JUSTICE BREYER, with whom THE CHIEF JUSTICE joins, dissenting.
The majority holds that the Constitution contains the following requirement: "any fact [other than recidivism] that increases the penalty for a crime beyond the prescribed statutory maximum must be submitted to a jury, and proved beyond a reasonable doubt." Ante, at 490. This rule would seem to promote a procedural ideal—that of juries, not judges, determining the existence of those facts upon which increased punishment turns. But the real world of criminal justice cannot hope to meet any such ideal. It can function only with the help of procedural compromises, particularly in respect to sentencing. And those compromises, which are themselves necessary for the fair functioning of the criminal justice system, preclude implementation of the procedural model that today's decision reflects. At the very least, the impractical nature of the requirement that the majority now recognizes supports the proposition that the Constitution was not intended to embody it.
In modern times the law has left it to the sentencing judge to find those facts which (within broad sentencing limits set by the legislature) determine the sentence of a convicted offender. The judge's factfinding role is not inevitable. One could imagine, for example, a pure "charge offense" sentencing system in which the degree of punishment depended only upon the crime charged (e.g., eight mandatory years for robbery, six for arson, three for assault). But such a system would ignore many harms and risks of harm that the offender caused or created, and it would ignore many relevant offender characteristics. See United States Sentencing Commission, Sentencing Guidelines and Policy Statements, Part A, at 1.5 (1987) (hereinafter Sentencing Guidelines or Guidelines) (pointing out that a "charge offense" [p556] system by definition would ignore any fact "that did not constitute [a] statutory elemen[t] of the offens[e] of which the defendant was convicted"). Hence, that imaginary "charge offense" system would not be a fair system, for it would lack proportionality, i.e., it would treat different offenders similarly despite major differences in the manner in which each committed the same crime.
There are many such manner-related differences in respect to criminal behavior. Empirical data collected by the Sentencing Commission makes clear that, before the Guidelines, judges who exercised discretion within broad legislatively determined sentencing limits (say, a range of 0 to 20 years) would impose very different sentences upon offenders engaged in the same basic criminal conduct, depending, for example, upon the amount of drugs distributed (in respect to drug crimes), the amount of money taken (in respect to robbery, theft, or fraud), the presence or use of a weapon, injury to a victim, the vulnerability of a victim, the offender's role in the offense, recidivism, and many other offense-related or offender-related factors. See United States Sentencing Commission, Supplementary Report on the Initial Sentencing Guidelines and Policy Statements 35–39 (1987) (table listing data representing more than 20 such factors) (hereinafter Supplementary Report); see generally Department of Justice, W. Rhodes & C. Conly, Analysis of Federal Sentencing (May 1981). The majority does not deny that judges have exercised, and, constitutionally speaking, may exercise sentencing discretion in this way.
Nonetheless, it is important for present purposes to understand why judges, rather than juries, traditionally have determined the presence or absence of such sentence-affecting facts in any given case. And it is important to realize that the reason is not a theoretical one, but a practical one. It does not reflect (JUSTICE SCALIA's opinion to the contrary notwithstanding) an ideal of procedural "fairness," ante, at 498 (concurring opinion), but rather an administrative need [p557] for procedural compromise. There are, to put it simply, far too many potentially relevant sentencing factors to permit submission of all (or even many) of them to a jury. As the Sentencing Guidelines state the matter,
"[a] bank robber with (or without) a gun, which the robber kept hidden (or brandished), might have frightened (or merely warned), injured seriously (or less seriously), tied up (or simply pushed) a guard, a teller or a customer, at night (or at noon), for a bad (or arguably less bad) motive, in an effort to obtain money for other crimes (or for other purposes), in the company of a few (or many) other robbers, for the first (or fourth) time that day, while sober (or under the influence of drugs or alcohol), and so forth." Sentencing Guidelines, Part A, at 1.2.
The Guidelines note that "a sentencing system tailored to fit every conceivable wrinkle of each case can become unworkable and seriously compromise the certainty of punishment and its deterrent effect." Ibid. To ask a jury to consider all, or many, such matters would do the same.
