Keith Undray Ford v. SUZANNE M. PEERY, Warden

FOR PUBLICATION
UNITED STATES COURT OF APPEALS
FOR THE NINTH CIRCUIT
KEITH UNDRAY F ORD ,
Petitioner-Appellant,
v.
S UZANNE M. P EERY, Warden,
Respondent-Appellee.
No. 18-15498
D.C. No.
2:15-cv-02463-
MCE-GGH
ORDER
Filed August 18, 2021
Before: William A. Fletcher and Ryan D. Nelson, Circuit
Judges, and Donald W. Molloy, District Judge.*
Order;
Dissent by Judge VanDyke
* The Honorable Donald W. Molloy, United States District Judge
for the District of Montana, sitting by designation.

-- 1 of 26 --

2 FORD V . P EERY
SUMMARY**
Habeas Corpus
In a case in which a judge of this court sua sponte
requested en banc rehearing, and in which appellant filed a
petition for rehearing en banc, the panel on behalf of the
court denied appellant’s petition for rehearing en banc and
denied rehearing en banc.
Judge VanDyke dissented from the denial of rehearing
en banc.
He wrote that this circuit’s cases misapplying AEDPA
deference are legion, but the panel majority took this court’s
habeas dysfunction to a new level by issuing, when first
confronted with an en banc petition, an amended opinion that
opined about how the majority would refuse to defer to a
purely hypothetical state court ruling as to whether the
prosecutor’s comments at the end of closing rebuttal
argument constituted misstatements of law.
As in its now-vacated opinion, the majority began its
analysis in its amended opinion by correctly describing the
issue as whether the prosecutor’s presumption-of-innocence
comments constituted misconduct in violation of due
process under Darden v. Wainright, 477 U.S. 168 (1986).
Judge VanDyke wrote that after laying the appropriate
prejudice-focused foundation for a proper Darden analysis,
the majority then diverted to an odd and lengthy dicta
** This summary constitutes no part of the opinion of the court. It
has been prepared by court staff for the convenience of the reader.

-- 2 of 26 --

FORD V . P EERY 3
discussion. Despite the California Court of Appeal’s explicit
assumption that the prosecutor did misstate the law, and
despite Darden’s primary focus on the prejudicial effect
from any alleged misconduct (and not the misconduct itself),
the majority concluded that “even if there were a state-court
decision holding that the prosecutor did not misstate the law,
we would conclude that such a holding would have been
unreasonable.”
Judge VanDyke wrote that the majority misconstrued the
Darden analysis and misapplied AEDPA—again—in its
hypothetical dicta. He explained that when appropriately
evaluated in context, the prosecutor’s presumption-of-
innocence remarks did not rise to the level of misconduct;
and that given the lack of clearly established law as to the
propriety of a prosecutor’s remarks in this context, even the
majority’s hypothetical state court decision would not
actually “unreasonably” apply any clearly established law.
Judge VanDyke wrote that the majority’s dicta
illuminates the potential for abuse of this court’s “binding
dicta” rule, under which this court views “well-reasoned”
dicta as binding. He emphasized that all of this could have
easily been avoided if the majority had voluntarily removed
its dicta during the first en banc proceedings. He concluded
that instead of starting down a path of issuing advisory
AEDPA fumbles, the court should have taken the panel’s
amended opinion en banc to nip this new practice in the bud,
and to clarify the badly amorphous binding dicta rule.

-- 3 of 26 --

4 FORD V . P EERY
ORDER
A judge of this court sua sponte requested a vote on
whether to rehear this case en banc. The parties were
directed to file simultaneous briefs setting forth their
respective positions as to whether this case should be reheard
en banc. Appellant filed a petition for rehearing en banc
following the panel’s order.
Judges W. Fletcher and R. Nelson have voted to deny the
petition for rehearing en banc, and Judge Molloy has so
recommended. The full court was advised of Appellant’s
petition for rehearing en banc.
A vote was taken on the sua sponte call and the matter
failed to receive a majority of the votes of the non-recused
active judges in favor of en banc consideration. See Fed. R.
App. P. 35(f). Appellant’s petition for rehearing en banc
(Dkt. No. 82) is DENIED. Rehearing en banc is DENIED.
Judge VanDyke’s dissent from the denial of rehearing en
banc is attached and filed concurrently with this order.

