control Model, with its administrative and managerial bias, he is a mere luxury; at no stage is he indispensable, and only in the very small proportion of cases that go to trial and the even smaller proportion that are reviewed on appeal is he to be regarded as more than merely tolerable. The Due Process Model, with its adversary and judicial bias, makes counsel for the accused a crucial figure throughout the process; on his presence depends the viability of this Model's prescriptions. The decision in Gideon v. Wainwright, that the states must provide counsel for criminal defendants who are financially unable to provide their own, is therefore the longest single step so far taken by any institution of government in moving the norms of the criminal process toward the Due Process Model. Many issues posed by this development remain to be clarified. In what kinds of criminal prosecutions does the right to assigned counsel apply? In "serious" offenses only? If so, what are the criteria of "seriousness"? When does the right to counsel "begin" and "end"? Are the limits the same for assigned counsel as for privately retained counsel? Does the guarantee require the reversal of convictions ob- tained before the decision in Gideon? Looming behind these questions are even more portentous ones: Does the effective assistance of counsel require that the state provide financial compensation for the lawyers who serve? Must provision be made for other expenses of an effective defense? The emerging shape of the criminal process will be sub- stantially affected by the answers given to questions such as these. It can be asserted, without depreciating the importance of questions of this order, that Gideon will remain for a long time the watershed decision in the evolution of the criminal process. It may also be the most durable. 75 Even if the cycle of change turns out to be near its end, the norms of the process have been ineradicably changed, and in far more than the mere insistence that counsel must be provided. Gideon makes no sense except on the acceptance of premises, all the stronger for being unarticulated, about the adversary and judicial quality of the criminal process. As long as Gideon remains in the law, there is in the normative content of state criminal processes a core of meaning that rests on the Due Process Model and provides a base for expansion of Due Process norms to other aspects of the process. Yet Gideon and decisions like it do not alone determine the shape of the criminal process. The response of other institutions of 75 Gideon was decided by a unanimous Court, although there were four opinions and although two members, Justices Clark and Harlan, did not join in the opinion of the Court. [Vol.l13:1 TWO MODELS OF THE CRIMINAL PROCESS government is equally important in determining whether the necessary resources will be provided to make the norm something more than a ground for reversing a few convictions. The implementation of Gideon may provide a paradigm of the tension between forces of change and of inertia. No one can doubt that the norms of the crim- inal process have been moved rapidly and spectacularly down the spectrum toward the Due Process Model. But a parallel development in the real-world operation of the process remains for the future. No estimate of direction and velocity for the criminal process can be realistic that fails to appraise not only the normative revolution that has occurred, but the forces of change and inertia that will govern the extent to which that revolution becomes a reality. III. THE TREND AND ITS IMPACT: A TENTATIVE APPRAISAL The criminal process as it actually operates in the large majority of cases probably approximates fairly closely the dictates of the Crime Control Model. Such systematically-gathered evidence as we have, reinforced by the impressionistic "feel" for the situation that is widely current in our culture, suggests that the real-world criminal process tends to be far more administrative and managerial than it does adver- sary and judicial. Yet, as we have seen, the officially determined norms of the process are rapidly providing a view that looks more and more like what has been described in these pages as the Due Process Model. It would be unnecessarily repetitive at this point to recapitulate this development in detail. Its principal thrusts have been to "judicial- ize" each stage of the criminal process, to enhance the capacity of the accused to challenge the operation of the process (both at the time adverse action is taken or threatened and subsequently), and to equalize the capacity of all persons accused of crime to take advantage of the opportunities thus created. In theory at least, to revert to a figure suggested at an earlier point in this Article, the process is being turned from an assembly line into an obstacle course. That is by far the dominant normative trend. We must now try to appraise its durability. There are some fairly obvious negative factors. First, the trend as it has so far developed is based almost exclusively on judicial deci- sions. Indeed, it has been derived from the lead taken by one judicial institution, the Supreme Court of the United States. Changes in atti- tude toward the criminal process or changes in personnel on the Court (which may come to the same thing) can slow or reverse the trend in two ways. First and more obviously, decisions can be overruled. 1964] 62 UNIVERSITY OF PENNSYLVANIA LAW REVIEW Much of the development of the last decade has been accomplished by overruling earlier precedents; decisional instability has been a feature of its evolution, and there is no reason to suppose that it can work in only one direction. Some of the most crucial decisions have been the result of closely divided judgments. Minorities within the Court have made powerful appeals to the reason of another day. A second, subtler, and probably more serious threat to the continued strength- ening of the Due Process trend is that the Court, out of a diminished enthusiasm either for the principles involved or for the continued combat their vindication entails, might cease or slacken its scrutiny of the criminal process as it operates in both state and federal criminal courts. It is well to remember that the Court's jurisdiction in these matters is almost entirely discretionary and exemplary. Any percep- tible slackening in the pace or the tone of its oversight will quickly convey a message to the lower courts who are necessarily the first- line custodians of the process' norms. That is not to say that the process of change can only be maintained if the Court continues to set new norms. We may well be coming to the point at which tightening the very open-textured pronouncements of the past decade will be the main task of a Court that continues to be committed to promoting the goals of the Due Process Model. But that sort of consolidating effort, no less than the innovating effort that has been going on, requires constant attention. Whether the Court will be willing or able to sup- ply this steady and unspectacular kind of leadership remains to be seen. Instability of decision and slowing of pace aside, there is another major reason why the predominantly judicial character of the trend we have been examining is a potential source of weakness. With in- significant exceptions, the courts can intervene in the criminal process only to impose the sanction of nullity. That is powerful enough, espe- cially when applied conscientiously by courts of first instance, but the sanction of nullity has its limitations. A court cannot ordain, admin- ister, and finance an adequate system for providing the assistance of counsel to persons unable to finance their own. It can only refuse to validate criminal proceedings that are the product of inadequate sys- tems or of no system at all. And, by and large, it can only invalidate such proceedings as happen to be brought before it for scrutiny, typi- cally requiring at some point the initiative of counsel whose very absence gave rise to the original objection. If it is the strength of courts that they can only deal, as Professor Freund has said, at retail, it is their weakness too. The sanction of nullity applied on a retail basis may provide the goad for change; but it is not a sufficient instru- ment of change. [Vol.l13:1 TWO MODELS OF THE CRIMINAL PROCESS That brings us to a second and related cautionary note about the durability of the trend toward the Due Process Model. However dif- fused among governmental and extragovernmental agencies the ca- pacity for promoting change in the criminal process may be, few would be disposed to deny the centrality of legislative assistance. Yet the legislative response has been slow and grudging. A single instance will suffice. It has been over twenty-five years since the decision in Johnson v. Zerbst " that persons accused of crime in federal courts are entitled to have counsel appointed for them if they are unable to hire their own. Repeated attempts have been made to get the Congress to set up a system for appointing and compensating defense counsel in the various federal districts. Finally, after a President explicitly made this an aspect of his program and an Attorney General put the force of his office behind a specific set of plans, both for the first time, bills to this end were this year passed in both houses of Congress and have finally been enacted into law. 77 The lesson of this experience seems clear. The legislative process is not, at best, a fast one. Powerful interests must normally be mo- bilized in order to get legislation approved. Reform in the criminal process has very little political appeal. There is no constituency of any consequence behind it, aside from a few professional organizations whose concern tends to exist in inverse ratio to their power. If it has taken twenty-five years to bring Johnson v. Zerbst to the brink of puberty, how long will the childhood of Gideon v. Wainwright have to last ?

Inertia is not the only force to be contended with. Hostility to the Due Process Model and its works is widely shared and is effectively mobilized by police and prosecutorial organizations. Every significant move in the Due Process direction has been greeted with predictions of an imminent breakdown in the criminal process. Because judicial activity has been based on the high ground of the Constitution, there have been not many instances of legislative riposte, although examples do exist. But those who unhesitatingly give an affirmative answer to the rhetorical question, "Are the courts handcuffing the police?" need not get their prescriptions enacted into law; theirs is the status quo, and they can maintain it well enough by resisting legislative efforts to pro- vide the resources required to translate Due Process prescriptions into operative fact. Behind all this stands that enigmatic force, public opinion. Just where it stands we cannot know. Television and the other mass media 76 304 U.S. 458 (1938). 77Criminal Justice Act, Pub. L. No. 455, 88th Cong., 2d Sess. (Aug. 20, 1964) (33 U.S.L. Wz 19 (Stat. Aug. 25, 1964)). 64 UNIVERSITY OF PENNSYLVANIA LAW REVIEW seem to be making the defender of the accused into a folk hero. Yet one suspects that a substantial if not a preponderant segment of the public has little sympathy with the tenets of the Due Process Model. It is hard to think that the balance of advantage in the criminal process, if that is a reckonable entity, now lies with the accused."' Yet there is some evidence that at least a segment of the public believes that it does, and that the pendulum has swung dangerously away from order. When speculations of this sort become the stuff of political campaigns, it is evident that there are powerful currents running against the trend. Who can say what will happen to the Due Process Model? These doubts may conduce to a picture too one-sided for accuracy. One of the most powerful features of the Due Process Model is that it thrives on visibility. People are willing to be complacent about what goes on in the criminal process as long as they are not too often or too explicitly reminded of the gory details. The more often specific cases are brought to light of invasions of privacy, of coerced confessions, of excessive bail, of lengthy periods of pretrial detention, and of depriva- tions of the assistance of counsel, the harder it becomes to maintain that the process should go on being primarily administrative and man- agerial. At root, the Due Process Model depends on the functioning of what has been called the sense of injustice. No one, Supreme Court Justices included, is immune to the force of the horrible example. And therein lies the Due Process Model's peculiar strength. It is self- sustaining because its own operations uncover the raw material that fuels its continued growth. It would take a conspiracy of silence to check the mobilization of energies that perpetuates the Due Process revolution. That is a conspiracy we are not likely to get. To start at the small end, the renaissance of criminal studies that has taken place in this country in the last fifteen or twenty years has produced a generation of scholars uniquely knowledgeable about and alert to the problems of the criminal process. They, in turn, are having an impact on students of the law that may well reverse the historic tendency of the American bar to ignore the problems of the criminal law and give us, if not a corps of professionals, at least a generation of dedicated amateurs. These tendencies are also producing a new journalism about the crimi- nal process whose publicists will help to ensure that the subject does not drop out of sight. The Due Process Model is to a significant extent the model of the schools. The next generation of lawyers and judges will have cut their teeth on its tenets. Cultural lag, then, is on the side of the Due Process Model. 78 See Goldstein, The State and the Accused: Balance of Advantage in Criminal Procedure, 69 YA.z L.J. 1149 (1960). [Vol.113:1 TWO MODELS OF THE CRIMINAL PROCESS Beyond the immediate arena of the criminal law, there are grow- ing interests in national life that may be expected as a kind of by- product to foster the development of the Due Process Model. Two in particular may be mentioned-civil rights and poverty. The criminal process has been and will probably continue to be an important forum in the struggle over civil rights. The coercive use of the criminal process-police brufality, arrests on inadequate grounds, denial of bail or excessively high bail, denial of access to counsel, prejudiced tri- bunals-has focused and will continue to focus attention on the prob- lem of adequate challenge in the process, that mainspring of the Due Process Model. Just as the Jehovah's Witnesses made much of our law on free speech, so may the Negro demonstrators be expected to make much of our law on the criminal process. Beyond that, the plight of the Negro as criminal defendant out of all proportion to his numbers in the population may be expected to produce legislative and extragovernmental interest in the workings of the criminal process. The problem of poverty is not far removed. As we have seen, an im- portant dimension of the Due Process ideology is in its insistence upon equality in the operation of the criminal process. The problem of crime is to an important extent a problem of poverty. The current national interest in poverty cannot fail, first indirectly and then directly, to confront the manifold relationships between poverty and the crimi- nal process. We have it on high authority that poverty and the admin- istration of criminal justice must be dealt with together in the formula- tion of national policies. 9 Unless the current that now appears to be running so strongly toward governmental concern in the problem of poverty is abruptly reversed, the exponents of the Due Process Model may expect to find powerful official support that has not hitherto been available in their cause. One can do no more than venture a guess about the continuation of present trends in the criminal process. In the short run, at least, it seems probable that the development and consolidation of norms drawn from the Due Process Model will not slacken. In particular norms that require the equalization of opportunity to challenge the process are likely to become firmly established. This trend, as sym- bolized by the decisions on the right to assigned counsel, may in the end be a far more momentous one than trends expanding particular substantive rights of the accused. If those rights are not further extended, or even if they are curtailed, there will remain a list of oppor- tunities to challenge the operation of the process that far exceeds the 79 See Address by Attorney General Robert F. Kennedy, University of Chicago Law School, May 1, 1964. 66 UNIVERSITY OF PENNSYLVANIA LAW REVIEW present capacity of criminal defendants to use it. If the recent trends toward the prescriptions of the Due Process Model in such matters as powers of arrest, use of illegally obtained evidence, confessions during police interrogation, pretrial liberty, and early access to counsel were now to come to an abrupt halt-an unlikely eventuality-the theoreti- cal operation of the process would still look very much like the Due Process Model. The burning question will be whether the great mass of criminal defendants, who are financially unable to invoke the chal- lenges now available to them, will find their capacity to do so materially improved by measures designed more closely to assimilate their oppor- tunities to those presently enjoyed by the small minority of the fully able. The interest now being displayed by governmental and extra- governmental organizations in devising ways to implement this norm of the Due Process Model suggests that this trend is both durable and influential.

What are the implications for the criminal sanction of the trend toward the Due Process Model? If it tends to become not merely a legal norm but an operational fact that the accused will have a much better opportunity to challenge the operation of the process than he presently enjoys, what, if anything, should that fact have to say to our hypothetical rational legislator 0 as he ponders the uses and limits of the criminal sanction? The problems are, of course, somewhat inter- dependent, in that official decision-makers will have a great deal to say about the extent to which the developing Due Process Model is allowed to become operational fact. Let us, nonetheless, make the simplifying assumption that to a degree the trend in the process is irreversible and that, for whatever complex of reasons and however reluctantly, the rational legislator finds himself confronted by a Given so far as the changed mechanics of the criminal process are concerned. The criminal process now looks to him as if it is becoming a some- what unwieldy instrument of public policy, especially for dealing with large numbers of defendants. It is not to the same extent the low-cost, high-speed process envisioned in the Crime Control Model and reflected in the present real-world situation. At the level of resources presently devoted to its operation, the capacity of the criminal process for dealing with its rapidly increasing annual intake must be seriously impaired. One line of solution is to throw more resources into the operation of the process-more policemen, more prosecutors, more judges, and more supporting services of all kinds. While there has undoubtedly been an expansion in the public resources devoted to the criminal so See note accompanying title supra. [Vo1.113:1 TWO MODELS OF THE CRIMINAL PROCESS process, perhaps even greater than would have been called for by in- creases in population or the so-called crime rate, we have not so far been noted for the steadiness of our attention to the resource needs of the process. Indeed, no systematic work has been done on the extent of those needs. Beyond that, even if conscious choice were to lead in that direction, there may come a point at which the quality of life in a free society is adversely affected by increases in the proportion of public resources employed to detect, prosecute, and punish activity that has been defined as criminal. Some increase in the resources available for operation of the process is undoubtedly warranted, even apart from the demands of the shift toward the Due Process Model that we have been exploring; but a conscious choice to meet the problems created by that shift entirely or even predominantly through increasing the resources of crime control seems unwise, if there is an alternative. The alternative that I would commend to the rational legislator is to reexamine the uses now being made of the criminal sanction with a view toward deciding which uses are relatively indispensable and which might with safety (and perhaps even with some net gain to the public welfare) be restricted or relinquished. There is nothing inherent in the nature of things about the penal code of any time and place. The behavior content of the criminal law has expanded enormously over the past century, mainly because declaring undesirable conduct to be criminal is the legislative line of least resistance for coping with the vexing problems of an increasingly complex and interdependent society. As a result we have inherited a strange m6lange of criminal proscriptions, ranging from conduct that offers the grossest kind of threat to important social interests to conduct whose potentiality for harm is trivial or nonexistent. It is always in order to question the uses made of this most awe- some and coercive of sanctions. It is especially appropriate to do so at a time when the processes that are invoked to apply the criminal sanction are undergoing a profound change that renders them unsuit- able for being lightly employed. What we require is a set of criteria for distinguishing the "mandatory" uses of the criminal sanction from the "optional" ones. Particular attention need be paid to that large group of consensual offenses in which it is not always easy to say who is being injured and by whom. Offenses of that kind-narcotics, gambling, and alcoholism are three statistically conspicuous examples- afford a special opportunity to canvass the important question of alternatives to the criminal sanction. And through that kind of exam- ination, the foundation that we do not now have for a jurisprudence 68 UNIVERSITY OF PENNSYLVANIA LAW REVIEW [Vo1.113:1 of sanctions may eventually be laid. It may be predicted that the change in our model of the criminal process will provide not merely a reason for pressing an inquiry into the appropriate criteria for legislative invocation of the criminal sanction, as has been argued in this Article, but also the source of some valuable clues to what some of those criteria ought to be.