United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT
Filed: August 13, 2013
No. 10-3091
UNITED STATES OF AMERICA,
APPELLEE
v.
DAVID A. DUVALL, ALSO KNOWN AS TONE,
APPELLANT
Consolidated with 11-3114
Appeals from the United States District Court
for the District of Columbia
(No. 1:09-cr-00236-RCL-1)
On Petition for Rehearing En Banc
______
Before: GARLAND, Chief Judge; HENDERSON, ROGERS**,
TATEL, BROWN, GRIFFITH, KAVANAUGH**, and SRINIVASAN*,
Circuit Judges; WILLIAMS**, Senior Circuit Judge
2
ORDER
Upon consideration of appellant’s petition for rehearing en
banc, the response thereto, and the absence of a request by any
member of the court for a vote, it is
ORDERED that the petition be denied.
Per Curiam
FOR THE COURT:
Mark J. Langer, Clerk
BY: /s/
Jennifer M. Clark
Deputy Clerk
* Circuit Judge Srinivasan did not participate in this matter.
** A statement by Circuit Judge Rogers concurring in the
denial of the petition for rehearing en banc is attached.
** A statement by Circuit Judge Kavanaugh concurring in the
denial of the petition for rehearing en banc is attached.
** A statement by Senior Circuit Judge Williams concurring
in the denial of the petition for rehearing en banc is attached.
ROGERS, Circuit Judge, concurring in the denial of
rehearing en banc: I write regarding Judge Kavanaugh’s
critique today of a related case, United States v. Epps (D.C. Cir.
2013). Our decision in Epps adhered to the interpretation of
Marks v. United States, 430 U.S. 188 (1977), adopted by the en
banc court in King v. Palmer, 950 F.2d 771 (D.C. Cir. 1991),
which is binding on the court. See LaShawn A. v. Barry, 87 F.3d
1389, 1395 (D.C. Cir. 1996) (en banc). Any “seriously flawed
application of the Marks principle,” Kavanaugh Op. at 3, if
present, is thus to be found in King v. Palmer, not Epps.
Critiquing Epps because “Marks means that, when one of the
opinions in a splintered Supreme Court decision has adopted a
legal standard that would consistently produce results with
which a majority of the Court in that case necessarily would
agree, that opinion controls,” Kavanaugh Op. at 5-6, is a direct
challenge to King v. Palmer.
I.
As examined at length in Epps, 707 F.3d at 348–51, King v.
Palmer precluded adoption of the interpretation of Marks
followed in other circuits that have declared binding Justice
Sotomayor’s concurring opinion in the splintered decision in
Freeman v. United States, 131 S. Ct. 2685 (2011). In King v.
Palmer, the en banc court interpreted Marks to mean that, to be
binding as representing the narrowest grounds for decision, an
opinion “must represent a common denominator of the Court’s
reasoning; it must embody a position implicitly approved by at
least five Justices who support the judgment.” 950 F.2d at 781
(emphasis added). Under Marks’ “narrowest grounds”
approach, for an opinion to be controlling it must contain a
“controlling rationale.” Id. at 781 n.6. “Marks is workable . . .
only when one opinion is a logical subset of other, broader
opinions.” Id. at 781. Otherwise, the en banc court reasoned,
“[i]f applied in situations where the various opinions supporting
the judgment are mutually exclusive, Marks will turn a single
opinion that lacks majority support into national law.” Id. at
2
782. So, “[w]hen . . . one opinion supporting the judgment does
not fit entirely within a broader circle drawn by the others,
Marks is problematic.” Id. According to the en banc court,
Marks applies when “the concurrence posits a narrow test to
which the plurality must necessarily agree as a logical
consequence of its own, broader position.” Id. (emphasis
added); see also Epps, 707 F.3d at 348 (quoting same).
King v. Palmer expressed no ambivalence regarding either
its general holding on the proper understanding of Marks’
“narrowest grounds” rule as requiring a binding opinion to
evidence a common rationale and not only a common outcome,
or its specific holding that, under its interpretation of Marks, in
the absence of compatible reasoning by a majority in a
splintered Supreme Court decision, “the appropriate course is to
hold that contingency enhancements [to plaintiffs’ counsel’s
award of attorneys’ fees] will not be available in this circuit.”
King v. Palmer, 950 F.2d at 784 (addressing the “splintered
decision” in Pennsylvania v. Delaware Valley Citizens’ Council
for Clean Air, 483 U.S. 711 (1987) (“Delaware Valley II”), and
reversing this court’s precedent in McKenzie v. Kennickell, 875
F.2d 330 (D.C. Cir. 1989), and portions of previous inconsistent
opinions). The en banc court emphasized that a concurring
opinion widely assumed to be controlling due to the perceived
presence of some sphere of outcome-determined “common
ground” constituted only persuasive authority. Id. at 775–77.
It identified “three distinct approaches to the issue of
contingency enhancements in fee-shifting statutes, none of
which enjoys the support of five Justices.” Id. at 782. Given
that the Supreme Court had denied a contingency enhancement
without a remand in Delaware Valley II, the court acknowledged
that it “could not authorize the routine awarding of contingency
enhancements of whatever size,” id. at 784, but determined there
was no “narrowest opinion” to be derived from the plurality and
concurring opinions because “there simply is no practical middle
3
ground between providing enhancements routinely and not
providing them at all,” id. Then, “[k]eeping in mind that a
majority of the Supreme Court clearly agrees that the question
of attorneys’ fees must not turn into major litigation in itself,”
the en banc court adopted its own view of “the appropriate
course.” Id. (rejecting “various tests for awarding contingency
enhancements” under Delaware Valley II adopted by the other
circuits because “most of the tests appear to be difficult, if not
impossible, to meet in practice”).
II.
Epps required the court to interpret a splintered Supreme
Court decision to determine if a defendant sentenced pursuant
to a Rule 11(c)(1)(C) plea agreement was eligible for a
sentencing reduction under 18 U.S.C. § 3582(c)(2). Application
of the Marks rule adopted in King v. Palmer to the Supreme
Court’s splintered 4-1-4 decision in Freeman revealed that
Justice Sotomayor’s concurring opinion cannot be understood
to be the “narrowest grounds” of the majority’s ruling, and
therefore ought not be treated as controlling.1 Epps explained:
“The plurality opinion [in Freeman] reject[ed] the concurring
opinion’s approach, stating its rationale is fundamentally
incorrect because § 3582(c)(2) ‘calls for an inquiry into the
reasons for a judge’s sentence, not the reasons that motivated or
informed the parties.’” 707 F.3d at 350 (quoting Freeman, 131
S. Ct. at 2694 (plurality opinion) (labeling the concurrence’s
rationale an “erroneous rule”)). The dissenting opinion
“agree[d] with the plurality that the approach of the concurrence
. . . is arbitrary and unworkable” and even “wrong.” Freeman,
131 S. Ct. at 2703 (dissent). In light of King v. Palmer, the only
1
In United States v. Duvall, by contrast, the court accepted
the parties’ stipulation that Justice Sotomayor’s concurring opinion
was controlling. 705 F.3d 479, 483 (D.C. Cir. 2013).
4
controlling holding evident from Freeman is that courts cannot
categorically bar defendants sentenced pursuant to Rule
11(c)(1)(C) plea agreements from eligibility for a sentencing
reduction under § 3582(c)(2). See Epps, 707 F.3d at 351.
Consequently, the Epps court had to determine, as the en
banc court instructed, which reasoning represented “the
appropriate course,” King v. Palmer, 950 F.2d at 784. The
Freeman plurality had observed that “the statute . . . calls for an
inquiry into the reasons for a judge’s sentence, not the reasons
that motivated or informed the parties,” and that the
concurrence, “[b]y allowing modification only where the terms
of the agreement contemplate it . . . would permit the very
disparities that the Sentencing Reform Act seeks to eliminate.”
