In anticipation of oral argument next week in Sykes v. United States, I’ve been thinking about the “categoric approach” the Court has been using to decide whether a prior conviction counts as a “violent felony” under the Armed Career Criminal Act. (See my post here for background.) The categoric approach requires an examination of the formal elements of the crime of conviction and a consideration of whether an offense so defined would typically (in the words of the statute) “involve conduct that presents a serious potential risk of physical injury to another.” The alternative would be to consider whether the specific offender’s conduct — as opposed to a typical offender’s conduct – presented a “serious potential risk.”
The statutory language itself strikes me as ambiguous on this point. Normally, I would favor falling back on the rule of lenity to resolve ambiguities in criminal statutes, but it is not clear to me which way lenity cuts in this setting. Some defendants would be better off if their own conduct were the focus of the inquiry, while other defendants would benefit from a focus on the typical case.
When the Court first embraced the categoric approach in Taylor v. United States, 495 U.S. 575 (1990), the Court relied on legislative history that seemed to indicate Congress was contemplating the use of generic offense categories.
Although the Court’s reading of the legislative history seems fair enough, I am not keen on the use of legislative history as a statutory interpretation device, especially where the legislative history does not indicate a deliberate choice to go one direction or another. While the chief proponents of the 1986 ACCA revision that produced the current statutory language may have had certain offense categories foremost in their minds, I don’t see anything in the Taylor discussion of legislative history that squarely rejects the case-by-case approach in favor of the categoric approach.
More recently, in Shepard v. United States, 544 U.S. 13 (2005), the Court suggested an alternative rationale for the categoric approach: an inquiry by the sentencing judge into the particulars of a defendant’s conduct in a prior case might run afoul of the Apprendi rule against increasing sentencing exposure on the basis of judicial fact-finding. I’m not much impressed by this argument. If the case-by-case approach triggers a right to jury fact-finding, so be it. Sure, it would be more burdensome for the government to get the fifteen-year minimum imposed, but it should be burdensome for the government to impose such harsh punishment. I don’t see much appeal in interpreting a statute in a particular way so as to permit the government to take advantage of a loophole in the constitutional right to a jury trial. I understand that the avoidance canon teaches that we should avoid interpretations that might render a statute unconstitutional, but we are not talking about the constitutionality of the ACCA here, only about the procedures by which it must be implemented.
The case-by-case approach also raises the possibility that the government will effectively be forced to retry many old cases in collateral proceedings to determine the actual circumstances of the defendant’s conduct. I’m not sure how serious a problem this would be. I suspect that the paper record would be adequate in many cases to establish that the defendant’s conduct was dangerous, and that in many other cases the facts of the earlier conduct could be resolved through negotiated stipulations. If the facts of an earlier case are sufficiently uncertain and disputed that a trial is necessary to sort them out, then we should want to see a trial rather than send a defendant away for fifteen years on the basis of speculation about the risks presented by a hypothetical “typical” case. And if the government is unwilling to invest the resources necessary to litigate the significance of a defendant’s criminal history, that is a pretty good sign that the defendant is not worth incarcerating for fifteen years.
Finally, if we are still concerned about transaction costs, the categoric approach has in practice hardly been a model of efficiency. The categoric approach requires careful linguistic parsing of the offenses defined in 51 sprawling, ever-expanding criminal codes, as well as the collection and analysis of empirical data regarding the dangerousness of each offense. The categoric approach has given rise to many circuit splits already and (as Justice Alito predicted in Chambers) seems likely to keep the Supreme Court busy for many years to come.
Although I don’t find the arguments for the categoric approach especially persuasive, the case-by-case approach admittedly has its difficulties, too. Still, in the end, what I find more attractive about the case-by-case approach is that it better focuses the inquiry on what we should really care about: not whether a hypothetical typical offender is dangerous, but whether this particular offender is dangerous. Fifteen years is a long time in prison, and I find it quite troubling that we might order such a long sentence without ever peeking beneath the labels on the defendant’s rap sheet to see what he really did. The current ACCA jurisprudence seems cold and detached from reality — reading the cases, I’m often reminded of the stereotype of the old scholastics debating how many angels could fit on the head of a pin. The case-by-case approach would give us a more principled jurisprudence and diminish the sense that those fifteen-year sentences are being handed out on the basis of highly abstract distinctions that no layperson could possibly think were important.