The quotation above from Justice Scalia, perhaps the most prominent proponent of legal formalism in the United States, articulates an intuitive connection between formalism and judicial constraint: to the extent judges are limited to the rote application of existing rules, they are prevented from deciding cases according to “their own predilections” or “preferences”. Justice Scalia’s approach implies that formalist principles are both sufficient and necessary to limit judicial discretion. In this chapter, I argue that they are neither.

For the purposes of this chapter, we can understand legal formalism as the idea that judges and other decisionmakers should decide particular cases, to the extent possible, by the mechanical application of existing legal rules. Those existing rules might come from constitutional provisions, from statutes, from treaties, from administrative regulations, or from the decisions of prior courts. To the extent a judge’s decision of a case is dictated solely by the content of an existing rule, that decision is not determined by anything else, including the values or beliefs or preferences of the judge herself. There is, then, a quite literal relationship of semantic or logical entailment between pure legal formalism and judicial constraint. A purely formalist decision is an entirely constrained one—a decision constrained completely by the content of the existing rule being applied.

Why might constraining judges and other point-of-application decisionmakers be a good idea? Judges constrained by formalism will not always produce the best possible decisions; sometimes the rule being applied will be a bad rule, and sometimes even the application of generally good rules will produce bad results, as Aristotle understood.² There may nonetheless be strong reasons of rule-consequentialism to require judges to apply existing rules even where the judge believes the result would be bad.³ No doubt there also are “rule of law” reasons such as predictability and consistency.⁴ In this chapter, however, I want to focus on the sorts of reasons to which Justice Scalia primarily appeals in his defense of legal formalism: reasons of democratic legitimacy.

As Cass Sunstain has written:

Justice Scalia is a democrat in the sense that much of his jurisprudence is designed to ensure that judgments are made by those with a superior democratic pedigree. Above all, he seeks to develop rules of interpretation that will limit the policymaking authority and decisional discretion of the judiciary, the least accountable branch of government.⁵

The American judiciary—in particular, the federal judiciary—is “the least accountable branch of government” in Justice Scalia’s view because its members serve during “good behavior” (that is, in most cases, for life or until they choose to retire),⁶ thus insulating their decisions from electoral accountability. For Justice Scalia, it is significant that the American judiciary appears to be saliently less accountable, politically speaking, than most of the institutions whose rules it is charged with interpreting and applying: the legislature, state or federal (whose members are directly elected by the voters); the administrative bodies whose chief policymakers typically are appointed and confirmed by elected officials; and the constitutional framers, whose efforts were channeled through an extraordinarily deliberative and participatory political process.⁷

The view that Justice Scalia represents therefore implies judicial constraint from legal formalism and democratic legitimacy from judicial constraint. For purposes of my arguments here, I will grant the premise that some meaningful degree of constraint upon judicial discretion is necessary for the proper functioning of democratic governance. But I will contend that the most important sources of this constraint lie in principles other than those urged by legal formalists.

In order for legal formalism to work as advertised—to constrain judges—it must be capable of doing so. Judges must actually be able to decide most or all cases primarily or exclusively by the mechanical application of existing legal rules. And whether or not judges are deciding cases in this way must be transparent to others—to the litigants, to superior judges (if any) in the hierarchy, to policymakers, to lawyers, to legal academics, perhaps to the media and to the public—if the constraint imposed by legal rules is to be real rather than merely professed. If it is not clear in most cases whether judges are in fact simply applying existing rules, then judges often will be able to avoid detection in not doing so, thus substantially reducing their incentive to simply apply existing rules.

We can illustrate why using an example that will be familiar to students of Anglo-American jurisprudence, arising as it does from a well-known mid-twentieth-century debate between the English legal positivist H.L.A. Hart and the American “Legal Process” theorist Lon Fuller.⁸ Suppose a city ordinance prohibits “vehicles” in the public park. A group of war veterans wants to erect a monument in the park featuring a now-inoperative truck once used in combat. Would this violate the ordinance?

Note, first of all, that the applicable text of the ordinance—banning “vehicles” from the park—will not, standing alone, answer this question. As Hart saw it, any word has both “a core of settled meaning” and “a penumbra of debatable cases in which [the word is] neither obviously applicable nor obviously ruled out”.⁹ For the latter category, Hart had in mind examples like whether “bicycles, roller skates, or toy automobiles” qualified as “vehicles” under the ordinance.¹⁰ Fuller, for his part, was skeptical of the existence even of “a core of settled meaning”; he used the war-monument example to suggest that even something we normally would call a “vehicle” without a second thought can, in some instances, fall within Hart’s “penumbra” of ambiguity.¹¹ Both theorists understood, however, that there will be many cases in which the proper application of a rule’s text to the facts of a particular case will be uncertain. As Hart put it, “the toy automobile” (or for that matter the inoperative combat truck) “cannot speak up and say, ‘I am a vehicle for the purpose of this legal rule’”.¹²

The hypothetical “no vehicles in the park” ordinance thus is an example of an under-determinate textual expression of a legal rule. Based upon the text alone, a judge faced with a case involving the war-monument combat truck could reasonably reach either alternative conclusion about the meaning of the ordinance: that the truck is a “vehicle” and thus is barred from the park, or that the truck is not a “vehicle” and thus is permitted. Neither interpretation would clearly be an unfaithful application of the text of the rule. And thus the judge is not constrained by the text alone to reach one interpretation rather than the other.