Freeman, 131 S. Ct. at 2694. The Freeman dissent, on the other
hand, despaired of being able to determine, where a judge
approved the parties’ agreed-upon fixed-term sentence under
Rule 11(c)(1)(C), what the sentence was “based on,” id. at 2703
(dissent) (quoting § 3582(c)(2)), a circumstance not present in
Epps, 707 F.3d at 352 (citing United States v. Epps, 756 F.
Supp. 2d 88, 92-93 (D.D.C. 2010)). This court adhered to the
plurality’s interpretation in view of the purpose of the
Sentencing Act to reduce unwarranted disparities, Rule
11(c)(1)(C)’s requirement that the sentencing judge approve the
plea agreement, and instructions in applicable Sentencing
Guidelines policy statements. Epps, 707 F.3d at 351–53. (The
third member of the panel concluded the issue of Epps’
entitlement to sentencing relief was moot. Id. at 353 (Brown,
J., dissenting)). In so doing, Epps also followed the approach
adopted in United States v. Berry, 618 F.3d 13 (D.C. Cir. 2010),
where this court relied on the Guidelines policy statement
1B1.10, which the Freeman plurality explained “seeks to isolate
whatever marginal effect the since-rejected Guidelines had on
the defendant’s sentence,” thereby “permitting the district court
to revisit a prior sentence to whatever extent the sentencing
5
range . . . was a relevant part of the analytic framework the
judge used to determine the sentence or to approve the [Rule
11(c)(1)(C)] agreement.” 131 S. Ct. at 2692–93.
The three asserted errors in Epps are non-existent: First,
Epps did not adopt “a novel standard for Marks cases.”
Kavanaugh Op. at 10. Instead, the court correctly concluded
that “[u]nder Marks [as interpreted in King v. Palmer] . . . there
is no controlling opinion in Freeman because the plurality and
concurring opinions do not share common reasoning whereby
one analysis is a ‘logical subset,’ King, 950 F.2d at 781, of the
other.” Epps, 707 F.3d at 350 (emphasis added for text omitted
by Judge Kavanaugh).
Second, a suggestion in the concurring opinion in United
States v. Duvall, 705 F.3d 479 (D.C. Cir. 2013), regarding
possible “weak” or “strong” readings of King v. Palmer, see
Duvall, 705 F.3d at 485-86 (Williams, J., concurring), cannot
change the en banc court’s unambiguous holding on the proper
interpretation of Marks’ “narrowest grounds.” See Kavanaugh
Op. at 12-13. Nevertheless, Epps addressed the suggested
“weak reading,” anticipatorily explaining why the concurring
opinion in Freeman still would not be deemed controlling. The
court referred under the “weak” reading to the circumstance (as
contemplated by the plurality in Freeman, 131 S. Ct. at 2694)
in which the concurrence in Freeman would grant relief and the
plurality would not. See Epps, 707 F.3d at 350–51 & n.8.
Other courts appear not to have considered this circumstance,
see, e.g., United States v. Rivera-Martinez, 665 F.3d 344, 348
(1st Cir. 2011); see also In re Sealed Case, 12-3012, 2013 WL
3305706 at *3 (D.C. Cir. July 2, 2013) (dictum).
Third, Epps did not “misappl[y] the ‘Marks corollary’ that
necessarily governs in cases where there is no one narrowest
opinion.” Kavanaugh Op. at 17. As described by Judge
6
Kavanaugh, that corollary requires a court to “run the facts and
circumstances of the current case through the various tests
articulated by the Supreme Court in the binding case.” Id. at 18.
Whatever its merits, the en banc court never endorsed this
corollary to Marks, but adhered to its “subset” analysis. King
v. Palmer, 950 F.3d at 784 & n.7; see also id. at 781-82. None
of the Supreme Court cases cited by Judge Kavanaugh in
discussing vertical stare decisis, Kavanaugh Op. at 7-8, includes
a binding instruction adopting his Marks corollary.2 So, this
court must follow the approach adopted by the en banc court —
identifying “the appropriate course,” King v. Palmer, 950 F.2d
at 784 — which Epps identified in terms of the undisputed
purpose of the Sentencing Reform Act, the plain texts of Rule
11(c)(1)(C) and the Sentencing Guidelines policy, and our own
§ 3582(c)(2) precedent relying on the policy statement..
Properly understood, Judge Kavanaugh’s objection is to the
interpretation of Marks adopted by the en banc court in King v.
Palmer. In disagreeing with the result in Epps, he revives
objections reminiscent of the dissent in King v. Palmer, 950
F.2d at 789 (dissent), but the three-judge panel in Epps was not
free to agree with this position. Judge Williams, concurring in
the denial of en banc in Duvall, elaborates on why Judge
Kavanaugh’s criticism of the en banc court’s interpretation of
Marks may lack traction and rebuts his criticisms of Epps.
Regardless, because the en banc decision in King v. Palmer is
binding, the suggestion that the court in Epps acted other than
2
In O’Dell v. J.D. Netherland, 521 U.S. 151 (1997), see
Kavanaugh Op. at 10-11, a capital case, the Supreme Court discussed
another capital case, Gardner v. Florida, 430 U.S. 349 (1977), in
which the concurring opinion appears to be consistent with King v.
Palmer’s “subset” interpretation of the Marks rule. See O’Dell, 521
U.S. at 174-75 (dissent) (pointing out that the Gardner opinion
identified in O’Dell as “the narrow one” embraced the due process
rationale of the Gardner plurality).
7
consistently with its constitutional obligation is without merit.
KAVANAUGH, Circuit Judge, concurring in the denial of
rehearing en banc:
When the Supreme Court issues a splintered decision –
that is, a decision where a majority of the Court agrees on the
result but not the reasoning – the binding holding is the
position taken by those Justices “who concurred in the
judgment[] on the narrowest grounds.” Marks v. United
States, 430 U.S. 188, 193 (1977). Under the Marks principle,
when one of the opinions in a splintered Supreme Court
decision has adopted a legal standard that would produce
results with which a majority of the Court in that case
necessarily would agree, that opinion controls.
In a recent case, Freeman v. United States, the Supreme
Court splintered over whether a defendant who enters into a
binding Rule 11(c)(1)(C) plea agreement specifying an
agreed-upon sentence is eligible for a sentence reduction after
the Guidelines sentencing range for the offense has been
retroactively reduced. 131 S. Ct. 2685 (2011). Under the
governing statute, defendants are eligible for a sentence
reduction if, among other things, they were sentenced “based
on” a Guidelines sentencing range that was later lowered by
the Sentencing Commission. See 18 U.S.C. § 3582(c)(2).
In an opinion by Justice Kennedy, a four-Justice plurality
of the Freeman Court reasoned that sentences in cases with
Rule 11(c)(1)(C) plea agreements are always “based on” a
Guidelines sentencing range because the Judge’s decision to
accept such plea agreements is framed by the Guidelines.
Thus, in the plurality’s view, those defendants are always
eligible for a sentence reduction when the applicable
Guidelines sentencing range has been retroactively reduced.
In an opinion by Chief Justice Roberts, a four-Justice dissent
reasoned that sentences in cases with Rule 11(c)(1)(C) plea
agreements are never “based on” a Guidelines sentencing
range because those sentences are based on the plea
2
agreement, not on the Guidelines. Thus, in the dissent’s view,
those defendants are never eligible for a sentence reduction.
Justice Sotomayor, alone concurring in the judgment,
adopted a middle ground in Freeman. She reasoned that
sentences in cases with Rule 11(c)(1)(C) plea agreements are
sometimes “based on” a Guidelines sentencing range –
namely, when the plea agreement itself makes clear that the
basis for the specified sentence is a Guidelines sentencing
range. Thus, in those circumstances, defendants are eligible
for a sentence reduction when the applicable Guidelines
sentencing range has been retroactively reduced.
The question at issue in this appeal is which opinion is
the binding opinion in Freeman. The answer should be
obvious. Justice Sotomayor’s opinion resolved the case on
the narrowest grounds and is therefore the binding opinion in
Freeman. Adhering to her opinion would produce results
with which a majority of the Freeman Court would agree.