At the same time, to require jury consideration of all such factors—say, during trial where the issue is guilt or innocence—could easily place the defendant in the awkward (and conceivably unfair) position of having to deny he committed the crime yet offer proof about how he committed it, e.g., "I did not sell drugs, but I sold no more than 500 grams." And while special postverdict sentencing juries could cure this problem, they have seemed (but for capital cases) not worth their administrative costs. Hence, before the Guidelines, federal sentencing judges typically would obtain relevant factual sentencing information from probation officers' presentence reports, while permitting a convicted offender to challenge the information's accuracy at a hearing before the judge without benefit of trial-type evidentiary rules. See Williams v. New York, 337 U.S. 241, [p558] 249–251 (1949) (describing the modern "practice of individualizing punishments" under which judges often consider otherwise inadmissible information gleaned from probation reports); see also Kadish, Legal Norm And Discretion In The Police And Sentencing Processes, 75 Harv. L. Rev. 904, 915–917 (1962).
It is also important to understand how a judge traditionally determined which factors should be taken into account for sentencing purposes. In principle, the number of potentially relevant behavioral characteristics is endless. A judge might ask, for example, whether an unlawfully possessed knife was "a switchblade, drawn or concealed, opened or closed, large or small, used in connection with a car theft (where victim confrontation is rare), a burglary (where confrontation is unintended) or a robbery (where confrontation is intentional)." United States Sentencing Commission, Preliminary Observations of the Commission on Commissioner Robinson's Dissent 3, n.3 (May 1, 1987). Again, the method reflects practical, rather than theoretical, considerations. Prior to the Sentencing Guidelines, federal law left the individual sentencing judge free to determine which factors were relevant. That freedom meant that each judge, in an effort to tailor punishment to the individual offense and offender, was guided primarily by experience, relevance, and a sense of proportional fairness. Cf. Supplementary Report, at 16–17 (noting that the goal of the Sentencing Guidelines was to create greater sentencing uniformity among judges, but in doing so the Guidelines themselves had to rely primarily upon empirical studies that showed which factors had proved important to federal judges in the past).
Finally, it is important to understand how a legislature decides which factual circumstances among all those potentially related to generally harmful behavior it should transform into elements of a statutorily defined crime (where they would become relevant to the guilt or innocence of an accused), and which factual circumstances it should leave to [p559] the sentencing process (where, as sentencing factors, they would help to determine the sentence imposed upon one who has been found guilty). Again, theory does not provide an answer. Legislatures, in defining crimes in terms of elements, have looked for guidance to common-law tradition, to history, and to current social need. And, traditionally, the Court has left legislatures considerable freedom to make the element determination. See Almendarez-Torres v. United States, 523 U.S. 224, 228 (1998); McMillan v. Pennsylvania, 477 U.S. 79, 85 (1986).
By placing today's constitutional question in a broader context, this brief survey may help to clarify the nature of today's decision. It also may explain why, in respect to sentencing systems, proportionality, uniformity, and administrability are all aspects of that basic "fairness" that the Constitution demands. And it suggests my basic problem with the Court's rule: A sentencing system in which judges have discretion to find sentencing-related factors is a workable system and one that has long been thought consistent with the Constitution; why, then, would the Constitution treat sentencing statutes any differently?
As JUSTICE THOMAS suggests, until fairly recent times many legislatures rarely focused upon sentencing factors. Rather, it appears they simply identified typical forms of antisocial conduct, defined basic "crimes," and attached a broad sentencing range to each definition—leaving judges free to decide how to sentence within those ranges in light of such factors as they found relevant. Ante, at 510–512, 518 (concurring opinion). But the Constitution does not freeze 19th-century sentencing practices into permanent law. And dissatisfaction with the traditional sentencing system (reflecting its tendency to treat similar cases differently) has led modern legislatures to write new laws that refer specifically to sentencing factors. See Supplementary Report 1 [p560] (explaining that "a growing recognition of the need to bring greater rationality and consistency to penal statutes and to sentences imposed under those statutes" led to reform efforts such as the Federal Sentencing Guidelines).
Legislatures have tended to address the problem of too much judicial sentencing discretion in two ways. First, legislatures sometimes have created sentencing commissions armed with delegated authority to make more uniform judicial exercise of that discretion. Congress, for example, has created a federal Sentencing Commission, giving it the power to create Guidelines that (within the sentencing range set by individual statutes) reflect the host of factors that might be used to determine the actual sentence imposed for each individual crime. See 28 U.S.C. § 994(a); see also United States Sentencing Commission, Guidelines Manual (Nov. 1999). Federal judges must apply those Guidelines in typical cases (those that lie in the "heartland" of the crime as the statute defines it) while retaining freedom to depart in atypical cases. Id., ch. 1, pt. A, 4(b).
