abstract. In Hughes v. United States, the Supreme Court will revisit a thorny question: how to determine the precedential effect of decisions with no majority opinion. For four decades, the clearest instruction from the Court has been the rule from Marks v. United States: the Court’s holding is “the position taken by those Members who concurred in the judgments on the narrowest grounds.” The Marks rule raises particular concerns, however, when it is applied to biconditional rules. Biconditionals are distinctive in that they set a standard that dictates both success and failure for a given issue. More formulaically, they combine an if-then proposition (If A, then B) with its inverse (If Not-A, then Not-B).

Appellate courts on both sides of the circuit split that prompted the grant of certiorari in Hughes have overlooked the special features of biconditional rules. If the Supreme Court makes the same mistake, it could adopt a misguided approach that would unjustifiably create binding law without a sufficient consensus among the Justices involved in the precedent-setting case. This Essay identifies these concerns and proposes ways to apply Marks coherently to nonmajority opinions that endorse biconditional rules.

Introduction

For forty years, this instruction from Marks v. United States has governed how to identify the holding of a Supreme Court decision that lacks a majority opinion: “When a fragmented Court decides a case and no single rationale explaining the result enjoys the assent of five Justices, ‘the holding of the Court may be viewed as that position taken by those Members who concurred in the judgments on the narrowest grounds . . . .’”1 The Supreme Court has recognized, however, that “pursu[ing] the Marks inquiry” has often “baffled and divided the lower courts that have considered it.”2

That may soon change. The Court has granted certiorari in Hughes v. United States, which explicitly asks the Court to clarify the Marks rule.3 Hughes involves how to interpret the Supreme Court’s 2011 decision in Freeman v. United States, in which the Court split four-one-four regarding when certain defendants are eligible to seek a sentence reduction based on a retroactive lowering of the Federal Sentencing Guidelines.4 Five Justices in Freeman found that the defendant was eligible to seek a reduction, with a four-Justice plurality embracing one rationale and a concurring Justice embracing a different rationale.5 Four Justices dissented, finding that the defendant was ineligible.6

Thanks to decades of judicial and scholarly engagement with the Marks rule, the Court in Hughes will be able to choose from a range of options for determining the precedential effect of decisions with no majority opinion.7 Under one approach, a nonmajority opinion is binding only when its analysis is the “logical subset” of the other opinions supporting the Court’s ultimate result; if no such logical-subset relationship exists, there is no controlling opinion under Marks.8 Under a different approach, Marks creates binding law as to the area of “shared agreement” or “partial overlap” between the nonmajority opinions supporting the Court’s ultimate judgment.9 Finally, the Marks analysis might include the rationales of dissenting Justices, so long as those rationales combine with enough Justices in the majority to garner five or more votes.10

Whichever path the Court takes, it will be crucial to appreciate the particular problems posed by biconditional rules. Biconditionals are distinctive rules in that they set a standard that dictates both success and failure for a given issue. More formulaically, biconditionals combine a conditional proposition (If A, then B) with its inverse (If Not-A, then Not-B).11 Because biconditionals are two rules rolled into one, they complicate the inquiry into which of two or more opinions provides the “narrowest grounds” under Marks.

Appellate courts on both sides of the circuit split that prompted the Court’s grant of certiorari in Hughes have misunderstood the nature of biconditional rules. Specifically, courts have assumed that Justice Sotomayor’s concurring opinion in Freeman would be binding under Marks if it would make a narrower universe of defendants eligible for a sentence reduction than Justice Kennedy’s Freeman plurality opinion would.12 But, as a matter of logic, this assumption is mistaken.

A key disagreement in the lower courts—and between the parties in Hughes—is whether the concurring opinion truly is narrower in the sense that it makes fewer defendants eligible for relief. The defendant in Hughes argues that the Freeman concurrence is not narrower; instead, he maintains that it makes some defendants eligible who would be ineligible under the plurality’s approach.13 The government contests this interpretation, arguing that any defendant who would be eligible under the concurrence’s approach would necessarily be eligible under the plurality’s approach, and thus that the Freeman concurrence is narrower.14 Due to some ambiguities in the Freeman plurality opinion, there is indeed some uncertainty on this point.

The dispute on this issue, however, misapprehends the nature of biconditionals. Even if the Freeman concurrence makes a narrower set of defendants eligible for a sentence reduction, it necessarily must make a broader set of defendants ineligible for a sentence reduction. Yet in Hughes, the government is invoking the part of the Freeman concurrence’s biconditional rule that dictates a defendant’s ineligibility. To say that the concurrence’s ineligibility rule is narrower than the plurality’s ineligibility rule turns the narrowest-grounds notion on its head.

If the Supreme Court makes the same mistake, the result could be a reading of Marks that unjustifiably creates binding law despite a lack of sufficient consensus among the Justices involved in Freeman itself. With respect to the particular issue in Hughes, the Freeman concurrence should not qualify as the “narrowest grounds” simply because it would deem fewer defendants eligible to seek a sentence reduction than would the Freeman plurality. As explained below, the only plausible way for the Freeman concurrence to be the Court’s binding holding is to take into account the views of dissenting Justices—an approach that is contrary to the prevailing understanding of Marks. Although the Supreme Court in Hughes might adopt the logic of the Freeman concurrence on its own merits,15 it should not do so based on the misconception that a consensus existed between the Freeman plurality and concurrence in which the concurrence was “narrowest.”

Part I of this Essay describes the Freeman decision and the disagreements in the lower courts that led the Supreme Court to grant certiorari in Hughes. This Essay then discusses how to account for biconditional rules under various potential approaches to the Marks rule. Part II addresses the significance of biconditional rules under an approach to Marks that would inquire whether one opinion is a “logical subset” of another. In particular, it shows why one biconditional rule can never be a complete logical subset of a competing biconditional rule. Part III applies a similar analysis to an approach to Marks that would identify areas of “shared agreement” or “partial overlap” between opinions. Finally, Part IV considers an approach to Marks that would both address biconditional rules and incorporate the views of dissenting Justices. To apply the Marks rule coherently under any approach, the Court must pay particular attention to the special features of biconditionals.

I. from freeman to hughes

This Part first summarizes the Supreme Court’s decision in Freeman. It then discusses the disagreements among circuit judges regarding how to apply the Freeman decision given the lack of a majority opinion.

A. The Supreme Court’s Decision in Freeman

The defendant in Freeman was indicted for possessing crack cocaine with the intent to distribute. He agreed to plead guilty under Federal Rule of Criminal Procedure 11(c)(1)(C), a provision that allows a defendant and the government to agree to a specific sentence or sentencing range, which is then binding on the court if the court accepts the plea agreement.16

After the defendant’s sentencing, the U.S. Sentencing Commission retroactively amended the Federal Sentencing Guidelines to reduce the disparity in penalties between crack- and powder-cocaine offenses.17 The defendant moved for a reduction in his sentence under 18 U.S.C. § 3582(c)(2), which gives the district court discretion to reduce a sentence that was “based on a sentencing range that has subsequently been lowered by the Sentencing Commission . . . .”18 This raised a fundamental statutory question: when is a sentence imposed following an 11(c)(1)(C) agreement “based on” a particular Guidelines range?

By a five-four vote, the Court held in Freeman that the defendant was eligible to seek a reduction in his sentence. Justice Kennedy’s four-Justice plurality opinion reasoned that a judge’s acceptance of an 11(c)(1)(C) plea agreement “is itself based on the Guidelines.”19 He observed that the Guidelines “require the district judge to give due consideration to the relevant sentencing range, even if the defendant and prosecutor recommend a specific sentence as a condition of the guilty plea.”20 Accordingly, the plurality found that a defendant may seek a sentence reduction when the sentence is imposed “in light of the Guidelines”21—that is, when “the sentencing range in question was a relevant part of the analytic framework the judge used to determine the sentence or to approve the agreement.”22 “Even where the judge varies from the recommended range, if the judge uses the sentencing range as the beginning point to explain the decision to deviate from it, then the Guidelines are in a real sense a basis for the sentence.”23 In Freeman, it was clear that the Guidelines were relevant to the judge’s acceptance of the plea agreement, because the district court “expressed its independent judgment that the sentence was appropriate in light of the applicable Guidelines range.”24

Justice Sotomayor’s solo concurring opinion—which provided the fifth vote in favor of the defendant in Freeman—adopted a different approach. She argued that a defendant is eligible for a sentence reduction only if the 11(c)(1)(C) agreement itself is based on a subsequently lowered guideline. This would occur if the plea agreement either “call[s] for the defendant to be sentenced within a particular Guidelines sentencing range” or “make[s] clear that the basis for the specified term is a Guidelines sentencing range applicable to the offense to which the defendant pleaded guilty.”25 She found that the defendant in Freeman was eligible for a sentence reduction because his plea agreement explicitly used a Guidelines range that was retroactively reduced.26

The four-Justice dissenting opinion, authored by Chief Justice Roberts, endorsed a rule that a defendant sentenced pursuant to an 11(c)(1)(C) agreement may never seek a sentence reduction based on a subsequently lowered guideline. Such sentences are “based on the agreement, not the Sentencing Guidelines,” and therefore are not eligible for reduction simply because a Guidelines range is retroactively modified in the defendant’s favor.27

B. Applying Freeman

A circuit split quickly developed over what qualified as Freeman’s holding under the Marks rule, as courts struggled to determine the “position taken by those Members who concurred . . . on the narrowest grounds.”28 Some circuits found that Marks required them to follow Justice Sotomayor’s concurring opinion in Freeman.29 Others found that neither the concurring opinion nor the plurality opinion was the narrowest, and therefore “no opinion in Freeman controls.”30 Those two circuits opted to follow Justice Kennedy’s plurality opinion.31

In applying Marks to the fragmented Freeman decision, judges have disagreed on two issues. The first area of disagreement was the content of the rule that the Freeman plurality adopted. This is significant for purposes of the Marks rule, because to assess which opinion’s “grounds” are “narrowest,” one must identify what each opinion’s “grounds” actually are.

A defendant would certainly be eligible for a sentence reduction under the plurality opinion if the district court considered the subsequently lowered guideline in accepting the plea and imposing the designated sentence.32 Some judges, however, read the plurality’s opinion more broadly. On one such reading, a defendant would “always” be eligible to seek a sentence modification in cases in which an applicable sentencing guideline has been lowered.33 On another reading, any defendant whose Rule 11(c)(1)(C) plea agreement indicates that the sentence was based on a subsequently lowered guideline would be eligible under the plurality’s rule, even if the judge did not rely on that guideline in her decision to approve the plea agreement.34 Some judges have justified this reading based on the plurality’s statement that a defendant would be eligible “if the judge uses the sentencing range as the beginning point” of her analysis.35 Even if the sentencing judge does not rely on the subsequently lowered guideline in approving the plea agreement, that guideline is still at least a “beginning point” insofar as it was the basis of the plea agreement that the judge ultimately approved. Judges adopting this reading have also emphasized the Freeman plurality’s characterization of the concurrence’s approach as an “intermediate position” that would permit a sentence reduction only for a “subset of defendants.”36 Under both this interpretation and the always-eligible interpretation, every defendant who would be eligible for a sentence reduction under the plurality’s rule would necessarily be eligible for a sentence reduction under the concurrence’s rule.37 Thus, judges adopting these interpretations have found that the concurrence’s rule was a “logical subset” of the plurality’s rule.38

Other judges, however, read the Freeman plurality opinion less generously. Because the plurality emphasized that the district court’s acceptance of the plea agreement is the ultimate basis for the sentence, it would arguably find a defendant ineligible for a sentence reduction in the following situation: the plea agreement explicitly relied on a subsequently lowered guideline, the district court found that guideline inapplicable, and the court nonetheless accepted the plea agreement “for reasons unrelated to the guideline range determined by the parties.”39 On this reading, “the set of cases where the defendant prevails under the concurrence is not always nestled within the set of cases where the defendant prevails under the plurality.”40

The second area of disagreement was the methodology for identifying the rule that qualifies as the “narrowest grounds” under Marks. Some judges insisted that Marks applies only when separate opinions supporting the Court’s ultimate decision “share common reasoning whereby one analysis is a ‘logical subset,’ of the other.”41 If there is no logical subset, there is no “controlling opinion.”42 For judges that adopted this methodology, the Marks inquiry hinged on the first area of disagreement. If they found that any defendant who was eligible under the Freeman concurrence was necessarily eligible under the plurality, then they held that the Freeman concurrence was binding under Marks. In that situation, the universe of eligible defendants under the concurrence’s reasoning seemed to be a logical subset of the universe of eligible defendants under the plurality’s reasoning.43 But if judges found that some defendants could be eligible for a sentence reduction under the Freeman concurrence but ineligible under the Freeman plurality, then judges applying a logical-subset approach held that neither opinion was “narrowest” and, therefore, neither opinion was controlling.44

Some judges, however, disagreed with the logical-subset approach. Rather, they argued that Marks makes binding any “legal standard which, when applied, will necessarily produce results with which a majority of the Court from that case would agree.”45 Notably, the views of dissenting Justices can count under this approach—provided their views combine with those of non-dissenting Justices to generate a five-Justice majority.46 On this understanding of Marks, failure to satisfy the Freeman concurrence—because the plea agreement itself did not indicate that it relied on a subsequently reduced sentencing guideline—would be fatal. That one vote against eligibility from the Freeman concurrence would combine with four votes from the Freeman dissenters (for whom all defendants who plead guilty under Rule 11(c)(1)(C) are ineligible) to compel the conclusion that the defendant is ineligible for a sentence reduction.47

In December 2017, the Supreme Court granted certiorari in Hughes v. United States.48 The defendant in Hughes pleaded guilty to a number of drug and firearm offenses, and he sought a reduction in his sentence after the Sentencing Commission retroactively lowered the offense levels for certain drug crimes. The Eleventh Circuit found that the Freeman concurrence was binding under Marks.49 Accordingly, it ruled that the defendant was not eligible to seek a sentence reduction because his plea agreement did not refer to the subsequently lowered guideline.50 The Hughes petition for certiorari explicitly asks the Supreme Court to decide whether the Eleventh Circuit correctly applied the Marks rule.51

II. biconditional rules and logical subsets

As described above, some lower courts have endorsed an approach to Marks that inquires whether one opinion is the logical subset of another. One of the key areas of disagreement between the defendant and the government in Hughes is whether the Freeman concurrence is a logical subset of the Freeman plurality.52 This Part defines the notion of a logical subset, and then turns to the particular challenges of applying a logical-subset approach to biconditional rules like those articulated in the competing Freeman opinions.

A. What Is a Logical Subset?

To clarify what is meant by a logical subset, it is helpful to conceptualize the competing rules in a case like Freeman as if-then propositions, also known as conditional statements.53 Formulaically, the Freeman plurality’s rule can be described as (If P, then X), where P (for plurality) defines the characteristics of cases in which a defendant is eligible to seek a sentence reduction, and X is the result that such a defendant is eligible. Using the formal terminology, P is the antecedent of the conditional statement, and X is the consequent.54

The Freeman concurrence’s rule can be described as (If C, then X), where C (for concurrence) defines the characteristics of cases in which a defendant is eligible to seek a sentence reduction, and X is the result that such a defendant is eligible. For the concurrence, C is the antecedent and X is the consequent.

To say that the concurrence’s rule is a logical subset of the plurality’s rule is to say that the concurrence’s antecedent (C) is a logical subset of the plurality’s antecedent (P).55 That is, the universe of cases that would be eligible for a sentence reduction under the plurality’s rule necessarily includes all of the cases that would be eligible for a sentence reduction under the concurrence’s rule. As discussed above, there is disagreement over whether this logical-subset relationship exists between the Freeman concurrence and plurality.56 It is clear, however, that the focus of this disagreement is on whether every case that satisfies the concurrence’s antecedent would necessarily satisfy the plurality’s antecedent. In other words, the disagreement is whether Justice Sotomayor’s rule would allow any defendants to seek sentence reductions whom the plurality’s rule would not; if so then the antecedent (C) of Justice Sotomayor’s rule (If C, then X)is not a logical subset of the antecedent (P) of the plurality’s rule (If P, then X).

B. The Freeman Opinions as Biconditionals

In Hughes, the government is not invoking the Freeman concurrence’s rule described above (If C, then X). The government is invoking the inverse of that rule (If Not-C, then Not-X). More specifically, the rule that was dispositive for the concurrence in Freeman was:

If a defendant’s plea agreement clearly relies on a subsequently lowered sentencing guideline, then the defendant is eligible to seek a reduction (If C, then X).57

But the government in Hughes is invoking the rule:

If a defendant’s plea agreement does not clearly rely on a subsequently lowered sentencing guideline, then the defendant is not eligible to seek a reduction (If Not-C, then Not-X).

These are distinct principles. Of course, it is not uncommon for a judicial opinion to endorse both a proposition and its inverse simultaneously. When a court does so, it articulates not only an if-then rule but an if-and-only-if rule. An if-and-only-if rule is called a biconditional rule precisely because it combines a conditional rule with its inverse.58 By using a biconditional, a court provides “a comprehensive resolution to the relevant question”59—a test that will determine both success (X) and failure (Not-X).60 We can think of the first rule (If C, then X) as the top half of the Freeman concurrence’s biconditional, and the second rule (If Not-C, then Not-X) as the bottom half of the biconditional.

One can reasonably read the opinions in Freeman as stating biconditional rules. That is, the concurrence and the plurality sought to set standards that would determine both eligibility and ineligibility for a sentence reduction. The Freeman concurrence stated that the defendant’s plea agreement “must use a Guidelines sentencing range that has subsequently been lowered by the Sentencing Commission.”61 The plurality wrote that a sentence reduction is available “to whatever extent the sentencing range in question was a relevant part of the analytic framework the judge used to determine the sentence or to approve the agreement.”62 Recognizing that these are biconditional rules, however, only complicates the logical-subset inquiry. In fact, as the next Section will show, it is logically impossible for one complete biconditional to be a logical subset of another complete biconditional.63

C. A Biconditional Rule Can Never Be a Logical Subset of Another Biconditional Rule

To begin, imagine the following simple example: Charlie and Pam are co-workers. They propose different rules for deciding when they will go out to lunch and when they will eat at the office. They work a typical five-day work week. Charlie wants to go out to lunch only on Friday, and Pam wants to go out to lunch only on Thursday and Friday.

Focus first on the top half of each biconditional. Charlie’s top-half rule is: if it is Friday (C), then we will go out to lunch (X). Pam’s top-half rule is: If it is Thursday or Friday (P), then we will go out to lunch (X). For the top half of the biconditional, C is clearly a logical subset of P. Every time Charlie’s antecedent is satisfied (because it is Friday), Pam’s antecedent is necessarily satisfied.

For the bottom half of the biconditional, however, the opposite is true. Charlie’s bottom-half rule is: If it is Monday, Tuesday, Wednesday, or Thursday (Not-C), then we will not go out to lunch (Not-X). Pam’s bottom-half rule is: If it is Monday, Tuesday, or Wednesday (Not-P), then we will not go out to lunch (Not-X). As to the two bottom-half rules, Pam’s antecedent is the logical subset of Charlie’s. Every time Not-P is true (because it is Monday, Tuesday, or Wednesday), Charlie’s bottom-half antecedent (Not-C) is necessarily satisfied.

This simple example confirms a crucial proposition: one biconditional can never be a logical subset of another biconditional. When the top-half antecedent of Charlie’s rule is a logical subset of the top-half antecedent of Pam’s rule, their roles are necessarily reversed as to their bottom-half antecedents.

This example also shows that it is possible to separate the biconditional into its component parts, and then to identify different logical subsets for the top half and the bottom half. Under a logical-subset approach, Charlie and Pam will go out to lunch on Friday and they will not go out to lunch on Monday, Tuesday or Wednesday. No logical subset exists with respect to Thursday.

Thus, we can only identify logical subsets that truly reflect consensus by looking at both halves of biconditional rules independently. If identifying areas of consensus is the goal, it would have been a mistake to look solely at the top-half of the biconditional. We would have found that Charlie’s rule is the logical subset because “Friday” is a logical subset of “Thursday or Friday,” and we therefore would have declared that they would not go to lunch on Thursday. The right thing to say about Thursday, however, is that no rule controls it.

D. Logical Subsets in Freeman

As mentioned earlier, lower courts have disagreed about whether any logical-subset relationship exists between the Freeman concurrence and the Freeman plurality.64 Let us assume, however, that the Freeman plurality is read so that every time a defendant is eligible for a sentence reduction under the concurring opinion, he or she will necessarily be eligible under the plurality opinion.

Recall that on this reading of the Freeman plurality opinion, the top half of the plurality’s rule would be:

If either the plea agreement clearly relies on a subsequently lowered sentencing guideline or that guideline is otherwise relevant to the judge’s analysis in approving the plea agreement, then the defendant is eligible to seek a reduction. (If P, then X.)

And recall the concurrence’s rule:

If a defendant’s plea agreement clearly relies on a subsequently lowered sentencing guideline, then the defendant is eligible to seek a reduction. (If C, then X.)

In this example, C is a logical subset of P. If the plea agreement clearly relies on a subsequently lowered sentencing guideline (C is true), then it is necessarily true that either the plea agreement clearly relies on a subsequently lowered sentencing guideline or that guideline was otherwise relevant to the judge’s approval of the plea agreement (P is true). The universe of cases where the plurality’s antecedent (P) is satisfied completely encompasses the universe of cases where the concurrence’s antecedent (C) is satisfied.

Now look at the bottom half of the biconditional. For the plurality, the inverse of its rule would be:

If the plea agreement does not clearly rely on a subsequently lowered sentencing guideline and that guideline is not otherwise relevant to the judge’s analysis in approving the plea agreement, then the defendant is not eligible to seek a reduction. (If Not-P, then Not-X).

And recall the bottom half of the concurrence’s rule:

If a defendant’s plea agreement does not clearly rely on a subsequently lowered sentencing guideline, then the defendant is not eligible to seek a reduction. (If Not-C, then Not-X).

Here, Not-P is a logical subset of Not-C. If it is true that both the plea agreement did not clearly rely on a subsequently lowered sentencing guideline and the subsequently lowered guideline was not otherwise relevant to the judge’s analysis (Not-P is true), then it is necessarily true that the plea agreement did not clearly rely on a subsequently lowered sentencing guideline (Not-C is true). As to the bottom half of the biconditional, then, the plurality’s rule is the logical subset of the concurrence’s rule—not the other way around.

This is precisely the problem illustrated by the example of Charlie and Pam’s lunch preferences examined above. Focusing on the bottom half of the biconditional (that is, the rules that dictate the conclusion that the defendant is not eligible for a sentence reduction), it cannot be said that the Freeman concurrence is the logical subset of the Freeman plurality. This holds even if—as the government argues in Hughes65—every defendant who would be eligible under the concurrence’s approach would also be eligible under the plurality’s approach.

E. Applying a Logical-Subset Approach to Biconditionals

While it is impossible for one biconditional rule to be a logical subset of another, there are two ways one might employ a logical-subset approach to the Marks rule in the context of biconditionals.

Option 1: There is binding precedent only as to one half of the biconditional.

One possibility is to read a fragmented case as setting binding precedent only as to the half of the biconditional that was dispositive in that case. With respect to Freeman, this approach would work as follows. Assume again that the Freeman plurality is broad enough that, every time a defendant is eligible for a reduction under the Freeman concurrence, he or she is necessarily eligible for a reduction under the Freeman plurality. This would mean that the top half of the Freeman concurrence is the logical subset of the top half of the plurality. On this approach, Freeman would establish the following proposition from Justice Sotomayor’s concurrence: If a defendant’s plea agreement clearly relies on a subsequently lowered sentencing guideline, then the defendant is eligible to seek a reduction (If C, then X). That would be the “narrowest grounds” for purposes of Marks, and future courts would be bound to accept that rule.

Freeman would not establish any binding law, however, with respect to whether a defendant is eligible in cases where the plea agreement does not clearly rely on a subsequently lowered sentencing guideline (Not-C). The simplest example of this would be the area of disagreement between the plurality and the concurrence—where the plea agreement does not refer at all to the subsequently lowered guideline, but that guideline is relevant to the judge’s analysis in accepting the plea agreement. In that situation, future courts would have the leeway to adopt the more defendant-friendly top-half rule embraced by the Freeman plurality: if either the plea agreement clearly relies on a subsequently lowered sentencing guideline or that guideline is relevant to the judge’s analysis in approving the plea agreement, then the defendant is eligible to seek a reduction (If P, then X). Or they might adopt the bottom half of the concurrence’s rule: If a defendant’s plea agreement does not clearly rely on a subsequently lowered sentencing guideline, then the defendant is not eligible to seek a reduction (If Not-C, then Not-X.) Or they might adopt some other rule. These would remain open questions. As one judge put it, the lack of an identifiable “narrowest ground” in the context of Marks is “similar to a 4‑4 split that affirms the lower court’s opinion but does not supply a national rule governing future litigation.”66

Option 2: Different opinions govern different halves of the biconditional.

A second approach would look at both halves of the biconditional independently and apply the logical-subset inquiry to each, even though this approach would follow the concurrence for one half and the plurality for the other. Assume again that the Freeman plurality’s rule is such that every time a defendant is eligible for a reduction under the Freeman concurrence he or she is necessarily eligible for a reduction under the Freeman plurality. As with Option 1 above, the Freeman concurrence would be the logical subset of the plurality with respect to the top half of the biconditional. Accordingly, Freeman would establish the concurrence’s top-half rule: If a defendant’s plea agreement clearly relies on a subsequently lowered sentencing guideline, then the defendant is eligible to seek a reduction (If C, then X).

This second approach, however, would also establish as binding law the bottom-half of the plurality’s biconditional.67 Assuming that C is a logical subset of P, the universe of cases where the plurality would find the defendant ineligible for a sentencing reduction (Not-P) would necessarily be a subset of the universe of cases where the concurrence would find the defendant ineligible (Not-C). Accordingly, Freeman would establish the plurality’s bottom-half rule: If the plea agreement does not clearly rely on a subsequently lowered sentencing guideline and that guideline is not relevant to the judge’s analysis in approving the plea agreement, then the defendant is not eligible to seek a reduction (If Not-P, then Not-X).

To summarize, this approach would generate the following rules:

If a defendant’s plea agreement clearly relies on a subsequently lowered sentencing guideline, then the defendant is eligible to seek a reduction. (If C, then X.).

And:

If the plea agreement does not clearly rely on a subsequently lowered sentencing guideline and that guideline is not relevant to the judge’s analysis in approving the plea agreement, then the defendant is not eligible to seek a reduction. (If Not-P, then Not-X).

Here too, however, no binding law would be made with respect to the key area of disagreement between the plurality and concurrence: that is, the situation where the plea agreement does not refer to the subsequently lowered guideline but that guideline is relevant to the judge’s analysis in accepting the plea agreement. As with Option 1, future courts would be free to adopt the plurality’s more defendant-friendly top-half rule—which would allow a sentence reduction based on the judge’s use of the subsequently lowered guideline. Or they might adopt the concurrence’s less defendant-friendly bottom-half rule—which would forbid a sentence reduction due to the plea agreement’s failure to clearly rely on a subsequently lowered guideline. Or they might adopt some other rule.

* * *

This Essay’s goal is not to endorse either one of the two options above—or even to endorse the logical-subset approach as a general matter. Option 2 would generate more law than Option 1, and it would do so only as to areas of consensus between the plurality and the concurrence. On the other hand, it would mean that no single opinion would constitute the Court’s holding. The crucial point here is that under either variant of the logical-subset approach, the Freeman concurrence’s complete biconditional rule would not be binding. This makes sense because the plurality and concurrence would not necessarily reach the same result for a defendant whose plea agreement did not clearly refer to a subsequently lowered sentencing guideline. Although future courts would have the option to adopt both halves of the Freeman concurrence, they would not be bound to do so.

III. biconditionals and partial overlaps

Under a logical-subset approach to the Marks rule, a nonmajority opinion would create binding law only if there is a complete overlap as to at least one-half of the biconditional. When there is only a partial overlap, one might still extract binding law using what has been called a shared-agreement approach.68 This is a more expansive approach because it can create binding law even if no logical-subset relationship exists between the plurality and the concurrence. But it still requires consensus between the judges in the majority in the sense that it would make law only as to the area of partial overlap between the antecedents of the competing rules. Even on this understanding of Marks, however, it is important to examine each half of the competing biconditional rules.

Focusing first on the top half of a biconditional rule, a shared-agreement or partial-overlap approach unfolds as follows. Assume, as above, the plurality’s rule is (If P, then X) and the concurrence’s rule is (If C, then X). Even if neither rule is necessarily narrower than the other, they logically encompass the rule (If both P and C, then X). When both P and C are true, the plurality demands the consequent X (because P would be true), and the concurrence would also demand the consequent X (because C would be true). The decision would establish as binding precedent the rule that reflects that area of partial overlap (If both P and C, then X).69

This would provide a workable solution even when the precise contours of the competing rules are unclear. For example, as described above, lower courts have disagreed about whether the Freeman plurality would deem a defendant eligible for a sentence reduction when the plea agreement clearly relied on a subsequently lowered guideline, but the district court judge did not consider that guideline when accepting the agreement.70 One might read the plurality opinion to endorse the general principle that a defendant is eligible for a sentence reduction when the subsequently lowered guideline is relevant to the judge’s analysis in approving the plea agreement, but not to conclusively resolve whether that guideline is relevant to the judge’s analysis solely because plea agreement itself relied on it.

On this reading, the top-half rule of the Freeman plurality would be:

If a subsequently lowered sentencing guideline is relevant to the judge’s analysis in approving the plea agreement, then the defendant is eligible to seek a reduction (If P, then X).

Although we might not know with certainty whether this rule necessarily would be satisfied in every case when the concurrence’s rule is satisfied, the following rule would cover the area of (at least partial) overlap between the plurality and the concurrence:

If a subsequently lowered sentencing guideline is relevant to the judge’s analysis in approving the plea agreement and the defendant’s plea agreement clearly relies on a subsequently lowered sentencing guideline, then the defendant is eligible to seek a reduction (If both P and C, then X).

A shared agreement would also exist with respect to the bottom half of the competing biconditionals. The plurality’s bottom-half rule is (If Not-P, then Not-X), and the concurrence’s bottom-half rule is (If Not-C, then Not-X). These logically encompass the rule (If both Not-P and Not-C, then Not-X).

This partial-overlap approach would establish the following bottom-half rule:

If a subsequently lowered sentencing guideline is not relevant to the judge’s analysis in approving the plea agreement and the defendant’s plea agreement does not clearly rely on a subsequently lowered sentencing guideline, then the defendant is not eligible to seek a reduction (If both Not-P and Not-C, then Not-X).

As with the logical-subset inquiry, this approach would make no law in the potential areas of disagreement between the plurality and the concurrence. This would leave future courts free to adopt other rules to handle those areas of disagreement, including the plurality’s entire biconditional or the concurrence’s entire biconditional. It follows that under this partial-overlap approach, courts would not be compelled to adopt the Freeman concurrence in its entirety.