Not surprisingly, every other court of appeals to consider the
question has therefore determined that Justice Sotomayor’s
opinion is the binding opinion. See United States v. Rivera-
Martínez, 665 F.3d 344, 347-48 (1st Cir. 2011); United States
v. White, 429 F. App’x 43, 47 (2d Cir. 2011) (unpublished);
United States v. Thompson, 682 F.3d 285, 289-90 (3d Cir.
2012); United States v. Brown, 653 F.3d 337, 340 & n.1 (4th
Cir. 2011); United States v. Smith, 658 F.3d 608, 611 (6th Cir.
2011); United States v. Dixon, 687 F.3d 356, 359-60 (7th Cir.
2012); United States v. Browne, 698 F.3d 1042, 1045 (8th Cir.
2012); United States v. Austin, 676 F.3d 924, 927 (9th Cir.
2012); United States v. Graham, 704 F.3d 1275, 1278 (10th
Cir. 2013); United States v. Lawson, 686 F.3d 1317, 1321 n.2
(11th Cir. 2012). Indeed, in Freeman itself, Justice
Kennedy’s plurality opinion for four Justices referred to
Justice Sotomayor’s position as the “intermediate position.”
3
131 S. Ct. at 2694 (opinion of Kennedy, J.). And Chief
Justice Roberts’s dissent for four Justices likewise treated
Justice Sotomayor’s opinion as binding. See id. at 2700-05
(Roberts, C.J., dissenting). And in United States v. Duvall, a
panel of this Court similarly assumed as much, as did both the
Government and the defendant in that case. 705 F.3d 479
(D.C. Cir. 2013).
But then in United States v. Epps, a separate panel of this
Court adopted Justice Kennedy’s plurality opinion in
Freeman. United States v. Epps, 707 F.3d 337, 350-51 (D.C.
Cir. 2013). Epps means that courts in this Circuit often will
find that a sentence was “based on” the Guidelines (and thus
the defendant is eligible for a sentencing reduction) even
when the sentence indisputably was not “based on” the
Guidelines under the approaches adopted by a majority of the
Freeman Court – namely, under the opinions of Justice
Sotomayor and Chief Justice Roberts. Indeed, Epps himself
would have lost under the approaches adopted by a majority
of the Supreme Court in Freeman. Under the opinions of
Justice Sotomayor and Chief Justice Roberts, Epps’s plea
agreement indisputably was not “based on” a Guidelines
sentencing range.
I find Epps to be a seriously flawed application of the
Marks principle. The Epps decision puts us out of step with
every other court of appeals that has considered Freeman and,
more fundamentally, will lead to results in many cases with
which a majority of the Supreme Court in Freeman would
necessarily disagree. The point of Marks, however, is for
lower courts to reach results with which a majority of the
Supreme Court in the relevant precedent would agree.
In his en banc petition here, Duvall understandably seeks
to reap the benefit of the Epps ruling. I am voting to deny the
4
petition because the problem is Epps, not Duvall, and we
should await an en banc petition that asks us to reconsider
Epps. If we receive such a petition, we should grant it and
overturn Epps so as to follow Justice Sotomayor’s binding
opinion in Freeman and bring our case law into line with
every other court of appeals that has considered the question.
I
Vertical stare decisis is absolute and requires lower
courts to follow applicable Supreme Court rulings in every
case. The Constitution vests Judicial Power in only one
Supreme Court. U.S. CONST. art. III, § 1. We are subordinate
to that one Supreme Court, and we must decide cases in line
with Supreme Court precedent.
Vertical stare decisis applies to Supreme Court precedent
in two ways. First, the result in a given Supreme Court case
binds all lower courts. Second, the reasoning of a Supreme
Court case also binds lower courts. So once a rule, test,
standard, or interpretation has been adopted by the Supreme
Court, that same rule, test, standard, or interpretation must be
used by lower courts in later cases. See, e.g., Seminole Tribe
of Florida v. Florida, 517 U.S. 44, 67 (1996) (“When an
opinion issues for the Court, it is not only the result but also
those portions of the opinion necessary to that result by which
we are bound.”); see also Randall v. Sorrell, 548 U.S. 230,
243 (2006) (opinion of Breyer, J.) (stare decisis “commands
judicial respect for a court’s earlier decisions and the rules of
law they embody”); County of Allegheny v. American Civil
Liberties Union, Greater Pittsburgh Chapter, 492 U.S. 573,
668 (1989) (separate opinion of Kennedy, J.) (“As a general
rule, the principle of stare decisis directs us to adhere not only
to the holdings of our prior cases, but also to their explications
of the governing rules of law.”).
5
In most cases, at least five Justices of the Supreme Court
join a single majority opinion that agrees on both the result
and the reasoning. Justices who join the majority may of
course express additional thoughts in a concurrence, but
concurrences do not bind lower courts in cases where there is
a majority opinion.
In some cases, however, no single opinion commands
majority support, even though five or more Justices agree on a
result. In those cases, there will be two or more opinions
supporting the judgment of the Court, without any one
opinion commanding a majority. For much of the Court’s
history, the precedential value of such splintered decisions
was unclear. See Justin Marceau, Plurality Decisions:
Upward-Flowing Precedent and Acoustic Separation, 45
CONN. L. REV. 933, 948-49 (2013); Joseph M. Cacace, Note,
Plurality Decisions in the Supreme Court of the United
States: A Reexamination of the Marks Doctrine After Rapanos
v. United States, 41 SUFFOLK U. L. REV. 97, 104-05 (2007).
In Marks v. United States, the Supreme Court furnished
important guidance regarding the stare decisis value of
splintered decisions. 430 U.S. 188 (1977). There, the Court
articulated a commonsense test to determine which opinion
(for example, a plurality opinion or an opinion concurring in
the judgment) is the binding opinion in a splintered decision:
“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 . . . .” Id. at 193 (internal
quotation marks omitted).
Marks means that, when one of the opinions in a
splintered Supreme Court decision has adopted a legal
6
standard that would produce results with which a majority of
the Court in that case necessarily would agree, that opinion
controls. In most cases, the commonsense way to apply
Marks is to identify and follow the opinion that occupies the
middle ground between (i) the broader opinion supporting the
judgment and (ii) the dissenting opinion. That middle-ground
opinion will produce results that represent a subset of the
results generated by the other opinions. By applying that
approach, lower courts will decide cases consistently with the
opinions of a majority of the Supreme Court in the relevant
precedent. See, e.g., O’Dell v. Netherland, 521 U.S. 151, 160
(1997) (adopting opinion from Gardner v. Florida, 430 U.S.
349 (1977), that predicts majority result even when Justices in
prior precedent had based reasoning on different
constitutional amendments); Dickens v. Brewer, 631 F.3d
1139, 1145 (9th Cir. 2011) (“When applying the Marks rule,
we look for a legal standard which, when applied, will
necessarily produce results with which a majority of the
Justices from that case would agree.”) (internal quotation
marks and brackets omitted); Planned Parenthood v. Casey,
947 F.2d 682, 693 (3d Cir. 1991) (“Where a Justice or
Justices concurring in the judgment in such a case articulates
a legal standard which, when applied, will necessarily
produce results with which a majority of the Court from that
case would agree, that standard is the law of the land.”); King
v. Palmer, 950 F.2d 771, 781 (D.C. Cir. 1991) (narrowest
opinion must be “logical subset of other, broader opinions”
and “must represent a common denominator of the Court’s
reasoning; it must embody a position implicitly approved by
at least five Justices who support the judgment”) (emphasis
added).
The Marks rule is an essential aspect of vertical stare
decisis: “The binding opinion from a splintered decision is as
authoritative for lower courts as a nine-Justice opinion. While
7
the opinion’s symbolic and perceived authority, as well as its
duration, may be less, that makes no difference for a lower
court. This is true even if only one Justice issues the binding
opinion.” Planned Parenthood, 947 F.2d at 694.1
In interpreting most splintered Supreme Court decisions,
the Marks rule is not especially complicated. But on rare
occasions, splintered decisions have no “narrowest” opinion
that would identify how a majority of the Supreme Court
would resolve all future cases. Marks itself did not have
reason to specifically address that situation. But in that
situation, the necessary logical corollary to Marks is that
lower courts should still strive to decide the case before them
in a way consistent with how the Supreme Court’s opinions in
the relevant precedent would resolve the current case. See
King, 950 F.2d at 784 (confronting this circumstance and
stating that when a decision “provides no controlling legal
holding,” it nevertheless has “binding impact”); see, e.g.,
Nichols v. United States, 511 U.S. 738 (1994). The easy way
to do that is for the lower court to run the facts and
circumstances of the current case through the tests articulated
in the Justices’ various opinions in the binding case and adopt
the result that a majority of the Supreme Court would have
reached. See, e.g., Bobby v. Dixon, 132 S. Ct. 26, 31-32
(2011) (reaching result that plurality and concurring opinion
from relevant precedent would reach); City of Ontario v.
Quon, 130 S. Ct. 2619, 2629 (2010) (reaching result when
two “approaches – the plurality’s and Justice Scalia’s . . . lead
1
To be clear, the goal for a lower court under Marks is not to
speculate or predict how a future Supreme Court might decide a
case. The goal is to determine how the principles set forth by the
Supreme Court in a prior decision would apply to the current case
facing the lower court. Thus, Judge Williams is incorrect to
suggest that I would require lower courts to predict future Supreme
Court action. See Williams Op. at 10.
8
to the same result”); Rapanos v. United States, 547 U.S. 715,
758 (2006) (Roberts, C.J., concurring) (cases with no
controlling opinion like Rapanos have to be interpreted on a
“case-by-case basis”); id. at 810 (Stevens, J., dissenting)
(when at least five Justices – the dissent plus either the
plurality or concurrence – would reach a given result, then
lower courts should reach that result).2
Indeed, if a lower court ever has doubt about the
predictive utility of a single opinion from a splintered
Supreme Court decision, this opinion-by-opinion
methodology is a foolproof way to reach the correct result in
the lower court’s subsequent decisions. Again, that is really
just common sense in a system of absolute vertical stare
decisis.
II
Now back to Freeman. Applying the Marks rule to
Freeman is fairly easy, which no doubt explains why every
other court of appeals has reached the same conclusion:
Justice Sotomayor’s Freeman opinion is the binding opinion
from that case.
In Freeman, the Supreme Court considered whether a
defendant who entered into a binding Rule 11(c)(1)(C) plea
2
When the Supreme Court itself applies Marks, it is not bound
in the same way that lower courts are bound by Marks to strictly
follow the narrowest opinion from a prior splintered Supreme Court
decision. That’s because the Supreme Court is free to reconsider or
refine or tweak its own precedents – including splintered precedents
– and it does so in appropriate cases. Lower courts, by contrast, are
not free to reconsider or refine or tweak Supreme Court precedents.
Marks is therefore even more important at the lower court level.
9
agreement specifying an agreed-upon sentence was eligible
for a sentence reduction after the relevant Guidelines
sentencing range for the offense had been retroactively
reduced. Federal law permits a sentence reduction when a
defendant “has been sentenced to a term of imprisonment
based on a sentencing range that has subsequently been
lowered by the Sentencing Commission.” 18 U.S.C.
§ 3582(c)(2) (emphasis added). The key question in Freeman
was whether the sentence in a case involving a Rule
11(c)(1)(C) plea agreement is “based on” a Guidelines
sentencing range.
In a plurality opinion written by Justice Kennedy, a group
of four Justices concluded that sentences in cases with Rule
11(c)(1)(C) plea agreements are always “based on” a
Guidelines sentencing range, because “the court’s
acceptance” of the plea agreement must itself be “based on
the Guidelines.” Freeman v. United States, 131 S. Ct. 2685,
2692 (2011) (opinion of Kennedy, J.) (citing U.S.
SENTENCING GUIDELINES MANUAL § 6B1.2); see id. at 2690-
95. Another group of four Justices, in a dissent written by
Chief Justice Roberts, concluded that Rule 11(c)(1)(C)
sentences are never based on a Guidelines sentencing range,
because the sentences are “based on” the plea agreement itself
rather than on the Guidelines. See id. at 2700-05 (Roberts,
C.J., dissenting). And Justice Sotomayor, who concurred in
the judgment alone, concluded that Rule 11(c)(1)(C)
sentences are sometimes “based on” a Guidelines sentencing
range. See id. at 2695-700 (opinion of Sotomayor, J.). Justice
Sotomayor explained that sentences in Rule 11(c)(1)(C) cases
are “based on” the plea agreements, but some such plea
agreements are in turn “based on” a Guidelines sentencing
range, whereas other Rule 11(c)(1)(C) plea agreements are
not. Under Justice Sotomayor’s theory, the sentence in a
Rule 11(c)(1)(C) case is “based on” a Guidelines sentencing
10
range if the plea agreement makes “clear that the basis for the
specified term is a Guidelines sentencing range.” Id. at 2697.
Following Justice Sotomayor’s opinion with regard to the
“based on” issue would produce results with which a majority
of the Supreme Court in Freeman would agree because – to
put it in simple terms – “sometimes” is a middle ground
between “always” and “never.” In other words, when Justice
Sotomayor concludes that a plea agreement was based on the
Guidelines, she would agree with the result reached under
Justice Kennedy’s opinion for four Justices. When she
concludes that a plea agreement was not based on the
Guidelines, she would agree with the result reached under
Chief Justice Roberts’s opinion for four Justices. But unlike
every other court of appeals, Epps did not follow this
commonsense approach to interpreting Freeman.
As I see it, Epps erred in several independent ways.
First, the Epps panel articulated a novel standard for
applying Marks to splintered Supreme Court decisions: If the
Supreme Court opinions in the majority do not share a
“common rationale,” then there is no binding opinion for
lower courts to follow, even if applying one of the opinions
would produce results with which a majority of the Supreme
Court would agree. United States v. Epps, 707 F.3d 337, 349
(D.C. Cir. 2013).
But Marks does not require the multiple opinions
supporting the Supreme Court’s judgment to employ a
“common rationale.” After all, in splintered cases, there are
multiple opinions precisely because the Justices did not agree
on a common rationale. See, e.g., O’Dell v. Netherland, 521
U.S. 151, 160 (1997) (adopting controlling opinion from
Gardner v. Florida, 430 U.S. 349 (1977), despite the fact that
plurality and concurrences based reasoning on different
11
constitutional amendments); Gregg v. Georgia, 428 U.S. 153,
169 n.15 (1976) (opinion of Stewart, Powell, and Stevens, JJ.)
(recognizing opinions of Justices Stewart and White as
controlling from Furman v. Georgia, 408 U.S. 238 (1972),
even though their opinions did not employ common rationale
with all other members of majority). Even though it is often
not possible to identify a “common rationale” in the multiple
opinions from a splintered decision, lower courts can still
reach a result consistent with the opinions of a majority of the
Supreme Court. They can do so by following the opinion that
would lead to an outcome that a majority of the Supreme
Court in the governing precedent would have reached if
confronted with the current case.3 With respect to Freeman
3
The joint Epps opinion by Judges Rogers and Williams cited
this Court’s precedent in King v. Palmer as support for Epps’s
“common rationale” principle. See Epps, 707 F.3d 337; King v.
Palmer, 950 F.2d 771 (D.C. Cir. 1991). But King v. Palmer does
not require that the opinions forming a majority share a “common
rationale” in order for Marks to apply. In his Duvall concurrence,
Judge Williams explained his view that King v. Palmer is
susceptible to either a “weak” or “strong” reading. United States v.
Duvall, 705 F.3d 479, 485 (D.C. Cir. 2013) (Williams, J.,
concurring in the judgment). Under the so-called weak reading of
King, the controlling opinion in a splintered decision is the opinion
that predicts a result with which a majority of the Supreme Court
from the relevant precedent would agree. See id. at 487 (reading
“would seem to require only that as a purely factual matter cases
producing an outcome in favor of the defendant under Justice
Sotomayor’s opinion would invariably yield an outcome in his
favor under the plurality view”). The strong reading of King would
find an opinion binding only if the opinion has a “common
rationale” with the other opinions necessary to form a majority. I
believe what Judge Williams described as the weak reading of King
is most consistent with Marks and with King itself. Indeed, that
reading of King must be correct because, among other things, King
identified the opinions of Justices Stewart and White in Furman v.
12
and the “based on” issue, Justice Sotomayor’s opinion does
just that, as every other court of appeals has recognized.4
Second, Epps suggested (apparently in the alternative)
that, even if the Marks rule focuses on identifying the opinion
that would achieve results with which a majority of the
Supreme Court would agree and does not depend on whether
there was a “common rationale” among the opinions making
up the Supreme Court majority, Freeman has no narrowest
opinion that would always lead to the result that a majority of
the Supreme Court would reach in future cases. Epps
hypothesized a situation where Justice Sotomayor, despite
taking the middle ground, would supposedly disagree with
both Justice Kennedy’s four-Justice plurality opinion and
Chief Justice Roberts’s four-Justice dissenting opinion. See
Epps, 707 F.3d at 350 n.8. As Justice Kennedy’s plurality
Georgia, 408 U.S. 238, as the narrowest opinions in Furman. King,
950 F.2d at 781. There were five separate opinions for the Justices
in the majority in Furman, and none of those five Justices joined
any of the others’ opinions. Those five opinions certainly did not
have a “common rationale.” The opinions of Justices Stewart and
White were binding because they would produce results in future
cases with which a majority of the Court in Furman would agree.
Judge Rogers and Judge Williams thus are not correct in
suggesting that I disagree with King on this point. Rather, of the
two plausible readings of King that Judge Williams previously
identified in his Duvall concurrence, I simply believe the weak
reading of King to be more consistent with Marks.
4
Judge Williams’s opinion respecting the denial of en banc
essentially collapses the Epps “common rationale” standard into the
requirement that lower courts follow the opinion that would lead to
a result that a majority of the Supreme Court would reach. If so,
then my disagreement with Judge Williams relates only to the
second and third points discussed in the text of my opinion here.
That is not, however, how the Epps opinion itself appeared to
describe its “common rationale” standard.
13
opinion stated, however, the sole question resolved by
Freeman was “whether defendants who enter into 11(c)(1)(C)
agreements that specify a particular sentence may be said to
have been sentenced ‘based on’ a Guidelines sentencing
range, making them eligible for relief under § 3582(c)(2).”
Freeman, 131 S. Ct. at 2691 (opinion of Kennedy, J.)
(emphasis added). And on the question whether a
defendant’s sentence was “based on” a Guidelines sentencing
range, there is no circumstance where Justice Sotomayor
would reach a result different from all of the other eight
Justices. Given Justice Sotomayor’s conclusion that the
sentences for Rule 11(c)(1)(C) defendants are sometimes
“based on” a Guidelines sentencing range, Justice
Sotomayor’s conclusion would necessarily coincide in any
given case with either the plurality’s conclusion (which
determined that such sentences are always “based on” the
Guidelines) or with the dissent’s conclusion (which
determined that such sentences are never “based on” the
Guidelines).5 So under Marks, Justice Sotomayor’s opinion is
binding with respect to the “based on” question.
5
Epps suggested that Justice Kennedy’s plurality opinion may
not always find a Rule 11(c)(1)(C) plea agreement to be “based on”
a Guidelines sentencing range. That is incorrect. The Freeman
plurality determined that “if the judge uses the sentencing range as
the beginning point” when imposing the sentence, “then the
Guidelines are in a real sense a basis for the sentence.” 131 S. Ct.
at 2692 (opinion of Kennedy, J.). And district court judges are
required to calculate the Guidelines sentencing range as their
beginning point before imposing a Rule 11(c)(1)(C) sentence. See
U.S. SENTENCING GUIDELINES MANUAL § 6B1.2(c). So under the
plurality approach, a sentence in a Rule 11(c)(1)(C) case will
always be “based on” a Guidelines sentencing range. Indeed, this
Court recently said just that about Justice Kennedy’s plurality
opinion. See In re Sealed Case, No. 12-3012, slip op. at 5 (D.C.
14
Epps premised its conclusion that Justice Sotomayor’s
opinion was not the narrowest on the idea that Justice
Sotomayor would allow a sentencing reduction in certain Rule
11(c)(1)(C) cases involving career offenders and offenders
subject to a mandatory minimum, whereas (according to
Epps) none of the other eight Justices supposedly would.
That is, moving beyond the “based on” question, Epps
claimed that Justice Sotomayor would reach a different
conclusion than the other eight Justices on the question
whether defendants in cases involving career offenders and
offenders subject to a mandatory minimum are ultimately
eligible for a sentence reduction under Section 3582(c)(2).
But in those cases, Justice Sotomayor would not allow a
sentence reduction; indeed, no Justice would.
To explain: To obtain a sentencing reduction under
Section 3582(c)(2), the defendant not only must show that the
sentence was “based on” a Guidelines sentencing range, but
also must meet an additional requirement. The defendant
must show that the “applicable guideline range” was
Cir. July 2, 2013). In Sealed Case, we stated: Using Justice
Kennedy’s plurality approach, “a sentence that emerges from a
Rule 11(c)(1)(C) plea agreement is always eligible for a subsequent
reduction because ‘[t]he 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.’ ” Id. (quoting Freeman, 131 S. Ct. at
2692 (opinion of Kennedy, J.)) (emphasis added); see also United
States v. Brown, 653 F.3d 337, 339 (4th Cir. 2011) (under plurality
approach, “a district court can always grant § 3582(c)(2) relief to a
defendant who enters into a Rule 11(c)(1)(C) plea agreement”)
(emphasis added); United States v. Rivera-Martínez, 665 F.3d 344,
347 (1st Cir. 2011) (“Even in cases in which sentencing follows the
execution of a C-type plea agreement, the sentencing judge is
required to take the guidelines into account when deciding whether
to accept the agreement and impose the agreed sentence.”).
15
subsequently lowered. U.S. SENTENCING GUIDELINES
MANUAL § 1B1.10(a)(2)(B) (emphasis added); see 18 U.S.C.
§ 3582(c)(2) (sentence revision must be “consistent with
applicable policy statements issued by the Sentencing
Commission”). In cases where a defendant’s applicable
Guidelines sentencing range involves a career offender or
mandatory minimum provision, the fact that the Guidelines
sentencing range for the offense of conviction – for example,
a crack-cocaine Guideline – was later lowered would not
alone suffice to support a Section 3582(c)(2) sentencing
reduction. That is because, in cases involving a career
offender or defendant subject to a mandatory minimum, the
“applicable” Guidelines sentencing range is the range for
career offenders or offenders subject to a mandatory
minimum. And in such a case, that Guidelines sentencing
range would not have been lowered even if, say, the relevant
crack-cocaine Guideline had been lowered. In those
circumstances, none of the opinions in Freeman – not Justice
Kennedy’s opinion, not Justice Sotomayor’s opinion, not
Chief Justice Roberts’s opinion – would support a Section
3582(c)(2) reduction.
Indeed, the Sentencing Commission has made clear that a
reduction is not available in those cases. The Application
Note to Section 1B1.10 of the Guidelines, which governs
sentence reductions, specifically states that a defendant may
not obtain a sentence reduction if “the amendment does not
have the effect of lowering the defendant’s applicable
guideline range because of the operation of another guideline
or statutory provision (e.g., a statutory mandatory minimum
term of imprisonment).” U.S. SENTENCING GUIDELINES
MANUAL § 1B1.10, Application Note 1(A); see id., amend.
759 (effective Nov. 1, 2011) (defining the term “applicable
guideline range”).
16
Notwithstanding the language of Guidelines Section
1B1.10, Epps premised its conclusion that Justice
Sotomayor’s Freeman opinion is not the narrowest opinion on
the notion that Justice Sotomayor – alone among the Freeman
Justices – would allow a sentence reduction for career
offenders and offenders subject to a mandatory minimum,
supposedly because the plea agreement itself would not have
rested on the career offender or mandatory minimum range.
The Epps Court’s assessment of Justice Sotomayor’s opinion
thus necessarily presupposes that she would ignore the
separate Guidelines requirement (which is binding on federal
courts) that the applicable Guidelines range, which in those
circumstances would be a career offender or mandatory
minimum provision, must also be lowered in order for a
defendant to be eligible for a sentence reduction. Unlike the
Epps panel, I see no basis for thinking that Justice Sotomayor
would ignore that Guidelines requirement.6
So contrary to Epps’s assertion, Justice Sotomayor’s
opinion would not reach a different result than Justice
6
Moreover, the Epps premise also ignores this Court’s
decision in Berry, where we already held that a Section 3582(c)(2)
sentencing reduction is not available when the applicable
Guidelines sentencing range has not been lowered because of the
operation of a career offender or mandatory minimum provision.
See United States v. Berry, 618 F.3d 13 (D.C. Cir. 2010). Notably,
in Duvall, Judge Williams specifically called for the en banc court
to overrule Berry. Judge Williams’s desire to overrule Berry
becomes understandable when one appreciates that the premise of
Epps – namely, that Justice Sotomayor would still grant relief in
career offender or mandatory minimum cases – all but collapses so
long as Berry is still on the books. However, not only is Berry still
on the books, but as I noted in the text, the Sentencing Commission
has made clear that Berry is correct and that a sentencing reduction
under Section 3582(c)(2) is not appropriate in those circumstances.
17
Kennedy’s opinion or Chief Justice Roberts’s opinion in those
career offender or mandatory minimum circumstances. Thus,
the hypothetical situation that Epps identifies as the basis for
not following Justice Sotomayor’s opinion simply does not
exist.
Third, despite all that, let’s suppose that I am wrong so
far and that, indeed, there was no narrowest opinion in
Freeman that governs all future cases. In other words,
suppose that Epps is right that Justice Sotomayor’s opinion
would allow a sentencing reduction under Section 3582(c)(2)
in some circumstances where the other eight Justices would
not. Even accepting that premise, Epps misapplied the Marks
corollary that necessarily governs in cases where there is no
narrowest opinion: In such cases, the lower court should still
strive to reach the result that a majority of the Supreme Court
would have reached under the opinions in the governing
precedent.
Epps said that lower courts analyzing a splintered
decision with no narrowest opinion may unilaterally adopt
any of the Justices’ opinions that supported the Supreme
Court’s judgment – even an opinion at the extreme end of the
spectrum – and apply that one opinion in all future cases. See
Epps, 707 F.3d at 349-51. Epps thus leads to an extraordinary
situation in which lower courts in all subsequent cases may
follow a broader opinion supporting the judgment and, by
doing so, decide future cases differently than a majority of the
Supreme Court in the governing precedent would decide the
cases. Consider the logic Epps employs: Because Justice
Sotomayor’s approach would allegedly lead to relief in some
small subset of Section 3582(c)(2) cases where the other eight
Justices would deny relief, lower courts should follow Justice
Kennedy’s broader approach in all cases, even though a
majority of the Supreme Court would definitely disagree with
18
the result achieved under Justice Kennedy’s approach in many
if not most of those cases. The logic behind Epps eludes me.
The core of the Marks principle cannot be sidestepped
simply because the multiple opinions supporting the Supreme
Court’s judgment did not produce a single narrowest opinion
for all future cases. As explained above, in those rare no-
narrowest-opinion cases, the lower court still must strive to
reach the result that a majority of the Supreme Court would
have reached in the current case, if such a result can be
ascertained. The simplest way to do that, again, is to run the
facts and circumstances of the current case through the
various tests articulated by the Supreme Court in the binding
case. Using that approach, lower courts can still reach a
result consistent with the decision of a majority of the
Supreme Court. That is what Marks, common sense, and
vertical stare decisis all require.7
7
Judge Rogers says that the Marks corollary is inconsistent
with this Court’s decision in King v. Palmer. I disagree. As I read
King, the core of the problem there was that the en banc Court
simply could not decipher Justice O’Connor’s opinion in Delaware
Valley II, the Supreme Court precedent at issue. See King, 950 F.2d
at 782 (Justice O’Connor’s “concurrence does not contain enough
independent reasoning on the question of availability to allow us to
compare her position analytically to that of the plurality.”); see also
City of Burlington v. Dague, 505 U.S. 557, 563 (1992) (overruling
Delaware Valley II on the ground that “we do not see how” Justice
O’Connor’s Delaware Valley II concurrence “can intelligibly be
applied”). As the Supreme Court itself later concluded, it was
essentially impossible to know how Justice O’Connor’s opinion
would sensibly apply to the situation presented in King (or in any
other situation). Indeed, King compared Justice O’Connor’s
opinion in Delaware Valley II to Justice Harlan’s opinion in
Coolidge v. New Hampshire, 403 U.S. 443 (1971), which the King
Court described as having “no reasoning by which one could
19
So even if one mistakenly thinks that Freeman itself has
no binding opinion that governs all future cases – as Epps said
about Freeman – the solution still would not be to rule for
Epps in his case. After all, Epps would lose under the
approaches adopted by five Justices in Freeman: Justice
Sotomayor’s opinion and Chief Justice Roberts’s opinion for
four additional Justices. Under the opinions of those five
Justices, Epps’s plea agreement was not “based on” a
Guidelines sentencing range. Under Justice Sotomayor’s
opinion, Epps’s sentence was not based on a Guidelines
sentencing range because Epps’s plea agreement did not make
“clear that the basis for the specified term is a Guidelines
sentencing range.” 131 S. Ct. at 2697 (opinion of Sotomayor,
J.). And of course, under the four-Justice opinion by Chief
Justice Roberts, Epps’s sentence was not based on a
Guidelines sentencing range because Rule 11(c)(1)(C)
sentences are, in those Justices’ view, never based on a
Guidelines sentencing range.
Epps never grapples with that critical point. Nor do
Judge Rogers and Judge Williams address that reality in their
opinions respecting the denial of en banc review.8
discern his position” on the legal question at issue. 950 F.2d at
782. Here, by contrast, we can readily discern Justice Sotomayor’s
reasoning and what result Justice Sotomayor would reach in Epps’s
case.
8
In his opinion respecting denial of en banc, Judge Williams
says that, under King, the dissent in a splintered decision does not
play any role in the Marks analysis. But that is not what King said.
To be sure, as Judge Williams notes, King said that we cannot
“combine a dissent with a concurrence to form a Marks majority.”
950 F.2d at 783. In other words, looking to just the concurrence
and dissent alone will never be enough to determine whether one of
the opinions is the binding opinion under Marks. As I have
20
***
In the event all of the above discussion is a bit difficult to
follow, just focus on one key point: If the approaches of the
nine Justices in Freeman were applied to Epps’s case, Epps
would lose. That’s because Epps would lose under the
approaches of Justice Sotomayor and the four dissenters. Yet
under this Court’s decision in Epps, Epps magically has won.
In my view, that cannot be correct. How can it be consistent
with vertical stare decisis to choose an approach that
contradicts the will of a majority of the Supreme Court as
expressed in the governing precedent?
I have great respect for my two distinguished colleagues
in the Epps majority. And I recognize that Marks can
sometimes seem like a Rubik’s Cube. But in my view, the
Epps decision jumped the rails. To be clear, I certainly do not
believe that my colleagues are deliberately flouting Supreme
Court precedent. Rather, I simply believe that they have
adopted a mistaken approach to Marks that in turn will lead to
our deciding certain cases in a manner inconsistent with
Supreme Court precedent.
explained, an opinion is the binding opinion only when it will lead
to results with which a majority of the Court would agree in all
future cases. And that analysis can be logically conducted only by
looking at all of the opinions in the Supreme Court case at issue.
In the rare cases where the Marks analysis shows that there
was no binding opinion, then the principle that I have described as
the Marks corollary necessarily applies. For such cases, King did
not say or purport to say that the views of the dissenters in the
Supreme Court case are irrelevant. Put simply, I do not read King
to direct that we decide a case contrary to how a majority of the
Supreme Court in the governing precedent would decide the case.
21
In his en banc petition, Duvall naturally seeks the benefit
of the Epps holding. After all, the relevant facts of his case
are indistinguishable from Epps. But in light of my views
about Epps, I am obviously not inclined to ditch Duvall – a
case that I believe was decided correctly – in favor of Epps.
So I have voted to deny Duvall’s en banc petition. For its
part, the Government has let our decisions in Epps and Duvall
co-exist for now, but the Government presumably will seek en
banc in an appropriate future case. If the proper case arises, I
will vote to reconsider Epps en banc, to overrule it, and to
make clear that, as every other court of appeals has held, we
must follow Justice Sotomayor’s binding opinion in Freeman.
WILLIAMS, Senior Circuit Judge, concurring in the denial
of en banc. I have not asked for a vote on whether the panel
opinion should be reviewed en banc, and write here only to
discuss Judge Kavanaugh’s suggestion that the related
decision in United States v. Epps, 707 F.3d 337 (D.C. Cir.
2013), be so reviewed on the first possible occasion.
Judge Kavanaugh argues that Justice Sotomayor’s
opinion in Freeman v. United States, 131 S. Ct. 2685 (2011),
states the “narrowest” grounds for allowing relief under 18
U.S.C. § 3582(c)(2), as applied to plea agreements under Rule
11(c)(1)(C) of the Federal Rules of Criminal Procedure, and is
therefore controlling Supreme Court precedent. Judge
Kavanaugh argues that Epps’s analysis of when a splintered
decision of the Supreme Court produces a binding precedent
is inconsistent not only with the Supreme Court’s decision in
Marks v. United States, 430 U.S. 188, 193 (1977), but with
our interpretation of Marks in King v. Palmer, 950 F.2d 771,
781 (D.C. Cir. 1991). I shall address first, my understanding
of Marks and King, and their application in Epps; and second,
Judge Kavanaugh’s critique of King.
It is perfectly true that Marks used the metaphor “narrow”
to describe the circumstances where one opinion of a
splintered set is binding, see, e.g., 430 U.S. at 193, but in King
we gave that metaphor an analytical interpretation.
Specifically, we held that one of several opinions in a decision
is binding precedent only when the cases governed by its rule
represent a subset of the cases governed by the other
opinion(s) prevailing in the decision.
Without King’s requirement that one be a subset of the
other, the idea of “narrowness” is inherently confusing and in
fact indeterminate. That is because its application requires
guesswork as to the distribution of situations governed by the
rules. Let’s assume Rules A and B, each setting out grounds
2
for relief (or any specific outcome) and partially overlapping,
as in the illustration below:
As I have drawn Rule B’s sphere distinctly smaller than Rule
A’s, Rule B appears “narrower.” But the fact of a non-
overlapping area for Rule B creates a measurement problem.
How do we know that B’s non-overlap area is smaller than
A’s? Moreover, why should Rule B’s set of outcomes be
viewed as the valid set where they fail to overlap with those
produced under Rule A?
This of course describes the situation in Freeman. As the
panel opinion in Epps recognized, there are cases where
Justice Sotomayor’s rule calls for relief and the plurality’s
does not. See Epps, 707 F.3d at 350-51 (citing United States
v. Duvall, 705 F.3d 479, 487-89 (D.C. Cir. 2013) (Williams,
J., concurring in the judgment)). These are cases where the
parties’ agreement rests the sentence on a Guidelines
provision that has been changed (calling for eligibility for
relief under Justice Sotomayor’s opinion), but the court,
though imposing the agreed sentence, rests it upon another
Guideline that the Sentencing Commission has left
undisturbed.
3
Contrary to Judge Kavanaugh’s suggestion, see
Kavanaugh, J., ante at 12-17, Freeman does not satisfy what I
earlier called the “weak” reading of King v. Palmer, see
Duvall, 705 F.3d at 487 (Williams, J., concurring in the
judgment). Cases producing an outcome in favor of the
defendant under Justice Sotomayor’s opinion would not
invariably yield an outcome in his favor under the plurality.
In my concurrence in Duvall, I used an example of the
interplay between the defendant’s alleged status as a career
offender and the sentence associated with the relevant
quantity of crack cocaine. In that hypo, the parties choose a
Guidelines sentence via the crack cocaine quantity but ignore
the career offender status (perhaps because there are
procedural defects in previous convictions not worth the
trouble of litigation), but the sentencing judge finds the career
offender status applicable. They arrive at the same sentence,
and the judge “accepts” the sentence agreed on by the parties
by pronouncing the agreed-upon sentence. The Sentencing
Commission subsequently lowers the penalty associated with
the relevant quantity of crack distribution.
In such a case, the Freeman plurality would deny relief,
as the sentencing court will have made clear that the sentence
was not “based on a sentencing range that has been
subsequently lowered by the Sentencing Commission.” 18
U.S.C. § 3582(c)(2). But Justice Sotomayor would grant
relief—in her view, it is only the agreement that matters. As I
said in my Duvall concurrence, “Cases such as this will occur
any time the parties to a Rule 11(c)(1)(C) agreement agree to
ignore some aspect of an alleged offense that would trigger a
mandatory minimum or a mandatory enhancement that the
sentencing judge deems inappropriate to ignore, but the
agreement yields an ultimate sentence that the judge regards
as otherwise ‘sufficient, but not greater than necessary,’ to
achieve the goals of sentencing as required by 18 U.S.C.
§ 3553(a).” Duvall, 705 F.3d at 488 (Williams, J., concurring
4
in the judgment). Thus the Freeman opinions on § 3582(c)(2)
present the sort of case reflected in the diagram above. In
such cases, the simple vision of relative “narrowness”
becomes a matter of guesswork as to the distribution of cases
among the various possible types.
Of course, even to suppose that such an estimated
quantitative disparity would justify treating the “narrower”
opinion as controlling rests on a dubious simplifying
assumption: that a rule could be characterized as narrower
merely because it called for relief in fewer situations. It is not
obvious why such a purely quantitative approach would
represent a sensible view of “narrow.”
The conditions that we laid out in King for the application
of Marks leave no such ambiguity. Where (1) Rule B calls for
relief in every case where Rule A does, and (2) Rule B calls
for relief in no other cases, Rule B is clearly “narrower” than
Rule A. The situation satisfying these criteria is illustrated
below:
5
Here the proposition that Rule B is narrower than Rule A is
completely independent of the distribution of factual patterns.
It is, in fact, the relationship between the opinions at issue in
Marks (to wit, the opinions in A Book Named “John Cleland’s
Memoirs of a Woman of Pleasure” v. Attorney General of
Massachusetts, 383 U.S. 413 (1966)). By contrast, when the
Supreme Court opinions yielding a result are on different
wavelengths, as in Freeman and in the Supreme Court case at
issue in King (namely Pennsylvania v. Delaware Valley
Citizens’ Council for Clean Air, 487 U.S. 711 (1987)),
characterizing one as narrower than the other is just a kind of
judicial force majeure.
Judge Kavanaugh repeatedly asserts that the Freeman
plurality “always” calls for eligibility under § 3582(c)(2)
because the judge is necessarily applying a Guidelines range.
Opinion of Kavanaugh, J., at 1, 9, 13 n.5. But that assertion
overlooks the requirement of § 3582(c)(2) that the sentence in
question have been based on “a sentencing range that has
been subsequently lowered by the Sentencing
Commission . . . .” In the class of cases discussed above, the
parties’ agreement has based the sentence on a range that has
been lowered (controlling under Justice Sotomayor’s opinion),
but the court has based its sentence (controlling under the
plurality) on one that has not. Accordingly, Justice
Sotomayor’s rule is not exclusively a subset of the plurality’s.
To put it another way, while the plurality may always
find that a sentence was based on a Guidelines range, it is not
the case that in every situation where Justice Sotomayor finds
a sentence to have been based on a Guideline that has been
changed the plurality will do so. To find this fact immaterial
to the application of King to Freeman, as Judge Kavanaugh
appears to do, is to distort the nature of the dispute in
6
Freeman. It would mean that a court purporting to apply
Freeman could, after scrutinizing the plea agreement to
determine whether a defendant’s sentence was based on the
Guidelines, then look either to the plea agreement or to the
analysis of the sentencing judge in assessing whether the
defendant was sentenced within a Guidelines range lowered
by the Commission. That clearly was not what Justice
Sotomayor had in mind; her assessment of defendant
Freeman’s eligibility for a sentence reduction under
§ 3582(c)(2) took no account of the reasoning of the
sentencing judge. See Freeman, 131 S. Ct. at 2699-2700.
Judge Kavanaugh is also mistaken to suggest that
Freeman can be shoehorned under King through use of the
phrase “applicable guideline range” from the Application
Note to § 1B1.10 of the Guidelines. See Opinion of
Kavanaugh, J., at 14-15. How Justice Sotomayor might read
“applicable guideline range,” which is made relevant through
§ 3582(c)(2)’s requirement that the requested sentence
revision be “consistent with applicable policy statements
issued by the Sentencing Commission” (plus the Application
Note itself), is irrelevant to whether Freeman meets King’s
requirements for establishing binding precedent: none of the
opinions in Freeman purports to address the meaning of the
Application Note to § 1B1.10 or its relationship to
§ 3582(c)(2). Rather, Freeman addressed only the
requirement of § 3582(c)(2) that a sentence be “based on a
sentencing range that has subsequently been lowered by the
Sentencing Commission.”
Thus the proper application of King to Freeman need
consider only whether the plurality would find that a
defendant’s sentence was based on a subsequently reduced
Guidelines range in every instance where Justice Sotomayor
would. King—and Marks, for that matter—are concerned
only with the discrete questions actually addressed in the
7
splintered opinions from which precedent is sought. Judge
Kavanaugh’s alternative would find a binding precedent (or
deny one) with respect to one issue based on speculation as to
how a justice might rule on an analytically distinct issue that
the splintered opinions had no cause to examine.
The proper application of King undeniably means that
some Supreme Court decisions yield no binding precedent,
but that reality does not trigger vertical stare decisis concerns
of the sort that trouble Judge Kavanaugh. Such instances are
similar to a 4-4 split that affirms the lower court’s opinion but
does not supply a national rule governing future litigation.
See, e.g., John Wiley & Sons, Inc. v. Kirtsaeng, 654 F.3d 210,
218 (2d Cir. 2011) (citing the lack of binding authority
produced by the 4-4 split in Costco Wholesale Corp. v.
Omega, S.A., 131 S. Ct. 565 (2010)). Of course, as actually
occurred in Wiley, the Supreme Court may follow its
splintered decision with a case that resolves the issue. See
Kirtsaeng v. John Wiley & Sons, Inc., 133 S. Ct. 1351 (2013)
(resolving the issue 6-3).
Moreover, where the Court resolves a case with a
splintered decision and a binding precedent cannot be found
under Marks/King, the disarray among Supreme Court
opinions is in important ways akin to the situation where one
or more (indeed, perhaps all but one) courts of appeals have
resolved an issue one way. In that case it is the duty of a court
of appeals facing the issue de novo to resolve it de novo, with
of course due recognition of the insights and arguments
reflected in the opinions of other courts. That independent
approach allows the issue to “percolate” and facilitates
ultimate Supreme Court resolution on the basis of a broad
pallet of lower court reasoning. Cf. United States v. Mendoza,
464 U.S 154, 160 (1984) (“Allowing only one final
adjudication [on a specified legal issue] would deprive this
Court of the benefit it receives from permitting several courts
8
of appeals to explore a difficult question before this Court
grants certiorari.”). Similarly, where the justices produce
splintered opinions that fail to satisfy King’s “subset”
principle, the process of continued percolation through
independent lower court reasoning yields important value.
Judge Kavanaugh’s quest for binding Supreme Court
precedent leads him to propose that when lower courts are
confronted with such complete disarray that no single view
meets even his standards (see Opinion of Kavanaugh, J., at 7-
8 (alluding to “no-narrowest-opinion cases”)), they should
“strive to decide the case before them in a way consistent with
how the Supreme Court’s opinions in the relevant precedent
would resolve the current case.” Well, of course, that is what
we always try to do. But the question is whether, looking at a
set of opinions that reveal no common core, we should
pretend that they have offered a unified body of coherent
reasoning and treat that synthetic body of reasoning as binding
precedent. Pursuing that approach, lower courts would look
more like lower officials seeking to discern the intent of their
superiors than like judges engaged in discerning and applying
rules of law. Courts are still, or should be, institutions of
reason, not will.
Judge Kavanaugh’s proposed rule for resolving a case
where there is no “narrowest” opinion, which entails
“run[ning] the facts and circumstances of the current case”
through the tests proposed by the various factions of the
splintered decision on a case-by-case basis to produce “the
result that a majority of the Supreme Court would have
reached,” Opinion of Kavanaugh, J., at 7, explicitly
contemplates including the opinions of dissenting justices in
the construction of such a “majority.” See id. at 8 (citing
Rapanos v. United States, 547 U.S. 715, 810 (2006) (Stevens,
J., concurring)), 10 (“When [Justice Sotomayor] concludes
that a plea agreement was not based on the Guidelines, she
9
would agree with the result reached under Chief Justice
Roberts’s opinion for four Justices.”), 16-17 (“[C]ontrary to
Epps’s assertion, Justice Sotomayor’s opinion would not
reach a different result than Justice Kennedy’s opinion or
Chief Justice Roberts’s opinion in those career offender or
mandatory minimum circumstances.”) (emphasis added).
“Foolproof” and “common sense” as Judge Kavanaugh
believes this case-by-case majority-seeking approach to be,
Opinion of Kavanaugh, J., at 8, it is in fact prohibited by King,
which stated plainly that “we do not think we are free to
combine a dissent with a concurrence to form a Marks
majority,” 950 F.2d at 783. It is telling that one of the two
circuits that have embraced the idea of reliance on dissenting
opinions expressly acknowledged that King precludes such an
approach. See United States v. Johnson, 467 F.3d 56, 65 (1st
Cir. 2006). Although some individual justices of the Supreme
Court appear to have endorsed the idea of relying on dissents,
see, e.g., Rapanos, 547 U.S. at 810 (Stevens, J., concurring);
Waters v. Churchill, 511 U.S. 661, 685 (1994) (Souter, J.,
concurring), the Court as a whole has never contradicted
King’s prohibition on counting dissenters. For that reason,
Judge Kavanaugh’s approach is unavailable to this court
(unless we re-en banc King, which Judge Kavanaugh does not
propose). That there is no Supreme Court decision embracing
such a viewpoint is not surprising. Dissenting judges enjoy
something of the liberty of a gadfly, as the outcome does not
in fact depend on what they say. Dissents of course often
prove bellwethers, but until they do so, they may inspire but
not guide.
In short, the relationship between the opinions in
Freeman precludes a finding that the Supreme Court has
provided binding authority on this aspect of § 3582. Judge
Kavanaugh frames his argument as advocating an en banc to
overrule Epps, but in fact his objection is to the principle of
10
King v. Palmer. In its place, he proposes a variation of Marks
that requires a kind of lower-court clairvoyance into the minds
of justices in future Supreme Court majorities. Yet he offers
neither a compelling reason to adopt this new rule, nor
evidence that King has caused the slightest difficulty in the
twenty-two years since its adoption. King in fact supplied
needed clarification for the challenge presented by splintered
Supreme Court decisions; I see no reason to drop it now.