Second, legislatures sometimes have directly limited the use (by judges or by a commission) of particular factors in sentencing, either by specifying statutorily how a particular factor will affect the sentence imposed or by specifying how a commission should use a particular factor when writing a guideline. Such a statute might state explicitly, for example, that a particular factor, say, use of a weapon, recidivism, injury to a victim, or bad motive, "shall" increase, or "may" increase, a particular sentence in a particular way. See, e.g., McMillan, supra, at 83 (Pennsylvania statute expressly treated "visible possession of a firearm" as a sentencing consideration that subjected a defendant to a mandatory 5-year term of imprisonment).
The issue the Court decides today involves this second kind of legislation. The Court holds that a legislature cannot enact such legislation (where an increase in the maximum is involved) unless the factor at issue has been charged, [p561] tried to a jury, and found to exist beyond a reasonable doubt. My question in respect to this holding is, simply, "why would the Constitution contain such a requirement"?
In light of the sentencing background described in Parts I and II, I do not see how the majority can find in the Constitution a requirement that "any fact" (other than recidivism) that increases the maximum penalty for a crime "must be submitted to a jury." Ante, at 490. As JUSTICE O'CONNOR demonstrates, this Court has previously failed to view the Constitution as embodying any such principle, while sometimes finding to the contrary. See Almendarez-Torres, supra, at 239–247; McMillan, supra, at 84–91. The majority raises no objection to traditional pre-Guidelines sentencing procedures under which judges, not juries, made the factual findings that would lead to an increase in an individual offender's sentence. How does a legislative determination differ in any significant way? For example, if a judge may on his or her own decide that victim injury or bad motive should increase a bank robber's sentence from 5 years to 10, why does it matter that a legislature instead enacts a statute that increases a bank robber's sentence from 5 years to 10 based on this same judicial finding?
With the possible exception of the last line of JUSTICE SCALIA's concurring opinion, the majority also makes no constitutional objection to a legislative delegation to a commission of the authority to create guidelines that determine how a judge is to exercise sentencing discretion. See also ante, at 523, n.11 (THOMAS, J., concurring) (reserving the question). But if the Constitution permits Guidelines, why does it not permit Congress similarly to guide the exercise of a judge's sentencing discretion? That is, if the Constitution permits a delegatee (the commission) to exercise sentencing-related rulemaking power, how can it deny the [p562] delegator (the legislature) what is, in effect, the same rulemaking power?
The majority appears to offer two responses. First, it argues for a limiting principle that would prevent a legislature with broad authority from transforming (jury-determined) facts that constitute elements of a crime into (judge-determined) sentencing factors, thereby removing procedural protections that the Constitution would otherwise require. See ante, at 486 ("constitutional limits" prevent states from "defin[ing] away facts necessary to constitute a criminal offense"). The majority's cure, however, is not aimed at the disease.
The same "transformational" problem exists under traditional sentencing law, where legislation, silent as to sentencing factors, grants the judge virtually unchecked discretion to sentence within a broad range. Under such a system, judges or prosecutors can similarly "transform" crimes, punishing an offender convicted of one crime as if he had committed another. A prosecutor, for example, might charge an offender with five counts of embezzlement (each subject to a 10-year maximum penalty), while asking the judge to impose maximum and consecutive sentences because the embezzler murdered his employer. And, as part of the traditional sentencing discretion that the majority concedes judges retain, the judge, not a jury, would determine the last-mentioned relevant fact, i.e., that the murder actually occurred.
This egregious example shows the problem's complexity. The source of the problem lies not in a legislature's power to enact sentencing factors, but in the traditional legislative power to select elements defining a crime, the traditional legislative power to set broad sentencing ranges, and the traditional judicial power to choose a sentence within that range on the basis of relevant offender conduct. Conversely, the solution to the problem lies, not in prohibiting legislatures from enacting sentencing factors, but in sentencing rules that determine punishments on the basis of properly defined [p563] relevant conduct, with sensitivity to the need for procedural protections where sentencing factors are determined by a judge (for example, use of a "reasonable doubt" standard), and invocation of the Due Process Clause where the history of the crime at issue, together with the nature of the facts to be proved, reveals unusual and serious procedural unfairness. Cf. McMillan, 477 U.S., at 88 (upholding statute in part because it "gives no impression of having been tailored to permit the [sentencing factor] to be a tail which wags the dog of the substantive offense").
Second, the majority, in support of its constitutional rule, emphasizes the concept of a statutory "maximum." The Court points out that a sentencing judge (or a commission) traditionally has determined, and now still determines, sentences within a legislated range capped by a maximum (a range that the legislature itself sets). See ante, at 481–482. I concede the truth of the majority's statement, but I do not understand its relevance.
From a defendant's perspective, the legislature's decision to cap the possible range of punishment at a statutorily prescribed "maximum" would affect the actual sentence imposed no differently than a sentencing commission's (or a sentencing judge's) similar determination. Indeed, as a practical matter, a legislated mandatory "minimum" is far more important to an actual defendant. A judge and a commission, after all, are legally free to select any sentence below a statute's maximum, but they are not free to subvert a statutory minimum. And, as JUSTICE THOMAS indicates, all the considerations of fairness that might support submission to a jury of a factual matter that increases a statutory maximum, apply a fortiori to any matter that would increase a statutory minimum. See ante, at 521–522 (concurring opinion). To repeat, I do not understand why, when a legislature authorizes a judge to impose a higher penalty for bank robbery (based, say, on the court's finding that a victim was injured or the defendant's motive was bad), a new crime is born; but [p564] where a legislature requires a judge to impose a higher penalty than he otherwise would (within a pre-existing statutory range) based on similar criteria, it is not. Cf. Almendarez-Torres, 523 U.S., at 246.
I certainly do not believe that the present sentencing system is one of "perfect equity," ante, at 498 (SCALIA, J., concurring), and I am willing, consequently, to assume that the majority's rule would provide a degree of increased procedural protection in respect to those particular sentencing factors currently embodied in statutes. I nonetheless believe that any such increased protection provides little practical help and comes at too high a price. For one thing, by leaving mandatory minimum sentences untouched, the majority's rule simply encourages any legislature interested in asserting control over the sentencing process to do so by creating those minimums. That result would mean significantly less procedural fairness, not more.
For another thing, this Court's case law, prior to Jones v. United States, 526 U.S. 227, 243, n.6 (1999), led legislatures to believe that they were permitted to increase a statutory maximum sentence on the basis of a sentencing factor. See ante, at 529–539 (O'CONNOR, J., dissenting); see also, e.g., McMillan, supra, at 84–91 (indicating that a legislature could impose mandatory sentences on the basis of sentencing factors, thereby suggesting it could impose more flexible statutory maximums on same basis). And legislatures may well have relied upon that belief. See, e.g., 21 U.S.C. § 841(b) (1994 ed. and Supp. III) (providing penalties for, among other things, possessing a "controlled substance" with intent to distribute it, which sentences vary dramatically depending upon the amount of the drug possessed, without requiring jury determination of the amount); N.J. Stat. Ann. §§ 2C:43-6, 2C:43-7, 2C:44-1a-f, 2C:44-3 (West 1995 and Supp. 1999–2000) (setting sentencing ranges for crimes, while providing for lesser or greater punishments [p565] depending upon judicial findings regarding certain "aggravating" or "mitigating" factors); Cal. Penal Code Ann. § 1170 (West Supp. 2000) (similar); see also Cal. Court Rule 420(b) (1996) (providing that "[c]ircumstances in aggravation and mitigation" are to be established by the sentencing judge based on "the case record, the probation officer's report, [and] other reports and statements properly received").
As JUSTICE O'CONNOR points out, the majority's rule creates serious uncertainty about the constitutionality of such statutes and about the constitutionality of the confinement of those punished under them. See ante, at 549–552 (dissenting opinion). The few amicus briefs that the Court received in this case do not discuss the impact of the Court's new rule on, for example, drug crime statutes or state criminal justice systems. This fact, I concede, may suggest that my concerns about disruption are overstated; yet it may also suggest that (despite Jones and given Almendarez-Torres) so absolute a constitutional prohibition is unexpected. Moreover, the rationale that underlies the Court's rule suggests a principle—jury determination of all sentencing-related facts—that, unless restricted, threatens the workability of every criminal justice system (if applied to judges) or threatens efforts to make those systems more uniform, hence more fair (if applied to commissions).
Finally, the Court's new rule will likely impede legislative attempts to provide authoritative guidance as to how courts should respond to the presence of traditional sentencing factors. The factor at issue here—motive—is such a factor. Whether a robber takes money to finance other crimes or to feed a starving family can matter, and long has mattered, when the length of a sentence is at issue. The State of New Jersey has determined that one motive—racial hatred—is particularly bad and ought to make a difference in respect to punishment for a crime. That determination is reasonable. The procedures mandated are consistent with traditional sentencing practice. Though additional proce- [p566] dural protections might well be desirable, for the reasons JUSTICE O'CONNOR discusses and those I have discussed, I do not believe the Constitution requires them where ordinary sentencing factors are at issue. Consequently, in my view, New Jersey's statute is constitutional.
I respectfully dissent.