IV. counting dissenting votes

Another possible approach would incorporate rules endorsed by dissenting Justices into either a logical-subset or partial-overlap analysis. The basic mechanics of identifying logical subsets or partial overlaps would be the same as described above. The difference would be that rules adopted by dissenting Justices must also be considered, as long as they combine with rules endorsed by enough other Justices to “get to five.”71

While it would not be implausible to consider the rationales of dissenting Justices,72 this is not the prevailing view. As a recent treatise explained, “the prevailing view is that the narrowest grounds are those that, when applied to other cases, would consistently produce results that a majority of the Justices supporting the result in the governing precedent would have reached.”73 Even the Eleventh Circuit—which found that the Freeman concurrence stated the Supreme Court’s complete holding—has explained that “Marks does not direct lower courts interpreting fractured Supreme Court decisions to consider the positions of those who dissented.”74

In Freeman, incorporating the rules of the dissenters could yield a range of outcomes, particularly in light of the uncertainty surrounding the scope of the plurality’s rule. This Essay will not itemize all of the possible permutations, but it will walk through one example to illustrate an important point. Counting the views of dissenting Justices is the only way to support the government’s position in Hughes: namely, that courts are bound to follow the bottom-half rule of the Freeman concurrence, which makes a defendant ineligible for a sentence reduction if the plea agreement does not clearly refer to a subsequently lowered guideline.

As described earlier, the dissenting Justices in Freeman endorsed a rule: if a defendant is sentenced pursuant to a Rule 11(c)(1)(C) plea agreement, then the defendant is not eligible for a sentence reduction (If D, then Not-X.)75 Because this rule compels the consequent (Not-X), the four dissenting Justices would add no votes to either the plurality’s or the concurrence’s top-half rules. But the dissenters can create five-Justice coalitions by combining with either the plurality’s or the concurrence’s bottom-half rules.

For illustration, focus on the relationship between the dissent and the concurrence. Recall the bottom-half rule of the Freeman concurrence: if a defendant’s plea agreement does not clearly rely on a subsequently lowered sentencing guideline, then the defendant is not eligible to seek a reduction. (If Not-C, then Not-X.)

The antecedent of this rule (Not-C)is a logical subset of the dissent’s rule. The universe of cases where a defendant is sentenced pursuant to an 11(c)(1)(C) plea agreement (D) includes both cases where the plea agreement does clearly rely on a subsequently lowered sentencing guideline (C) and cases where it does not clearly rely on a subsequently lowered sentencing guideline (Not-C).

Therefore, when we factor in the Freeman dissenters—using either a logical-subset or partial-overlap methodology—there are five Justices whose rules support the proposition that a defendant is ineligible for a sentence reduction when an 11(c)(1)(C) plea agreement does not clearly refer to a subsequently lowered sentencing guideline. The dissenting Justices would agree because in their view no defendants who are sentenced pursuant to an 11(c)(1)(C) plea are eligible. And the concurrence would agree because such a defendant’s plea agreement did not clearly refer to the subsequently lowered guideline. The big picture here is not surprising. When we incorporate the restrictive approach of the Freeman dissenters, we end up with case law that is less friendly to defendants.

It is beyond the scope of this Essay to comprehensively assess the advantages and disadvantages of an approach to Marks that considers the rationales of dissenting Justices. As discussed above, however, allowing dissenting Justices to determine the binding content of a Supreme Court decision would be a significant departure from the prevailing understanding of Marks.76 Yet the government’s position in Hughes—and the Eleventh Circuit’s reading of Freeman—can only be correct if the Supreme Court decides to embrace an approach to Marks that incorporates the views of dissenting Justices.

Conclusion

Courts and scholars have articulated a variety of ways to determine the law-generating content of Supreme Court decisions that lack a majority opinion. Whether the Court embraces an approach that requires logical subsets, an approach that identifies partial overlaps, or an approach that counts dissenting votes, a coherent methodology must take into account the unique features of biconditional rules. There is simply no way to do Marks right without analyzing biconditionals, and courts and litigants have been trying unsuccessfully for too long. Hughes is a perfect opportunity for the Supreme Court to direct courts and litigants to attend to biconditionals when applying the Marks rule, however conceived.

Adam Steinman is the University Research Professor of Law at the University of Alabama School of Law. Thanks to Jenny Carroll for her helpful comments and suggestions, and to the editors of the Yale Law Journal—especially Salil Dudani, Cody Poplin, Zoe Jacoby, and Jordan R. Goldberg—for their exemplary editorial work.