Nor is the “vehicles” example anomalous as a representation of textual indeterminacy in American law (or, I suspect, in the law of any reasonably complex legal system). In the context that most concerns Justice Scalia—American constitutional law—vague or ambiguous text is more the rule (as it were) than the exception. The “penumbras” of terms like “the equal protection of the laws” in the Fourteenth Amendment, “liberty” and “due process of law” in the Fifth and Fourteenth Amendments, and “the freedom of speech” in the First Amendment are considerably larger than that of the word “vehicles” in Hart’s hypothetical ordinance.

Even relatively determinate legal texts have instances of indeterminacy. Justice Scalia himself¹³ discusses an example of this (although he denies that it is an example): a federal statute mandating increased jail time for a defendant who “uses … a firearm” “during and in relation to… [a] drug trafficking crime”.¹⁴ Suppose a defendant offers to trade an unloaded gun for drugs; has he “use[d] … a firearm” in the sense meant by the statute? In Smith v. United States, the Supreme Court answered yes; Justice Scalia dissented on the ground that the decision was inconsistent with the “ordinary meaning” of the phrase “uses a firearm”.¹⁵ The Justice won two other votes with his dissent; a six-Justice majority disagreed with his interpretation of the statute’s text. That the text was susceptible to at least two reasonable, and mutually exclusive, interpretations seems obvious from the nonunanimous vote.

Add to this the fact that legal rules rarely stand alone in any legal system. Often they interact with other legal rules that apply in particular cases. For example, in a recent lawsuit challenging the validity of a federal law refusing recognition to same-sex marriages, the Supreme Court had to interpret not only the statute being challenged, but also the Equal Protection Clause of the federal Constitution, the Due Process Clause of that Constitution, and many prior judicial decisions applying those clauses.¹⁶ And it is virtually always true that legal rules are situated within a larger matrix of rules that, while not directly applicable to the case at hand, nonetheless may influence which interpretations of the applicable rule are most reasonable. Calling an inoperative war-memorial truck a prohibited “vehicle” will seem more or less reasonable to the extent it is consistent or inconsistent with, for example, the use of the word “vehicle” in other ordinances, or with a separate ordinance promoting “natural” uses of the park. A judge applying the indeterminate text of any given legal rule therefore often must also apply the indeterminate text of other rules and determine how the rules interact with each other.

So, in many cases within a complex legal system like that in the United States, the text of legal rules standing alone will be indeterminate, or at least will not be transparently determinate. Judges will be able to reach more than one reasonable result that is consistent with the text. This is not to say that text does not constrain at all in such cases; often there will be many applications of the text clearly within (or outside) its “core of settled meaning”. An operative combat truck driven through the park clearly would violate the “no vehicles” ordinance; a toddler’s small wooden pull-toy clearly would not. Nor is it to deny that some textual expressions of rules will be more determinate than others, or even that some will be determinate in every or nearly every case (consider a law setting the speed limit at 65 mph, or the provision of the US Constitution composing the Senate of “two Senators from each State”). The point is only that there will be a great many cases in which the text does not completely constrain judges. In many cases, the formalist search for a criterion of decision “quite separate from the preferences of the judge himself” (in Scalia’s words) will have to extend beyond the text.

If text alone rarely can underwrite formalism, then what? Among American legal formalists, the answer is almost always some form of originalism. Originalists seek to supplement indeterminate legal texts with empirical facts about the process of the text’s adoption: the “intent” of the authors or ratifiers of the text, or—in the currently dominant form of originalism—the “meaning” that would have been attributed to the text by a reasonable member of the public at the time of its adoption as law. Justice Scalia, a leading originalist, was an early proponent of the latter approach.¹⁷ This “original meaning” or “original public meaning” originalism superseded its “original intent” progenitor in part because the former was exposed as highly indeterminate,¹⁸ and so I will focus on the supposedly more-determinate recent version here.

The problems with this aspiration, however, are threefold. Often an original public meaning will not exist, at least not at the level required to decide a concrete case. Sometimes, to the contrary, there will be multiple inconsistent original public meanings. And locating a single original meaning, assuming it exists, frequently will be difficult or impossible, especially for judges, who are not in fact linguistic historians.