-- 4 of 26 --

FORD V . P EERY 5
VANDYKE, Circuit Judge, dissenting from denial of
rehearing en banc:
Our circuit’s cases misapplying AEDPA deference are
legion, and the resultant game of whack-a-mole the Supreme
Court has been forced to play with our habeas decisions is
so well known at this point as to need no supporting
citation.1 But this case takes our habeas dysfunction to a
new level. Initially, the panel majority here refused to
provide AEDPA deference, granting habeas relief to
Petitioner Keith Ford in a split opinion. Confronted with an
en banc petition, the panel was forced to reverse itself,
issuing an amended opinion that, this time, begrudgingly
deferred to the state court’s conclusions on the part of Ford’s
case that mattered, and so appropriately denied habeas relief.
As Judge Nelson observed in his partial dissent from the
panel’s amended opinion, this was a commendable move
that likely saved the panel majority from being reversed
either by our own court en banc or by the Supreme Court.
Ford v. Peery, 999 F.3d 1214, 1227 (9th Cir. 2021) (R.
Nelson, J., dissenting in part and concurring in the
judgment).
If that had been all that the panel majority did, there
would be cause for celebration in the West and hope that
perhaps our court was really turning over a new leaf. But
alas, like a sullen kid who spits in the cookie jar after being
caught red-handed, the panel majority decided that if they
couldn’t get away with directly defying AEDPA in this case,
they could at least opine in their revised opinion about how
1 To give credit where credit is due: my diligent clerk did prepare a
very nice string-cite spanning multiple pages. But including it felt
awkward—like trying to shame a career offender with his rap sheet.

-- 5 of 26 --

6 FORD V . P EERY
they would refuse to defer to a purely hypothetical state court
ruling not presented in this case at all.
This appears to be an entirely new phenomenon. Our
court has a well-documented habit of not properly deferring
to actual state court rulings in AEDPA cases, including a
long list of summary reversals from the Supreme Court. But
I’m not sure I’ve ever seen our court make up a pretend state
court ruling just so it could refuse to apply AEDPA
deference to it while pummeling a strawman of its own
making. Weird.
Have things gotten so bad for my AEDPA-disdaining
colleagues that they are forced to invent stuff that they can
then hypothetically refuse to defer to, secure in the
knowledge that at least those advisory rulings won’t get
reversed? It’s possible. But I tend to think they’re likely
trying to do something more nefarious. Instead of starting
down a new path of issuing advisory AEDPA fumbles, our
court should have taken the panel’s amended opinion en
banc to nip this new practice in the bud. And in doing so,
we could have taken this opportunity to clarify our badly
amorphous “binding dicta” rule,2 making clear that attempts
like the panel majority’s here won’t work. Because we
didn’t, I respectfully dissent from the denial of rehearing en
banc.
2 Though we have referred to this rule as the “[w]ell-reasoned dicta”
rule, see Li v. Holder, 738 F.3d 1160, 1165 n.2 (9th Cir. 2013), the line
between well-reasoned dicta and not-well-reasoned dicta seems to lie
largely in the eye of the beholder. I therefore refer to this rule as the
“binding dicta” rule.

-- 6 of 26 --

FORD V . P EERY 7
I.
The majority’s amended opinion is a product of what
preceded it, so some background is necessary. A jury
convicted Ford of first-degree murder for shooting Ruben
Martinez point-blank in the head. People v. Ford, No.
A137496, 2014 WL 4446166, at *1 (Cal. Ct. App. Sept. 10,
2014). Throughout Ford’s trial, the court repeatedly
admonished the jurors to refrain from deciding any issue in
the case until after the entire case was presented and the jury
was released for deliberations. And immediately prior to
closing arguments, the trial court instructed the jury that “[a]
defendant in a criminal case is presumed to be innocent,”
which “requires that the People prove a defendant guilty
beyond a reasonable doubt.” The trial court also reminded
the jury that “[i]n their . . . closing arguments, the attorneys
discuss the case, but their remarks are not evidence.” It
warned the jury that “[i]f you believe that the attorneys’
comments on the law conflict with my instructions, you must
follow my instructions.” Later, the trial court again
reminded the jury that “[y]ou may not convict the defendant
unless the People have proved his guilt beyond a reasonable
doubt,” that “[i]t is up to you to decide whether an assumed
fact has been proved,” and that no juror should “make up
your mind about the verdict or any issue until after you have
discussed the case with the other jurors during
deliberations.”
In his closing arguments, the prosecutor reiterated the
trial court’s instruction that “[e]vidence is not anything that
I say up here. This is just argument . . . . What I am saying
here is not evidence.” He also informed the jury that “[t]he
real work is going to be starting in a little bit, and that’s
where we’re not involved. It’s when you guys are all back
there together talking about this case.” But for the purposes

-- 7 of 26 --

8 FORD V . P EERY
of his closing, the prosecutor stated, “I’m going to go back
over the facts of this case and show you why I have proven
beyond a reasonable doubt that the defendant committed
murder in this case, beyond a reasonable doubt that he killed
Ruben Martinez . . . .” The prosecutor proceeded to discuss
at length the evidence of Ford’s guilt, all the while
referencing the government’s beyond-a-reasonable-doubt
burden to show that he had met that high burden. “In
combination with all the other information,” he explained, a
statement by Ford during a phone call was “proof beyond a
reasonable doubt.” He concluded that “when you . . . follow
all the evidence and you follow all the law, you’re going to
reach the same conclusion that I asked you to reach at the
beginning of this case that the defendant is guilty of murder
. . . .” Ford’s counsel then delivered her closing arguments,
where she repeatedly emphasized Ford’s presumption of
innocence.
Lastly, the prosecutor began his lengthy closing
argument rebuttal by noting that “[t]his is now my
opportunity just to respond to what [Ford’s counsel] said.”
He acknowledged that “it’s true [the defendant’s counsel]
doesn’t have to present any evidence. It is my burden of
proof,” but that “[w]e’re way past that point. It’s been
proven to you every which way . . . that [palm]print was the
defendant’s . . . .” He continued, “there’s always two sides
to every story . . . . but if that other side is a clearly
unreasonable version of the events, then it’s your job as
jurors to reject that . . . . [T]hat’s part of the beyond-a-
reasonable-doubt analysis that you do in this case.”
“Context is everything. You consider all of the evidence,
not just some of it . . . .” “Bottom line, . . . I’ve provided you
with all the information that you need to feel the abiding
conviction of the truth of these charges. I have provided the
information for you to make that decision” and “to follow

-- 8 of 26 --

FORD V . P EERY 9
through with your promise to not hesitate to convict once the
case has been proven to you beyond a reasonable doubt.
This idea of this presumption of innocence is over. Mr. Ford
had a fair trial. . . . He’s not presumed innocent anymore.”
At this point, Ford’s counsel objected that the prosecutor
“misstate[d] the law.” After a sidebar, the judge overruled
the objection, reasoning that the prosecutors’ comments
were “the final comment in the context of argument. [The
jury has] been reminded continuously that they’re not to
form or express any opinions until after they deliberate with
their fellow jurors, so I don’t think there’s any particular
harm in that and that this was the final argument, closing
argument.” So the prosecutor continued with his closing
rebuttal argument before the jury: “And so we’re past that
point. We’re at the point now where you go back, look at
the information that you have before you . . . . and you
should feel comfortable with your decision. . . . And the
evidence before you, when you take all of that information
together, is that the defendant is guilty of murder.”
Following the closing arguments and before releasing
the jury to deliberate, the trial court provided the jury with
instructions that again reiterated Ford’s presumptive
innocence. By the time the jury went into deliberations, it
had repeatedly received instructions from the trial court, the
prosecutor, and Ford’s counsel reiterating Ford’s
presumption of innocence.
After deliberating, the jury returned with a verdict of
first-degree murder. Ford, 999 F.3d at 1217. The jury could
not reach a verdict on some charged enhancements, which,
as Judge Nelson pointed out in his original dissent, was
consistent with the jury’s willingness to hold the government
to its burden of proof. Ford v. Peery, 976 F.3d 1032, 1049
(9th Cir. 2020) (R. Nelson, J., dissenting), opinion

-- 9 of 26 --

10 FORD V . P EERY
withdrawn and superseded on reh’g, 999 F.3d 1214 (9th Cir.
2021).
Ford appealed to the California Court of Appeal, arguing
that the prosecutor’s comments regarding the presumption
of innocence made at the tail end of the closing rebuttal
argument constituted reversable misconduct. Ford, 2014
WL 4446166, at *6. In evaluating this argument, the Court
of Appeal surveyed several state cases that came to different
conclusions as to whether comments in contexts like this
constituted misstatements of law. Id. at *6–7. Ultimately,
however, the Court of Appeal determined that “[w]e need
not resolve any conflict between [the state cases] because we
conclude any assumed error is harmless under either the state
([People v. Watson, 299 P.2d 243 (Cal. 1956)]) or federal
constitutional standard (see [Chapman v. California 386
U.S. 18, 24 (1967)]).” Id. at *8. Any assumed error was
harmless, it reasoned, because: (a) “[t]he [trial] court
instructed the jury Ford was presumed innocent until the
contrary was proven beyond a reasonable doubt and to
disregard any conflicting statements made by the attorneys
concerning the law”; (b) “the prosecutor repeatedly
reminded the jury of his burden to establish guilt beyond a
reasonable doubt;” and (c) “the evidence of Ford’s guilt was
strong.” Id. As the California Court of Appeal observed,
“[t]he jury was properly informed about the prosecution’s
burden.” Id.
The California Supreme Court denied Ford’s petition for
review, and the California courts and the federal district
court all denied his habeas petitions. Ford, 999 F.3d at 1223.
The federal district court also certified three questions for
appeal—none of which, notably, included Ford’s claim that
the prosecutor’s presumption-of-innocence comments
violated due process under Darden v. Wainwright, 477 U.S.

-- 10 of 26 --

FORD V . P EERY 11
168 (1986).3 The district court thereby concluded that Ford
failed to make “a substantial showing of the denial of a
constitutional right” on this claim. 28 U.S.C. § 2253(c)(2).
In a split opinion with Judge Nelson dissenting, the panel
majority granted a certificate of appealability on Ford’s
Darden claim and granted habeas relief on that claim. Ford,
976 F.3d at 1041, 1045. There is no need to belabor the
errors in its original opinion since the majority has
withdrawn and superseded it with the new opinion. To
summarize though, the majority: (a) manufactured its own
de novo review by ignoring the California Court of Appeal’s
reliance on a harmlessness test identical to how our own
court has applied Darden, while simultaneously
emphasizing irrelevant omissions from the California Court
of Appeal’s analysis, (b) engaged in the same prejudice
analysis as the California Court of Appeal—but just reached
the opposite conclusions on essentially every part of the
analysis; and (c) purported to apply AEDPA deference only
at the tail end of its opinion, while actually just
reincorporating its prior de novo analysis. The majority held
that the prosecutor’s presumption-of-innocence remarks
violated due process under Darden. Id. at 1044. Judge
Nelson wrote a compelling dissent, explaining how the
3 In Darden, a petitioner argued “that the prosecution’s closing
argument . . . rendered his conviction fundamentally unfair.” 477 U.S.
at 178. The Supreme Court acknowledged that the prosecutor’s
comments “undoubtedly were improper,” but determined that “[t]he
relevant question is whether the prosecutors’ comments so infected the
trial with unfairness as to make the resulting conviction a denial of due
process.” Id. at 180 (citation and internal quotation marks omitted). The
Supreme Court then evaluated several factors—including the context
surrounding the prosecutor’s remarks, the trial court’s instructions, and
the weight of the evidence—and ultimately concluded that the trial was
not fundamentally unfair. Id. at 182–83.

-- 11 of 26 --

12 FORD V . P EERY
majority misapplied AEDPA deference and inappropriately
evaluated the prosecutor’s remarks out of context. Id.
at 1045–55 (R. Nelson, J., dissenting). After the panel
issued its opinions, the government petitioned for panel
rehearing or rehearing en banc.
* * *
Up to this point, things had progressed in relatively
typical fashion for our court: a panel majority had published
an opinion blatantly defying the deference required by
AEDPA, the state government petitioned for en banc review,
and various off-panel judges had the opportunity to weigh
in. But in a surprising and initially welcome turn of events,
the panel majority took the unusual step of reversing course,
withdrawing its original opinion, and issuing a new,
superseding opinion that denied habeas relief.
The welcome surprise, however, was short-lived. As in
its now-vacated opinion, the majority began its analysis by
correctly describing the issue as whether the prosecutor’s
presumption of innocence comments constituted
“misconduct in violation of due process under Darden.”
Compare Ford, 999 F.3d at 1224, with Ford, 976 F.3d
at 1041. It also acknowledged that prosecutorial misconduct
“rises to the level of Darden error only if there is a
reasonable probability that it rendered the trial
fundamentally unfair.” Ford, 999 F.3d at 1224 (citation
omitted). The majority further correctly observed that,
because the California Court of Appeal assumed without
deciding that the prosecutor misstated the law, there was no
state-court decision to defer to on that point—which, of
course, was not particularly relevant, given that the majority
had just acknowledged that the key inquiry for a Darden
claim is whether any misstatements actually prejudiced the
defendant.

-- 12 of 26 --

FORD V . P EERY 13
After laying the appropriate prejudice-focused
foundation for a proper Darden analysis, the majority then
diverted to an odd and lengthy dicta discussion. Despite the
California Court of Appeal’s explicit assumption that the
prosecutor did misstate the law, and despite Darden’s
primary focus on the prejudicial effect from any alleged
misconduct (and not the misconduct itself), the majority
concluded that “even if there were a state-court decision
holding that the prosecutor did not misstate the law, we
would conclude that such a holding would have been
unreasonable.” Id. (emphasis added). But no one contended
(or now contends) that the state court did anything other than
what it did, which was to simply assume that the
prosecutor’s presumption-of-innocence remarks constituted
misstatements of law so it could address the heart of the
Darden inquiry: prejudice. And no party urged the majority
to stray from the actual issues presented in the case and make
up its own hypothetical state court record to create its own
faux controversy.
After informing the world what it would conclude if
presented with a legal question not actually present in this
case, the majority engaged in an AEDPA-like analysis by
purporting to evaluate the reasonableness of its made-up
state court decision. Id. at 1224–25. It launched into this
“analysis” by quoting a single statement from the
prosecutor’s lengthy closing remarks, with that sole
statement forming the entire factual basis for the majority’s
analysis of the made-up issue. See id. at 1224. Nowhere
does the majority mention the trial court’s repeated
admonitions to the jury to refrain from forming any opinions
until deliberations, or the trial court’s several instructions
regarding the presumption of innocence. Nor does the
majority mention any part of the context immediately
surrounding the prosecutor’s remarks, or both counsels’

-- 13 of 26 --

14 FORD V . P EERY
emphasis throughout their closing arguments on the
prosecution’s beyond-a-reasonable-doubt burden.
After plucking one statement out of the prosecutor’s
entire closing arguments, the majority cited several Supreme
Court cases that discuss the general importance of the
presumption of innocence. Id. at 1225. Later, in language
meant to resemble the AEDPA standard, the majority
concluded that the prosecutor’s presumption-of-innocence
remarks “were misstatements of clearly established law as
articulated by the Supreme Court.” Id. at 1227. But the
Supreme Court cases cited by the majority in its
presumption-of-innocence analysis have nothing to do with
a prosecutor’s remarks in the context of closing arguments.4
So the Supreme Court cases cited by the majority can’t
“squarely address the issue in the case or establish a legal
principle that clearly extends to [this] context.” Moses v.
Payne, 555 F.3d 742, 754 (9th Cir. 2009) (cleaned up). And,
perhaps most fundamentally, the majority simply ignored
constitutionally imposed judicial restraints that limit our
review to actual controversies.
After concluding this bizarre and gratuitous frolic, the
majority jumped back to its Darden analysis by
acknowledging that “[a] violation of due process under
Darden requires more than a prosecutorial misstatement.”
Ford, 999 F.3d at 1225. In contrast to its prior opinion, the
majority now correctly concluded that California’s “Watson
standard is indistinguishable from the Darden ‘reasonable
4 Id. at 1224–25 (citing Betterman v. Montana, 136 S. Ct. 1609,
1612, 1618 (2016); Delo v. Lashley, 507 U.S. 272, 278 (1993) (per
curiam); Herrera v. Collins, 506 U.S. 390, 399 (1993); Reed v. Ross,
468 U.S. 1, 4–5 (1984); Estelle v. Williams, 425 U.S. 501, 503 (1976);
Coffin v. United States, 156 U.S. 432, 453 (1895)).

-- 14 of 26 --

FORD V . P EERY 15
probability’ standard.” Id. Because the California Court of
Appeal applied Watson in concluding that the prosecutor’s
remarks were harmless, the majority “therefore conclude[d]
that the Court of Appeal applied the functional equivalent of
the Darden harmlessness test in holding that the prosecutor’s
statement was harmless.” Id. And because the majority was
“required to give deference to the decision of the Court of
Appeal that the prosecutor’s misstatements were harmless
under the Darden standard,” the majority held that “a
reasonable jurist could have concluded that there was no
reasonable probability that, in the absence of the
prosecutor’s statements that the presumption of innocence
was ‘over,’ the jury would have reached a different
conclusion.” Id. at 1226.
Judge Nelson continued to dissent in part, explaining that
the majority’s “reversal on rehearing[] is only half noble.”
Id. at 1227 (R. Nelson, J., dissenting in part and concurring
in the judgment). Judge Nelson observed that the majority
reached its conclusion regarding the prosecutor’s comments
under a “make-believe, hypothetical de novo review,” and
that “[i]n context, the comments do not rise to the level of
prosecutorial misconduct.” Id. at 1228, 1229. “The
prosecutor made numerous statements supporting the more
reasonable interpretation (still largely ignored by the
majority on rehearing)” that the presumption of innocence
still applied, but the prosecutor had just successfully rebutted
it. Id. at 1230. And “[t]he surrounding context of the
prosecutor’s statements also explains the trial court’s
decision to overrule defense counsel’s objection to the
contested statements.” Id. at 1231. “In short, no reasonable
juror would interpret the prosecutor’s statements, in context,
consistent with the majority’s isolated gloss.” Id. And
“[e]ven assuming the prosecutor’s statements viewed in
context rose to the level of a misstatement of clearly

-- 15 of 26 --

16 FORD V . P EERY
established Supreme Court precedent,” Judge Nelson
concluded, “the statements are harmless under any
standard.” Id. He therefore concurred in the judgment but
“disagree[d] with the decision to grant the COA and much
of the majority’s convoluted reasoning.” Id. at 1233.
II.
The majority’s misstatement-of-law dicta misconstrues
the Darden inquiry, is substantively wrong, misapplies
AEDPA deference, and inappropriately predicates its
conclusion on a made-up state-court decision. Given the
dicta’s utter irrelevancy to the prejudice-focused Darden
analysis or the record before it, the majority’s insistence on
its inclusion in its revised opinion suggests that the dicta is
not really meant for this case. Its refusal to remove the dicta
evinces a more sinister motivation: after being forced to
withdraw its original, deeply flawed opinion, the panel
majority is seeking to wring some drop of lemonade from
what it now perceives to be a lemon of a case by squeezing
a completely advisory rule into its revised opinion. In doing
so, the majority is trying to lay the groundwork for future
AEDPA cases by proffering dicta that it hopes will simply
be accepted as binding by future parties and panels of our
court.
This would not be such a problem were it not for our
circuit’s hopelessly vague binding dicta rule. The panel
majority’s maneuvering here highlights the potential for
abuse of this rule: panels of our court can sneak utterly
irrelevant and erroneous dicta into published opinions, with
little threat of challenge (or even close review) given its
minimal impact on that case’s outcome, yet potentially
create new law in our circuit. This case presents a
particularly egregious and completely indefensible example
of the rule’s potential abuse and therefore warranted our en

-- 16 of 26 --

FORD V . P EERY 17
banc court’s further scrutiny and clarification of the rule’s
limits.
A. The Majority Misconstrues The Darden Analysis And
Misapplies AEDPA—Again—In Its Hypothetical
Dicta.
First, the majority’s misstatement-of-law analysis is
obvious dicta in light of the state court’s assumption that the
prosecutor here did misstate the law. When evaluating a
Darden claim, “[t]he relevant question is whether the
prosecutors’ comments so infected the trial with unfairness
as to make the resulting conviction a denial of due process.”
Darden, 477 U.S. at 181. The Supreme Court in Darden
merely asserted without explanation that the prosecutors’
remarks in that case were unlawful, id. at 180 (“These
comments undoubtedly were improper.”), and then spent all
its analysis on the comments’ prejudicial effects. Id. at 180–
82 (citation omitted). Thus, especially under AEDPA, “the
Darden standard is a very general one, leaving courts more
leeway . . . in reaching outcomes in case-by-case
determinations.” Parker v. Matthews, 567 U.S. 37, 48
(2012) (per curiam) (citation and internal quotation marks
omitted); see also Rowland v. Chappell, 876 F.3d 1174,
1189 (9th Cir. 2017).
Indeed, given the “[t]he highly generalized standard for
evaluating claims of prosecutorial misconduct set forth in
Darden,” the Supreme Court has expressly castigated a sister
circuit for imposing an “elaborate, multistep test” on state
courts when conducting a Darden analysis on AEDPA
review. Parker, 567 U.S. at 49. Because Darden focuses
primarily on the prejudicial effect of alleged prosecutorial
misconduct, here the majority’s misstatement-of-law
analysis in its revised opinion is “clearly unnecessary to its
resolution of the case, does not affect its outcome in any

-- 17 of 26 --

18 FORD V . P EERY
manner, . . . constitutes an advisory opinion” and is “entirely
dicta.” Spears v. Stewart, 283 F.3d 992, 998–99 (9th Cir.
2002) (Reinhardt, J., joined by Hawkins, Thomas, and Paez,
JJ., and joined in part by, inter alios, W. Fletcher, J.,
dissenting from denial of rehearing en banc). It is especially
ironic that the author of the majority opinion here, Judge
Fletcher, previously agreed that when faced with such
clearly unnecessary dicta, “the panel could and should have
avoided the highly controversial question it unnecessarily
reached out for and purported to decide.” Id. at 999. One
wonders: does Judge Fletcher now disagree with his earlier
self, or is something other than a principled position about
dicta driving his conflicting positions in these cases? Given
his dicta’s utter irrelevancy to this case’s outcome, the only
reason that the panel majority would insist on its inclusion is
to hopefully bind future panels with its new rule under our
court’s binding dicta rule. More on that later.
Second, the majority’s misstatement-of-law dicta is
wrong for the reasons well-explained by Judge Nelson’s
dissent. Ford, 999 F.3d at 1227–33 (R. Nelson, J., dissenting
in part and concurring in judgment). When appropriately
evaluated in context—context that the majority studiously
ignores—the prosecutor’s presumption-of-innocence
remarks did not rise to the level of misconduct. The
Supreme Court has repeatedly made clear that context is
critically important when evaluating a prosecutor’s
statements. See Boyde v. California, 494 U.S. 370, 385
(1990) (“[T]he arguments of counsel . . . must be judged in
the context in which they are made.”); Darden, 477 U.S.
at 179 (“It is helpful as an initial matter to place these
remarks [in the prosecution’s closing argument] in
context.”). In context, the prosecutor’s presumption-of-
innocence remarks don’t mean what they could mean in
isolation. Here, the context surrounding the prosecutor’s

-- 18 of 26 --

FORD V . P EERY 19
statements shows that the prosecutor was informing the jury
that he had met his beyond-a-reasonable-doubt burden, and
that a reasonable juror would have perceived the comments
(again, in context) as such. The statements’ surrounding
context, combined with the trial court’s and the prosecutor’s
repeated references to the government’s high burden, reveal
that the prosecutor’s statements did not impinge on Ford’s
presumption of innocence, much less contradict or
unreasonably apply any clearly established law on the issue.
Third, the majority’s conclusion that a hypothetical state
court decision would fail even under AEDPA’s deferential
standard renders its dicta especially problematic. The
majority purports to conduct an AEDPA analysis by
concluding that its hypothetical state court decision would
have been “unreasonable” (as opposed to simply erroneous)
and that the prosecutor’s presumption-of-innocence remarks
were “misstatements of clearly established law as articulated
by the Supreme Court,” Ford, 999 F.3d at 1224, 1227; cf. 28
U.S.C. § 2254(d)(1). But as discussed, the Supreme Court
in Darden merely asserted without explanation that the
challenged statements “undoubtedly were improper,” and
spent all of the Court’s analysis in that case on the
statements’ prejudicial effect. 477 U.S. at 180–82. And
none of the presumption-of-innocence cases cited by the
majority concern the propriety of a prosecutor’s remarks
made in the context of a closing argument. See supra n.4
and accompanying text. Because these cases do “not
squarely address the issue in [this] case or establish a legal
principle that clearly extends to a new context . . . , it cannot
be said, under AEDPA, there is clearly established Supreme
Court precedent addressing the issue before us, and so we
must defer to the state court’s decision.” Robertson v.
Pichon, 849 F.3d 1173, 1182 (9th Cir. 2017) (cleaned up)
(citation omitted). Given the lack of clearly established law

-- 19 of 26 --

20 FORD V . P EERY
as to the propriety of a prosecutor’s remarks in this context,
even the majority’s hypothetical state court decision would
not actually “unreasonably” apply any clearly established
law. See Wright v. Van Patten, 552 U.S. 120, 126 (2008)
(per curiam) (“Because our cases give no clear answer to the
question presented, let alone one in [the petitioner]’s favor,
it cannot be said that the state court unreasonably applied
clearly established Federal law.” (cleaned up) (citation
omitted)). In short, the majority reaches out to opine on a
completely hypothetical issue subject to full AEDPA
deference—and then refuses to actually give the appropriate
deference in its advisory analysis.
Fourth, the majority’s erroneous AEDPA application to
its own hypothetical reveals the depth of the sickness that
afflicts our court when it comes to our habeas jurisprudence.
Not satisfied with merely defying AEDPA’s strict
parameters (which our court habitually does), the majority
now brushes aside the inconvenient constitutional
boundaries of Article III to make up its own imaginary
controversy on which to misapply AEDPA deference. Cf.
Princeton Univ. v. Schmid, 455 U.S. 100, 102 (1982) (per
curiam) (“We do not sit to decide hypothetical issues or to
give advisory opinions about issues as to which there are not
adverse parties before us.”); Thomas v. Anchorage Equal
Rts. Comm’n, 220 F.3d 1134, 1139 (9th Cir. 2000) (en banc)
(“[T]he Constitution mandates that prior to our exercise of
jurisdiction there exist a constitutional case or controversy,
that the issues presented are definite and concrete, not
hypothetical or abstract.” (citation and internal quotation
marks omitted)). No one disputes that the state court
assumed that the prosecutor misstated the law in this case.
And no party was so audacious as to request an advisory
opinion from us on the majority’s hypothetical. There is
simply no controversy on which we should opine with

-- 20 of 26 --

FORD V . P EERY 21
respect to the hypothetical that the majority raises sua sponte
in its new opinion. See id.
Given the majority’s plucking of the prosecutor’s
remarks out of context, misapplication of AEDPA
deference, and frolicking beyond the constitutional
parameters of our jurisdiction, to say that the majority’s
misstatement-of-law analysis is fraught with error is an
understatement. It’s also utterly irrelevant dicta under the
Darden analysis, yet the majority refused to remove it. The
inevitable question is: Why? Why was the majority so
insistent on including an irrelevant and erroneous discussion
in its opinion?
B. The Majority’s Dicta Illuminates Our Binding Dicta
Rule’s Potential For Abuse.
The probable answer lies, unfortunately, in our court’s
binding dicta rule. Under this rule, our court views “well-
reasoned” dicta as binding. See Li, 738 F.3d at 1165 n.2.5
5 As support for the proposition that “[w]ell-reasoned dicta is the
law of the circuit,” id., Li cites to Judge Kozinski’s separate, plurality
opinion in United States v. Johnson, 256 F.3d 895, 914 (9th Cir.2001)
(en banc) (per curiam) (Kozinski, J., concurring, joined in relevant part
by Trott, T.G. Nelson, and Silverman, JJ.). Li, 738 F.3d at 1165 n.2. But
only three other judges out of the 11-judge en banc court joined the
portion of Judge Kozinski’s Johnson plurality opinion discussing well-
reasoned dicta. See Johnson, 256 F.3d at 914; see also Alcoa, Inc. v.
Bonneville Power Admin., 698 F.3d 774, 796 (9th Cir. 2012) (Tashima,
J., concurring). Four years after Johnson, however, an en banc panel
majority in Barapind v. Enomoto, 400 F.3d 744, 750–51 (9th Cir. 2005)
(en banc) (per curiam), determined that when an issue “was . . . presented
for review” and the panel “addressed the issue and decided it in an
opinion joined in relevant part by a majority of the panel,” then the
panel’s “articulation of [the issue] became law of the circuit, regardless
of whether it was in some technical sense ‘necessary’ to our disposition

