What Should Fair Notice Require in Criminal Statutes?

by Michael Dorf

Last week, I wrote about a Missouri Supreme Court opinion that construed that state's theft statute in a way that resulted in most instances of stealing being classified as misdemeanors rather than felonies. I criticized the opinion as an instance of "textualism run amok." In addition to some comments agreeing with my analysis, there were three lines of criticism.

One group of critics said that the Missouri Supreme Court ruling did not discredit textualism because it was an instance of bad textualism. Any approach to statutory construction can be used badly, the argument goes, and so it was unfair of me to hang this particular albatross around the neck of textualism. My post anticipated this response by noting that some textualists allow for less literal readings of statutes when literal readings produce terrible results, but that even so, the gravitational pull of textualism will make this option less readily available than it ought to be.

A second group of critics disagreed with my particular alternative reading of the relevant Missouri statute, with some of them proposing their own alternatives. Perhaps these critics are right. I gave very little thought to my alternative, because my aim was to criticize the Missouri Supreme Court for not even trying to come up with an alternative reading that might make sense of the text.

Here I want to focus on a point made by a third group of critics. They said that the Missouri Supreme Court was right because the case involved a criminal law, and the requirement of fair notice (which is an element of due process) forbids creative statutory construction to the detriment of a criminal defendant. In this view, even if it might be appropriate to depart from the literal text to achieve a sensible result in a civil context, where the possible result is the defendant's loss of liberty through the criminal law, a more rigid textualism should prevail. Indeed, under the so-called rule of lenity, ambiguity should be construed in favor of the defendant; hence, if the statute on its face favors the defendant, then certainly it should not be construed against him.

I want to concede that this third group of critics has a point, insofar as they are characterizing existing doctrine. But I think that is largely a cost of the rule of law, rather than a feature to be celebrated--at least with respect to criminal acts that are obviously wrongful, like theft of high-value items.

Let's grant that no one should be held criminally liable for acts that he was not on notice were criminal. And as a corollary, let's grant that if there is notice only that an act is a misdemeanor, then it violates the requirement of fair notice to punish someone for a felony if she commits that act. Conceding all of that--and conceding, at least for the sake of argument, that the Missouri law did not give clear notice that stealing of firearms was a felony--it still strikes me as a cost of the rule of law that the defendant got off.

Why? For the simple reason that knowledge of the law is a fiction. The law does not, after all, require actual notice. Instead, it presumes that people know the law. The adage, stated in the contrapositive, is that ignorance of the law is no excuse. In the Missouri case, it is extremely unlikely that the defendant committed her crime because she thought that it was only a misdemeanor rather than a felony. It's extremely unlikely that she knew what the law defining stealing said. Do you know the details of the definition in your state? I don't in mine.

Nonetheless, let's concede that a defendant who timely raises an objection that the law doesn't clearly define her alleged conduct as criminal (or as a serious crime) should be able to succeed in that objection. How far must the fiction that everyone knows the law be pushed? Let me suggest a couple of limitations.

First, I think the Missouri Supreme Court could have been creative with respect to prospective remedy. Case law states that a law that is unclear on its face may be made sufficiently clear to satisfy notice requirements through a judicial construction--at least prospectively. Accordingly, the Missouri Supreme Court could have said that it was giving the defendant the benefit of its construction of the law, but that henceforth it would construe the law broadly. The court might even have said that it would accelerate the coming into force of the new--quite clear--version of the law, currently scheduled to come into force on January 1, 2017, making it immediately operative.

If that sounds unprecedented, consider the 1985 judgment of the Supreme Court of Canada in the Manitoba Language Rights Case. The 1870 Manitoba Act required that all laws be published in French as well as English. However, they were only published in English. The Court accordingly held that all of Manitoba's laws were invalid. But rather than plunge Manitoba into a state of anarchy, the Court gave the province a grace period to translate and publish its laws. During the grace period, the old laws (just in English) remained in effect. I am proposing that the Missouri Supreme Court could have accomplished something similar by accelerating the effective date of the amended criminal law.

Now it could be objected that there is a difference between leaving an old law on the books and putting a new one on. But note that both involve judicial creativity. And because in my proposal for Missouri, the new law only applies to conduct occurring after the court's decision, there is no notice problem.

Second, I think that another way to cabin the damage from recognizing that clearly harmful conduct falls into a gap in the criminal law is to limit its availability based on waiver. Earlier this year, in Welch v. United States, the SCOTUS ruled that a "new rule" that a statute fails to give fair notice (in that case because it was vague) is substantive, and therefore can be used to challenge a criminal conviction in a collateral proceeding. For readers who are not fed courts nerds, the background idea is that when the Supreme Court announces a new rule it is typically not available as the basis for habeas corpus relief; however, there is an exception for new rules that make the underlying conduct uncriminalizable. Recently, the Court has somewhat expanded this exception. Thus, insofar as these principles apply in state courts (and Welch was a case involving federal habeas for a federal prisoner, so its applicability is not entirely clear), one might think that a rule like the one announced by the Missouri Supreme Court would also be the basis for relief for people whose convictions have already become final.

But I think that such a conclusion would over-read Welch (even assuming it applies in state court). Welch and related cases say that the newness of the rule is not a bar to asserting it in collateral proceedings. However, it does not speak to other possible limits. One such limit is the ordinary requirement that a defendant raise an objection at trial or else be deemed to have waived it. Now there are quite a few circumstances in which such "procedural defaults" can act harshly. But I don't think that failure to object to a lack of fair notice is one of them. A defendant who didn't know that what she did was a crime--or didn't know that it was a serious crime--is likely to tell that to her lawyer, who, in turn, is likely to raise the absence of fair notice as grounds for quashing the indictment if there is a colorable argument for that result. Thus, where a defendant did not raise such a claim, it is fair to assume that's because the defendant was not relying on the terms of the statute in attempting to conform her conduct to law. And while that's true in nearly every case--because of the fictive quality of notice--it seems reasonable for the legal system to cut its losses by not engaging the fiction for old cases where the social cost of freeing people who committed clearly bad acts is high.

Finally, I want to be clear that I mean the foregoing discussion of non-retroactivity to apply only to crimes (such as most stealing) that we can clearly classify as malum in se, and not to acts that are not clearly wrongful (such as possessing marijuana) but are nonetheless illegal. I realize that there are people who dispute the distinction between malum in se and malum prohibitum. If you count yourself among those people, then you may have good reason to disagree with my proposal with respect to non-retroactivity.