With the Sentencing Reform Act of 1984, Congress enacted what may be the most important single change in federal judicial procedure, since the promulgation of the Federal Rules of Civil Procedure in 1938. In reforms that would have been of great significance even by themselves, the Act eliminated the institution of parole from federal jurisprudence, and made sentencing decisions appealable, as they had not been since a brief period in the late nineteenth century. Most significantly, however, the 1984 Act set up an administrative agency, the United States Sentencing Commission, to establish mandatory rules to guide federal judges in the exercise of their sentencing authority. This sweeping reform and bureaucratization of the sentencing system reflected the popular belief -based on studies of dubious methodological soundness- that discretion in sentencing led to vast, arbitrary disparities among judges’ sentencing decisions.

I

Since the early years of this century, federal sentencing policy had emphasized the rehabilitation of offenders. To this end, legislation had deliberately permitted sentencing to re-main somewhat unpredictable and discretionary: trial judges had leeway to be severe or lenient and the United States Parole Commission could greatly reduce sentences for good behavior and presumptive rehabilitation. Claiming to see in all this discretion the rampant abuse of judicial power, the reformers replaced the judgment of judges in individual cases with the judgment of bureaucrats with mandatory blanket rules -euphemistically known as guidelines.

Reformers were particularly concerned that the previous system gave free reign to the racial, ethnic, and class-based prejudices of judges, permitting them to impose more severe sentences on nonwhites and on members of other disadvantaged groups. The American Civil Liberties Union, for example, advocated strict legislation prohibiting judges from considering education, vocational skills, and employment record and family or community ties in sentencing.2 In keeping with this important goal, the Sentencing Commission was required to ensure that its Guidelines were “entirely neutral as to the race, sex, national origin, creed, and socio-economic status of offenders. This requirement reflected a central aspiration of the Guidelines: imposing strict limits on what was described as the terrifying and almost wholly unchecked discretionary power of federal judges. In place of the human, discretionary element, the Guidelines would rely on a body of appointed experts to draw up rules covering every circumstance and contingency. The ultimate aim was to create a nearly automatic process, a sentencing algorithm that would require all judges to compute the same sentences in the same types of cases.

II

It is increasingly clear to practitioners at the federal criminal bar -and particularly to the judges tasked with untangling this web for every convicted criminal- that the ambitious system of sentencing automation has been a dismal failure. No one familiar with contemporary federal criminal practice will need reminding of the ponderousness and complexity of the Guidelines themselves. They take the form of a 258-box grid and 700 pages of accompanying commentary -nearly four pounds of documents to be use in determining every sentence. In light of the Rube Goldberg qualities of the Guidelines system, it is not surprising that there has developed, among “federal judges and other who work daily in the system,” the “pervasive concern that the Commission’s guidelines are producing fundamental and deleterious changes in the way federal courts process criminal cases and federal judges use their time.”

Even on their own terms, the Guidelines have failed. First of all, the Guidelines have not eliminated -indeed, they have arguably exacerbated- the problem of arbitrary disparities among sentences imposed for similar offenses. For example, a drug dealer convicted of selling a given quantity of LSD might receive anywhere from as few a 10 months to as many as 235 months -a disparity of nearly 19 years- depending upon whether he sold it in pure form, in gelatin capsules, on blotter paper or in sugar cubes. It is especially striking that members of racial and ethnic minorities continue to fare worse under the Guidelines than wealthier non-minorities do. Since members of these groups are statistically more likely to be convicted of crimes involving narcotics and firearms, they tend to receive the most severe sentences meted out under the Guidelines system, which has been designed to punish such offenses with special severity. Indeed, it appears that “denying the judges the opportunity to mitigate sentences on the basis of social disadvantage has worked against poor and minority defendants.” Today, I know of no federal trial judge from a minority group who can be counted as a supporter of the Federal Guidelines -while distinguished minority judges, such as Harry Edwards of the District of Columbia Circuit, Nathaniel Jones of the Sixth Circuits, and Terry Hatter of the Central District of California, are in the forefront of the opposition.