Often an original public meaning will not exist in the required sense because, at the time the text was adopted as law, nobody considered what that text might mean in a given circumstance. Sometimes nobody at the time of adoption could have considered the question. For example, when the Free Speech Clause became part of the Constitution in 1791, no one could have thought about its potential application to campaign-finance laws in the late-twentieth and early twenty-first centuries. It would have been impossible to anticipate what American political campaigns would look like two centuries later—the existence of television and the Internet, the dominance of two entrenched political parties, the system of primary elections, the importance of independent “political action committees”, the rise of for-profit corporations. To ask a reasonable and well-informed citizen in 1791 whether restrictions on so-called “issue advertisements” funded by corporations or unions would “abridge[e] the freedom of speech” would be to ask a nonsensical question. The words of the Free Speech Clause simply have no “original meaning” with respect to that question.

On the other hand, sometimes the evidence will suggest contradictory original meanings. Within a few years after ratification of the Bill of Rights, the Federalist Congress enacted the Sedition Act, which made it a federal crime to “write, print, utter, or publish … any false, scandalous and malicious writing or writings against the government of the United States … with intent to defame the said government … or to bring [it] … into contempt or disrepute”.¹⁹ Many citizens at the time—among them James Madison, a key delegate at the 1787 Constitutional Convention and the principal author of the Bill of Rights—argued that the Sedition Act was unconstitutional, either as a violation of the Free Speech Clause of the First Amendment or as an encroachment on the powers of the State governments.²⁰ Other members of the founding generation, including Alexander Hamilton and, quite obviously, a majority of the Congress that passed the Sedition Act, disagreed (though Hamilton himself did so halfheartedly).²¹ What then was the “original meaning” of the Free Speech Clause with respect to “defamation” of the government, or the original meaning of the Tenth Amendment, which “reserved to the States” all powers not “delegated” by the Constitution to the federal government? The answer depends on whether one credits the stated views of the (mostly Federalist) supporters of the Sedition Act or those of its (mostly Republican) opponents. Here also, it is far too simplistic to assign a single “original meaning” to these provisions, which by all evidence were contested from the moment of their origin.

Finally, consider the practical difficulties facing the judge seeking to identify original meaning even where it exists. The judge must, first, locate relevant historical evidence regarding what the appropriate collection of people thought the text meant (and in so doing must decide what evidence is relevant, which collection of people is appropriate, and what understandings or beliefs or other mental states of those people matter). She must then determine whether some of the relevant mental states of some of the people in question are in conflict and, if so, how to resolve that conflict. And she must, finally, figure out how to apply those historical mental states to the potentially very different and unforeseen facts as they exist today. As Justice Scalia himself puts it:

[I]t is often exceedingly difficult to plumb the original understanding of an ancient text. Properly done, the task requires the consideration of an enormous mass of material … Even beyond that, it requires an evaluation of the reliability of that material … And further still, it requires immersing oneself in the political and intellectual atmosphere of the time – somehow placing out of mind knowledge that we have which an earlier age did not, and putting on beliefs, attitudes, philosophies, prejudices and loyalties that are not those of our day. It is, in short, a task sometimes better suited to the historian than the lawyer.²²

All of which raises the obvious question: are we really constraining judges—who typically are not, after all, historians—in any meaningful way by asking them to search for original meaning in deciding contemporary cases? Many originalist theorists have been forced to admit that the answer is no. Randy Barnett and Lawrence Solum, two leading contemporary originalists, thus endorse what they call “constitutional construction”—essentially, discretionary judicial lawmaking—as a means of filling the many gaps in originalist interpretation.²³ In Barnett’s words:

For better or worse, … the US Constitution requires more than originalist interpretation to be applied to cases and controversies. Owing to the vagueness of language and the limits of historical inquiry, originalist interpretation may not result in a unique rule of law to be applied to a particular case or controversy.²⁴

Original meaning, like the text, is endemically under-determinate. And while the problem may be most egregious in the context of constitutional provisions, which typically “use general concepts and abstract principles in place of specific rules”,²⁵ it will appear to some extent in the interpretation of any legal text.

The assumption that formalist techniques can meaningfully constrain judges is therefore deeply problematic. In this and the following two sections, I argue that this is not a great cause for concern—because the assumption that formalism is necessary to meaningfully constrain judges also is flawed. American judges are not constrained by the formalistic interpretation of legal rules. But they are constrained by the internal dynamics of the adversary system of adjudication; and they are constrained by the need to keep faith with substantive legal principles, even when those principles are not reducible to formalistic rules. I discuss the former source of constraint in this section and the latter in Sect. 5.²⁶

In the United States, as in most legal systems influenced heavily by the British, adjudication follows an “adversary” model: primary responsibility for initiating court cases, framing the factual and legal issues, investigating and proving the relevant facts, advancing the legal arguments, and shaping the remedies lies with the affected parties and their attorneys, not with the court itself (I will focus the discussion here on civil disputes rather than criminal prosecutions, but most of the same party-driven dynamics apply in criminal cases—often to an even greater degree—although of course one of the parties to a criminal case is the government).

In American civil adjudication: