464 F. 3d 1072, reversed and remanded.

HTML version
PDF version
HTML version
PDF version
HTML version
PDF version

Souter, J., dissenting



on writ of certiorari to the united states court ofappeals for the ninth circuit

[May 19, 2008]

Justice Souter, with whom Justice Stevens and Justice Ginsburg join, dissenting.

The Court chooses one reading of the Armed Career Criminal Act (ACCA), 18 U. S. C. §924(e) (2000 ed. and Supp. V), over another that would make at least as much sense of the statute’s ambiguous text and would follow the counsel of a tradition of lenity in construing perplexing criminal laws. The Court’s choice, moreover, promises hard times for the trial courts that will have to make the complex sentencing calculations this decision demands. I respectfully dissent.


The ACCA mandates a 15-year minimum sentence for anyone convicted of violating §922(g) (2000 ed.) who “has three previous convictions [for] a serious drug offense” among his prior crimes. §924(e)(1) (2000 ed., Supp. V). Section 924(e)(2)(A) (2000 ed.) defines “serious drug offense” as an offense under state or federal drug laws, “for which a maximum term of imprisonment of ten years or more is prescribed by law.” This limitation leaves open the question whether a given conviction qualifies as “serious” by reference to the penalty for the acts making up the basic offense, regardless of who commits it, or whether account must also be taken of further facts (such as an offender’s criminal record that qualified him for an enhanced penalty at the time of that earlier conviction). If the first alternative is the reading Congress intended, a sentencing judge needs to look only to the penalty specified for the basic offense committed by a first-timeoffender. But if the second is the intended one, a judge may have to consider sentencing variations (for using a gun, say, or for repeating the offense) set out in other provisions.

It all turns on the meaning of the word “offense,” to which the “maximum term” is tied. One can naturally read “an offense” at a general level as synonymous with “a crime,” which would tend to rule out reference to maximums adjusted for other facts; we do not usually speak of a crime of “burglary while having a criminal record and while out on bail.” Those details would come up only if we were speaking about a specific instance, described as a burglary “committed by someone with a record while out on bail,” in which case the other facts may “enhance” his sentence beyond what would have been the maximum term for burglary. The trouble is that “offense” could easily refer to a specific occurrence, too; looking at it that way would make it less jarring to suggest that the circumstances around an event that authorize higher penalty ranges (such as the use of a gun) or the defendant’s history (like a prior conviction) ought to count in identifying the maximum penalty for the offense committed on the given day, at the given place, by the particular offender, in a given way. Either reading seems to offer a plausible take on the “offense” for which the ACCA court will have to identify or calculate “maximum” penalties, under state law.

We get no help from imagining the circumstances in which a sentencing court would ask which reading to adopt. The choice of answer would be easy if the question arose in the mind of a lawyer whose client is thinking about a guilty plea and asks what maximum term he faces. See ante, at 4–5. His lawyer knows that he means the maximum term for him in his case. When a repeat offender wants to know, counsel understands that the penalty prescribed for the basic crime without the recidivist add-on is not the baseline for comparison that may make or break the potential plea agreement. And if the repeat offender faces a further statutory enhancement for carrying a gun during the offense, or for being out on bail, his lawyer would not tell him the maximum term for repeat offenders without guns or bail restrictions. By the same token, if the offender faced (as Rodriquez did) a lower sentence ceiling than what the statute says, by grace of mandatory sentencing guidelines, his lawyer would know enough to tell him that his maximum was capped in this way.

When the issue comes up not in a particular client’s questions about his own prospects, however, but in a trial judge’s mind wondering about the meaning of the general statute, context gives no ready answer. Nor does it break the tie to say, as the Court does, that taking “maximum” to refer to the basic offense would mean that a recidivist with add-ons could be sentenced above the ACCA “maximum,” see ante, at 4 (“even if respondent had been sentenced to, say, six years’ imprisonment, ‘the maximum term of imprisonment’ prescribed by law still would have been five years”). That description, after all, might be just a verbal quirk showing the statutory design in proper working order: if Congress meant an offense to be viewed generically and apart from offender characteristics, a gap between the maximum for ACCA purposes, and a heavier, actual sentence accounting for a defendant’s history is to be expected.1

The text does not point to any likelier interpretive choices, and as between these alternatives, it is simply ambiguous.2 Because I do not believe its ambiguity is fairly resolved in the Government’s favor, I would affirm.



None of the Court’s three principal points or ripostes solves the puzzle. To begin with, there is something arbitrary about trying to resolve the ambiguity by rejecting the maximum-for-basic-offense option while declining to consider an entire class of offender-based sentencing adjustments. If offender characteristics are going to count in identifying the relevant maximum penalty, it would seem to follow that in jurisdictions with mandatory sentencing guidelines, the maximum “prescribed by law” would be what the guidelines determine. The original Federal Guidelines, and the mandatory state guidelines I am aware of, were established under statutory authority that invests a guideline with the same legal status as a customary penalty provision. Cf. United States v. R. L. C., 503 U. S. 291, 297 (1992) (“The answer to any suggestion that the statutory character of a specific penalty provi-sion gives it primacy over administrative sentencing guidelines is that the mandate to apply the Guidelines is itself statutory”).

The Court tries to deflect the implication of its position by denying that state sentencing guidelines really do set maximum penalties, since typically they allow a judge to depart from them, up or down, when specified conditions are met. See ante, at 12. But while this is true, the objection stands. However a particular mandatory guideline scheme works, it sets a maximum somewhere; if it includes conditions affecting what would otherwise be a guideline maximum, the top of the range as affected should be the relevant maximum on the Court’s reading of the statute. Indeed, the factual conditions involved are usually offender characteristics, and if the ACCA is going to count them under offense-defining statutes or freestanding recidivism laws, those same facts ought to count under a guideline rule (whether setting, or authorizing a departure from, a particular limit). There is no practical difference whether maximums are adjusted by a statute, a statutorily mandated guideline, or a guideline-specified departure; wherever a “prescri[ption] by law” resides, it ought to be honored by the ACCA court.

If we were to follow the Court’s lights, then, I think we would have to accept the complication that guidelines schemes present, and face the difficulty of calculating enhanced maximums in guidelines jurisdictions.3 What we cannot do is to resolve statutory ambiguity by looking to the sentencing range for an imaginary offender who meets statutory conditions for altering the basic sentence, but is artificially stripped of any characteristic that triggers a guideline rule also “prescribed by law.”


The more fundamental objection, though, goes to the Court’s basic conclusion that it makes the better sense to read the ACCA as resting the federal treatment of recidivists on the maximum sentence authorized by state recidivist schemes, in cases where state law must be considered. The Court says it would have been natural for Congress to think in terms of state judgments about repeat criminals when thinking about what to do at the national level, and the Court is quite possibly right about this; the fact that the federal penalty may turn on a state felony classification at all shows that Congress was thinking about state law. But the chances are at least equally good that the Court is wrong; it is odd to think that Congress would have piggybacked the federal system on state repeat-offender schemes, given the extraordinaryand irreconcilable variations among state policies on the subject.

For one thing, the States’ recidivism schemes vary in their methods for augmenting sentences. Iowa’s law, for example, subjects repeat drug offenders to triple penalties, Iowa Code §124.411(1) (2005); but in Wisconsin a repeat drug distributor will see his maximum term increased bya fixed number of years, whatever the starting point,see, e.g., Wis. Stat. §961.48(1)(b) (2003–2004) (4-year in-crease for Class H felony such as selling 1 kilogram of marijuana).

More striking than differing structures, though, are the vast disparities in severity from State to State: under Massachusetts drug laws, a third conviction for selling a small amount of marijuana carries a maximum of 2.5 years. Mass. Gen. Laws Ann., ch. 94C, §32C(b) (West 2006). In Delaware, a third conviction means a mandatory sentence of life in prison without parole. See Del. Code Ann., Tit. 11, §4214(b) (2007) (third-felony penalty of life without parole for violations of non-narcotic controlled substances law, Tit. 16, §4752 (2004)). That Congress might have chosen to defer to state-law judgments about “seriousness” that vary so widely for the same conduct is at least open to doubt. And that doubt only gets worse when we notice that even where two States have similar maximum penalties for a base-level offense, their recidivist enhancements may lead the same conduct to trigger the ACCA sanction in one State but not the other: on the Court’s view, an offender’s second conviction for selling, say, just over two pounds of marijuana will qualify as an ACCA predicate crime if the conviction occurred in Arizona (maximum of 13 years), Iowa (15 years), Utah (15 years), and the District of Columbia (10 years), for example;4 but it will fall short of the mark in California (8 years), Michigan (8 years), and New York (8 years).5 Yet in each of these States, the base-level offense has a maximum term falling within a much narrower range (between 3.5 and 5.5 years).6 With this backdrop of state law, the Government can hardly be heard to say that there would be something “incongruous” about a federal law targeting offenses flagged by the penalties assigned only to bare conduct, without regard to recidivism or other offender facts. Brief for United States 17.

Nor does it show what the ACCA means by “maximum” or “offense” when the Court points to language from our prior cases saying that enhanced recidivist penalties are not to be viewed as retroactive punishment for past crimes, for purposes of double-jeopardy and right-to-counsel enquiries. See ante, at 7 (citing Nichols v. United States, 511 U. S. 738, 747 (1994) , and Gryger v. Burke, 334 U. S. 728, 732 (1948) ). The quotations show that a separate offense is identified by an enhanced penalty, the Court says, because from them we can draw the conclusion that “[w]hen a defendant is given a higher sentence under a recidivism statute,” nonetheless “100% of the punishment is for the offense of conviction,” leaving nothing to be attributed to “prior convictions or the defendant’s ‘status as a recidivist,’ ” ante, at 7.

Still, the fact is that state-law maximums for repeat offenders sometimes bear hardly any relation to the gravity of the triggering offense, as “three-strikes” laws (not to mention the Delaware example, above) often show. See, e.g., Ill. Comp. Stat., ch. 720, §5/33B–1 (2004) (mandatory life sentence for third “Class X” felony, such as dealing heroin, without regard to the specific penalty gradation for the latest Class X felony or to any similarity with prior offenses); W. Va. Code Ann. §61–11–18(c) (2005) (if offender was “twice before convicted in the United States of a crime punishable by confinement in a penitentiary,” third such conviction incurs a mandatory life sentence). Cf. Ewing v. California, 538 U. S. 11, 30, n. 2 (2003) (plurality opinion) (the “California Legislature therefore made a deliberate policy decision … that the gravity of the new felony should not be a determinative factor in triggering the application of the Three Strikes Law” (internal quotation marks omitted)). And there is no denying that the fact of prior convictions (or a defendant’s recidivist status) is necessary for the “ ‘stiffened penalty’ ” to be imposed for “ ‘the latest crime,’ ” ante, at 7, the necessary fact being specific to the offender, and falling outside the definition of the offense. This is, after all, what it means to apply an “enhancement.”

The upshot is that it may have been natural for Congress to think of state recidivism schemes, but it may well not have been. If there is anything strange about ignoring enhanced penalties, there is something at least as strange about a federal recidivist statute that piles enhancement on enhancement, magnifying the severity of state laws severe to begin with.


Whatever may be the plausibility of the offender-based reading of the statute as the Court describes it, the Court’s description avoids a source of serious doubt by glossing over the practical problems its take on the statute portends. The Court is unmoved by the argument that Congress probably did not expect federal courts applying the ACCA to master the countless complications of state sentencing schemes; because all jurisdictions provide for enhanced sentencing some way or another, the Court thinks there is nothing threatening in the subject, which it tries to simplify by offering a few practical pointers. It notes that there will be cases with a qualifying enhancement “evident from the length of the sentence imposed” by the state court; sometimes, it says, a court’s “judgment of conviction will … list the maximum possible sentence”; or the state prosecutor will have “submit[ted] a formal charging document in order to obtain a recidivist enhancement.” Ante, at 10. And in cases involving pleas, the Court notes, “the plea colloquy will very often include a statement by the trial judge regarding the maximum penalty.” Ibid. Even when there are no pointers to help, says the Court, and “the records that may properly be consulted” yield no clear answer, the worst that can happen will be the Government’s inability to show that a prior conviction qualifies. Ante, at 11.

But it is not that easy, and the Court’s pointers are not much comfort. To start with, even where a “maximum” sentence is mentioned in state records, how will the ACCA court be supposed to know that the “maximum” written down there is what the Court today holds that “maximum” means? A State’s number below 10 years may refer to the base-level offense, or it may be the reduced maximum required by mandatory guidelines; and a number over 10 years may be the product of other enhancements (as for weapons use or being out on bail at the time of commission). Having to enquire into just what imposed sentences or what trial documents really mean would seem to leave plenty of sorting out for the federal courts to do (or at least, for federal prosecutors, if they end up with the job).

Another example: State laws are not written to coordinate with the ACCA, and if a State’s specific repeat drug-offender provisions, say, are supposed to be read together with its general habitual-offender statutes, the resulting “maximum” may not be the Court’s “maximum.” Indeed, a federal court may have to figure out just how those state statutes may be read together to avoid conflict between them, when the way to avoid conflict is not clear cut even for the state courts, see, e.g., Goldberg v. State, 282 Ga. 542, 651 S. E. 2d 667 (2007) (general recidivist statute trumps more specific one; overruling same court’s decision in Mikell v. State, 270 Ga. 467, 510 S. E. 2d 523 (1999)); State v. Keith, 102 N. M. 462, 697 P. 2d 145 (App. 1985) (specific trumps general). Cf. Clines v. State, 912 So. 2d 550 (Fla. 2005) (relying on rule of lenity to resolve whether multiple recidivist categories in same habitual-offender law could apply to a single sentence).

And there is more: as Rodriquez reminds us, just deciding what counts as a “prior” offense under state law is not always an easy thing. See People v. Wiley, 9 Cal. 4th 580, 583, 889 P. 2d 541, 542 (1995) (noting difficulty of applying requirement that “prior” charges have been “brought and tried separately,” where defendant had been convicted in trials occurring one day apart and sentenced at the same court session; in the end, drawing the needed inference from docket numbers revealed on documents requested from the municipal trial court); id., at 595, 889 P. 2d, at 550 (Werdegar, J., dissenting) (protesting the court’s solicitation and use of extra-record documents). Nor would that sort of enquiry get any easier, or be more likely to benefit from well-settled state law, when a given State’s law takes account of prior offenses in other States, see Timothy v. State, 90 P. 3d 177 (Alaska App. 2004) (holding Oklahoma burglary not to be analogous to one in Alaska, for purposes of Alaska’s recidivism enhancements, thus overruling its own 2-year-old decision, Butts v. State, 53 P. 3d 609 (2002)); or, to take a specific example, when what qualifies a prior offense under one State’s recidivism scheme is the length of the sentence authorized by another State’s law (raising the question whether that first State would see recidivist enhancements the same way the Court does today). See, e.g., N. J. Stat. Ann. §2C:44–4(c) (West 2005) (“A conviction in another jurisdiction shall constitute a prior conviction of a crime if a sentence of imprisonment in excess of 6 months was authorized under the law of the other jurisdiction”); N. M. Stat. Ann. §31–18–17(D)(2)(b) (2007 Supp.) (defining “prior felony conviction” as, inter alia, a felony “punishable [by] a maximum term of imprisonment of more than one year”).

A still thornier problem is how federal courts are supposed to treat a State’s procedural safeguards for using prior convictions at sentencing. Saying that congressional deference to the States’ judgments about the severity of crimes also extends to their judgments about recidivism raises, but does not answer, the question whether such deference goes only as far as the state courts themselves could go in raising penalties. (The Court’s disregard of mandatory sentencing guidelines would seem to suggest that the answer is no.) In those States that require notice before the prosecutor can seek a recidivism enhancement, for example, how will a federal court decide whether the ACCA counts a prior conviction that would have qualified for recidivism enhancement if the state prosecutor had not failed to give timely notice? See, e.g., Commonwealth v. Fernandes, 430 Mass. 517, 522, 722 N. E. 2d 406, 409 (1999) (noting longstanding rule that the indictment must give notice of prior convictions “that may subject the defendant to enhanced punishment”).

I could go on, but this is enough to show that the Court’s interpretation promises that ACCA courts will face highly complicated enquiries into every State’s or Territory’s collection of ancillary sentencing laws. That is an unconvincing answer to the ambiguity.


At the end of the day, a plainly superior reading may well be elusive; one favoring the Government certainly is. It does not defy common English or common sense, after all, to look at a statute with one penalty range for the basic crime and a higher one for a repeat offender and say that the former setsthe maximum penalty for the “offense”; but neither is it foolish to see the “offense” as defined by its penalty, however that is computed. What I have said so far suggests that I think the basic-crime view of “offense” is the better one, but I will concede that the competing positions are pretty close to evenly matched. And on that assumption, there is a ready tie-breaker.

The interpretation adopted by both the District Court and the Court of Appeals is the one counseled by the rule of lenity, which applies where (as here) we have “ ‘ seiz[ed] every thing from which aid can be derived,’ ” but are “left with an ambiguous statute,” United States v. Bass, 404 U. S. 336, 347 (1971) (quoting United States v. Fisher, 2 Cranch 358, 386 (1805) (opinion of the Court by Marshall, C. J.)). The rule is grounded in “ ‘the instinctive distaste against men languishing in prison unless the lawmaker has clearly said they should,’ ” Bass, supra, at 348 (quoting H. Friendly, Benchmarks 209 (1967)), and we have used it to resolve questions both about metes and bounds of criminal conduct and about the severity of sentencing. See Bifulco v. United States, 447 U. S. 381, 387 (1980) (collecting cases). “This policy of lenity means that the Court will not interpret a federal criminal statute so as to increase the penalty that it places on an individual when such an interpretation can be based on no more than a guess as to what Congress intended.” Ladner v. United States, 358 U. S. 169, 178 (1958) .

This is why lenity should control here. Even recognizing the best that can be said for the Government’s side, its position rests on debatable guesswork to send a man to prison for 180 months, as against 92 months on the basic-crime view. And the District Courts will be imposing higher sentences more than doubling the length of the alternative in a good many other cases, as well.

The “fair warning” that motivates the lenity rule, McBoyle v. United States, 283 U. S. 25, 27 (1931) (opinion of the Court by Holmes,J.), may sometimes be a benign fiction, see R. L. C., 503 U. S., at 309 (Scalia, J., concurring), but there is only one reading of this statute with any realistic chance of giving fair notice of how the ACCA will apply, and that is the reading the District Court and the Court of Appeals each chose. Their choice should be ours, too.


1 Indeed, if today’s decision is read to mean that enhancements only for recidivism need to be counted, then it too permits a defendant’s actual sentence for a predicate conviction to be higher than what a federal court identifies as an offense’s “maximum term” for ACCA purposes: actual sentences can outstrip the maximum term for recidivists if nonrecidivism factors such as weapons enhancements can also raise a given defendant’s statutory ceiling. The Government seems to accept this possibility, noting that “if a statute is as a formal matter structured in such a way as to create broad tiers of punishment for categories of offenders” based on factors other than recidivism, “then certainly that would seem to be an alternative maximum term of imprisonment.” Tr. of Oral Arg. 21. The Court, however, does not address this prospect, despite having seen the same kind of result as a dealbreaker for Rodriquez’s view.

2 Even adopting the “alternative” of accounting for an offender’s circumstances and record does not resolve the ambiguity, for this rubric actually comprises multiple possibilities under its generic umbrella. Most simply, it might be thought to refer to the actual offender’s sentencing range as applied by the state court. At the other extreme, it might mean the maximum for a purely hypothetical “worst” offender who incurs all possible add-ons. Or perhaps it means a fictional version of the actual offender, say, one qualifying for some statutory add-ons but not for any guidelines rules (as the Court would have it); or maybe one who qualifies for both the statutory and the guidelines departures for which the actual offender was eligible, even though not all of those departures were applied by the state court. This menagerie of options would be multiplied, if a court directly confronted the choice whether to count enhancements for offender-based factors other than recidivism, and if so, which.

3 In this case, doing so would likely result in affirmance, because as the Government admits, Rodriquez’s guidelines ceiling was just shy of five years. Brief for United States 28.

4 See Ariz. Rev. Stat. Ann. §13–604(B) (West Supp. 2007) (maximum set at 13 years); Iowa Code §§124.401(1)(d), 902.9(5), 124.411 (2005) (basic-offense maximum is tripled to 15 years); Utah Code Ann. §§58–37–8(1)(b)(ii) (Lexis 2007 Supp. Pamphlet), 76–3–203(2) (Lexis 2003) (15 years); D. C. Code §§48–904.01(a)(2)(B) (2007 Supp. Pamphlet), 48–904.08(a) (2001) (basic offense maximum is doubled to 10 years).

5 See Cal. Health & Safety Code Ann. §11360 (West 2007); Cal. Penal Code Ann. §1170.12(c)(1) (West 2004) (basic-offense maximum is doubled to 8 years); Mich. Comp. Laws Ann. §§333.7401(2)(d)(iii) (West Supp. 2008), 333.7413(2) (basic-offense maximum is doubled to 8 years); N. Y. Penal Law Ann. §§221.55 (West 2001), 70.70(3)(b)(ii) (West Supp. 2008) (maximum set at 8 years).

6 See Ariz. Rev. Stat. Ann. §§13–3405(B)(5), 13–701(C) (West 2001) (maximum set at 3.5 years); Cal. Health & Safety Code Ann. §11360(4 years); D. C. Code §48–904.01(a)(2)(B) (5 years); Iowa Code §§124.401(1)(d), 902.9(5) (5 years); Mich. Comp. Laws Ann. §333.7401(2)(d)(iii) (4 years); N. Y. Penal Law Ann. §§221.55, 70.70(2)(a)(ii) (5.5 years); Utah Code Ann. §§58–37–8(1)(b)(ii), 76–3–203(3) (5 years).