Moreover, the Guidelines have not eliminated, or even severely constrained, the exercise of discretion in sentencing decisions. Instead, the Guidelines have concealed and distorted the discretionary decisions that affect sentencing. Federal prosecutors can dramatically affect sentencing outcomes by their decisions about which charges to bring and what pleas to accept. Federal probation officers, who have been called the guardians of the Guidelines, possess broad powers as fact-finders and sentencing advisors. The Sentencing Commission itself, of course, also possesses substantial discretion in deciding the sentences applicable to particular categories of offenses and offenders. Finally -somewhat paradoxically- federal trial judges themselves retain substantial discretion over sentencing decisions, despite the presence of the Guidelines, but they are now permitted to exercise this discretion only by performing elaborate gyrations within the confines of the Guidelines calculations. This element of judicial discretion within a purportedly mandatory system of rules produces a game of tug-or-war between the bureaucracy and the bench, as the Sentencing Commission struggles to incorporate or repudiate the exceptions articulated by individual judges or appellate courts. The result is clear: rather than eliminating discretionary sentencing decisions, the Guidelines have reduced the comprehensibility of those decisions and the accountability of the decision markers. Finally and perhaps worst of all, the Guidelines and sentencing hearings generally are largely incomprehensible to both victims and defendants- if not to lawyers and judges themselves. Nothing can be more disconcerting to a District Judge than to watch a defendant and his family and members of the general public -and perhaps even the family of his victim- sitting in a courtroom, bewildered by an hour of intricate discussion between court and counsel over computations, additions, deductions, exclusions, inclusions, departures and non-departures. These arcane and mechanistic computations are intended to produce a form of scientific precision, but in practice, they generate a dense fog of confusion that undermines the legitimacy of judges’ sentencing decisions. The Guidelines were designed to limit, if not eliminate, judges’ efforts to make individualized decisions reflecting the particular circumstances of individual cases. It should not surprise us that they have substantially succeeded in doing so. Rather than relying on jurists to exercise Weston and judgment, we now merely ask them to perform an automaton’s function by applying stark formulae set by a central power. Can this really be an improvement in the quality of the justice we all administer?

III

If we are to restore the legitimacy of the federal sentencing system in the eyes of both, criminal defendants and the general public, we must recognize that the Sentencing Guidelines are based on a fundamental misconception about the administration of justice: the belief that just outcomes can always be defined by a comprehensive code applicable to all persons and circumstances, one that could be applied as easily by a computer as by a human being. We must recognize, in other words, that no system of rigid rules -no system devoid of discretion- can fully capture all of our intuitions about what justice requires.

The desire to achieve rational, technocratic solutions to human problems has deep roots in our intellectual traditions. The Enlightenment thinkers who inspired our nation’s founders possessed an abiding faith in the power of reason and in their own power to explain the human and physical world in terms of a finite number of fixed laws. Similarly, the legal positivism of the nineteenth century hoped to reduce all laws to a single comprehensive system of abstract rules. In our century, well-intentioned reformers have often proposed complex, centrally controlled regulatory regimes as a means of addressing many kinds of social ills. These thinkers have contributed much to the progress of our nation, producing incalculable benefits in science, technology, medicine, and, to be sure, the law. But our experience has also taught us that technocratic solutions often backfire, undermining the very purposes they were intended to serve.

Yet we still cling to the illusions that have produced what Vaclav Havel has called the crisis of modern thought. We persist (as Havel has written) in the “proud belief that man, as the pinnacle of everything that exists, [is] capable of objectively describing, explaining and controlling everything… and of possessing the one and only truth about the world.” If we are to learn, anything from the recent past, it should be the error of trying to govern human affairs through a single, centrally planned and scientifically prescribed model.

Our faith in technology and planning can serve us well, despite its dangers, if we remain equally committed to another important element of our political and intellectual tradition: our distrust of governmental institutions, and our awareness of the ways in which institutions designed to serve us can become our masters. This skepticism acts as a powerful check upon creeping accretions of power and upon the erosion of liberty and justice. But in the federal sentencing reforms of the 1980s, we have let the first of these traditions capture and corrupt the second. Driven by our rationalist ambitions for an all-encompassing technocratic Solution, we vent our fears of authority against the only sort of power that is recognizably human, and embrace its coldest and most impersonal alternative.

In a world in which the exercise of some authority cannot be avoided -in which justice must be administered every minute of every day by mortals- we must learn once again to trust the exercise of judicial discretion. We would do well to remember Vaclav Havel’s words: “We cannot devise”, he reminds us, “a system that will eliminate all the disastrous consequences of previous systems. We cannot discover a law or theory whose technical application will eliminate all the disastrous consequences of the technical application of earlier laws and technologies. We have to abandon the arrogant belief that the world is merely a puzzle to be solved, a machine, with instructions for use waiting to be discovered, a body of information to be fed into a computer in the hope that, sooner or later, it will spit out a universal solution.”

Once we recognize the irreducible need for individualized judgment, and for humanity, as well as rationality in sentencing, we can begin to see the shape that sentencing reform must take. First, if judges are to exercise discretion rather than plugging numbers into formulae, the process that should concern us must is that of choosing the men and women of whom we ask judgment: judges. This truth was so fundamental to the Founders, that the Constitution itself protects the independence of the federal judiciary -through provisions for life tenure and salary protection- and ensures that no single official will fully control the appointment of federal judges. In this way, the Founders sought to assure the reasonably good judgment (if not the wisdom) of the federal bench, and to guarantee the independence necessary for the exercise of that judgment.

Second, if judges’ discretion must be constrained (as it surely must), we must look to the mechanisms that have been used for centuries to impose such constraints: requiring trial judges to give reasons for their decisions, and then permitting litigants to seek review of those decisions in appellate courts. Since 1789, the discretion of each judge in most other important matters has also been reviewable by a higher court, all the way to the Supreme Court itself. Each disposition, affordance or reversal are placed permanently on the record and are publicly available for all to see. A similar model could also work well in sentencing: trial judges could be required to explain their sentencing decisions in their own terms on the record, with the understanding that those decisions would be appealed by the Government and the defendant. This approach would build on one of the strengths of the system established by the Sentencing Reform Act of 1984: the automatic appealability of sentences by both sides. If this approach were adopted, we would enjoy all the strengths of tailorable discretion and all the accountability of appellate review, but without the rigidities of our present mechanical determinism.

The authors of the Guidelines have begun to recognize the unworkability of the system they have created. We now see, for example, that Judge Stephen Breyer -a principal architect of the Guidelines, as an original member of the Sentencing Commission- is willing to permit trial judges to exercise greater discretion in departing upward or downward from the Guideline sentencing ranges. Under his guidance, the court of Appeals for the First Circuit has now arguably changed the standard of appellate review of sentencing departure from a de novo approach to a review, in effect, only for reasonableness. This is indeed the direction to go, but the solution is not more leeway in departure, but outright abolition of the Guidelines framework. Slowly allowing more discretion to coexist with the mandatory rules of the Guidelines would add still more incoherence and incomprehensibility and would further burden courts and confuse litigants.

Replacing the Guidelines with a system of appealable discretion is clearly the answer. To the extent that there might ever be any temptation for judges to abuse their discretion, this would be remedied by appellate review of sentencing decisions. This, after all, is routinely done with other matters of law decided at trial. Judges, however, would no longer be vexed with applying the 700-page Guidelines, litigants, and the public would no longer be so baffled by sentencing, and better justice would be done. Judges would again try to tailor punishment to the particular circumstances of the offense, as they understand it after presiding throughout the trial. They would, however, be forced to record their reasoning and to subject it to review from above, on appeal as of right by either party.

British courts, in this respect, could teach us a great deal. There, while there are no rigid guidelines for sentencing, sentences are reviewable on appeal. British appellate courts set standards for trial judges to apply in sentencing, just as they do in other areas of the law. Civil law jurisdictions employ individual judgment and peer review simultaneously by giving judicial panels the job of sentencing. Thus they ensure that decisions “will be the product of the collective judgment of [the] judicial panel and not dependent on the arbitrary judgment of a single, all-powerful judge”. With such models so near at hand, it would be a mistake to assume that the only alternative to retaining the Guidelines is a return to the pre-Guidelines system.

Appellate review of discretionary sentencing allows us to have our cake and eat it too; it offers us a way to preserve the humanity of the sentencing system while preventing potential “disparities” from eroding the basic principle that, all other things being equal, similar crimes should entail similar punishment. This proposal would allow judges to exercise the judgment that is their calling while forcing them to explain themselves to their peers, to the public, and to posterity. It would also restore the flexibility necessary in sentencing while avoiding the unintelligible and the bizarre.

Perhaps best of all, an appellate review process for federal criminal sentencing would once more be comprehensible to the public and to parties before the court. Punishment serves ends beyond the mere application of official injury to a convicted criminal: it has a powerful social function as the concretization of our disapproval of a particular act. Whatever penal theory one wishes to apply,19 the clarity of the connection between wrong and punishment is vital. We must once again establish a clear connection between the offense and the sanction in criminal sentencing.

Sentencing that becomes incomprehensible to the convicted defendant being punished, mysterious to his victim, and baffling to the public, has taken a dangerous step toward being perceived as being no more than arbitrary force. The legitimacy of our system of justice demands that its workings be understood by those upon whom it works. Who, it might be asked, can really be said to have had this proverbial “day in court” if the court’s sentencing decision is nothing more than a mechanistic computation to all but the most expert. If in 1993 we no longer trust central planners with our pocketbooks or our political liberties, what business have we ceding to them administration of our criminal justice system?

Justice may sometimes require more leniency than the Guidelines permit. It may sometimes require more severity. Justice, if submit, will always, however, require more nuance and flexibility than can be provided by a set of “guidelines” fashioned by a bureaucracy.