-- 21 of 26 --

22 FORD V . P EERY
Specifically, “[w]here a panel confronts an issue germane to
the eventual resolution of the case, and resolves it after
reasoned consideration in a published opinion, that ruling
becomes the law of the circuit, regardless of whether doing
so is necessary in some strict logical sense.” McAdory,
935 F.3d at 843 (opinion of Hawkins, J., joined by W.
Fletcher and Bennett, JJ.) (quoting Cetacean Cmty. v. Bush,
386 F.3d 1169, 1173 (9th Cir. 2004)). “[B]ut we are not
bound by a prior panel’s comments made casually and
without analysis, uttered in passing without due
consideration of the alternatives, or done as a prelude to
another legal issue that commands the panel’s full attention.”
McAdory, 935 F.3d at 843 (citation omitted) (cleaned up).
“[T]he only dicta by which we are bound is well-reasoned
dicta.” Alcoa, Inc., 698 F.3d at 796 (citation and internal
quotations omitted).6
Our binding dicta rule, while no doubt well-intentioned,
has serious difficulties. The rule was originally established
with the salutary goal of preventing judges from casting
aside binding precedent just by labeling it as “dicta.” Cf.
Spears, 283 F.3d at 1006 (“But it is quite a different matter
to suggest, as do Judges Reinhardt and Tashima, that the
of the case.” Barapind, 400 F.3d at 750–51. A three-judge panel has
subsequently relied on Johnson and Barapind in determining that
reasoned considerations of issues “germane to the eventual resolution of
the case” are binding. United States v. McAdory, 935 F.3d 838, 843 (9th
Cir. 2019).
6 Although Judge Tashima stated this rule in his concurrence, he
wrote separately to note that he also “concur[red] in Judge Bea’s
interpretation of Miranda B. v. Kitzhaber, 328 F.3d 1181, 1186–87 (9th
Cir. 2003), that the only dicta by which we are bound is well-reasoned
dicta.” Alcoa, Inc., 698 F.3d at 796 (citation and internal quotation
marks omitted)).

-- 22 of 26 --

FORD V . P EERY 23
work product of a panel of this court can simply be
disregarded because a later panel finds a way to call it ‘dicta’
or ‘advisory’ or some similar invective.”); Johnson,
256 F.3d at 915 (“If later panels could dismiss the work
product of earlier panels quite so easily, much of our circuit
law would be put in doubt.”).
The first difficulty with our binding dicta rule is that,
while the problem it seeks to address is real, it doesn’t
actually prevent judges so inclined from simply ignoring
precedent by recharacterizing it as merely “dicta.” See, e.g.,
E. Bay Sanctuary Covenant v. Biden, 993 F.3d 640, 698 (9th
Cir. 2021) (VanDyke, J., dissenting from the denial of
rehearing en banc) (recounting a panel majority’s bare
assertion that binding precedent was “dicta” so it could
ignore inconvenient precedent). Moreover, since the
binding dicta rule’s inception, several judges on our court
have raised legitimate concerns about its scope—and
specifically, its tension with Article III.7
7 See Barapind, 400 F.3d at 759 (9th Cir. 2005) (Rymer, J., joined
by Kleinfeld, Tallman, Rawlinson, and Callahan, JJ., concurring in the
judgment in part and dissenting in part) (arguing that “[i]t is one thing
for a court of last resort to announce that whatever it says in a published
opinion is binding. . . . It is another for an intermediate court such as
ours to make every reasoned discussion in a published opinion binding
whether it is necessary or not,” and raising further concerns about the
rule’s tension with Article III); see also Nat’l Fed’n of the Blind v. United
Airlines Inc., 813 F.3d 718, 745–46 (9th Cir. 2016) (Kleinfeld, J.,
concurring) (arguing that Part I of the majority’s opinion in that case “is
a prime example of what Judge Rymer, in her dissent in Barapind, called
overwriting invited by the Barapind majority opinion. The Constitution
gives us authority to decide only ‘Cases and Controversies.’ The federal
courts do not have authority to issue advisory opinions. Yet that is what
Part I is.” (citation omitted)); Irons v. Carey, 506 F.3d 951, 952 (9th Cir.
2007) (Kleinfeld, J., joined by Bea, J., dissenting from denial of

-- 23 of 26 --

24 FORD V . P EERY
And well-intentioned as it may be, the rule creates
confusion: it is unclear what qualifies as “well-reasoned,”
and even the articulation of the rule itself has changed over
time. Compare Spears, 283 F.3d at 1006 (“[S]o long as the
issue is presented in the case and expressly addressed in the
opinion, that holding is binding and cannot be overlooked or
ignored by later panels of this court or by other courts of the
circuit.”), with McAdory, 935 F.3d at 843 (“Even if [an issue]
was not before us in that case, our conclusion with respect to
[that issue] is the very type of ‘well-reasoned dicta’ by which
we are bound.”). This is because the rule carries with it a
healthy dose of an in-the-eye-of-the-beholder quality. For
example, when Judge Fletcher (the author of the majority
opinion here) did not like the substance of dicta in another
published opinion, he joined the parts of a dissental in which
Judge Reinhardt argued that a panel’s advisory declarations
were “wholly improper dicta” that were “clearly
unnecessary to its resolution of the case.” Spears, 283 F.3d
at 998–99 (Reinhardt, J., joined by Hawkins, Thomas, and
Paez, JJ., and joined in part by, inter alios, W. Fletcher, J.,
dissenting from denial of rehearing en banc). But in the
majority opinion here, Judge Fletcher includes the very type
of “clearly unnecessary” statements he previously objected
to. See id. And in another opinion, Judge Fletcher agreed
that, pursuant to the binding dicta rule, dicta from another
case was dispositive in his panel’s case—even though a
conviction under the statutory provision at issue in his
panel’s case was not before the panel in the prior case. See
McAdory, 935 F.3d at 843–44. Our judges’ variable
treatment of “dicta” demonstrates that the binding dicta
rehearing en banc) (“The traditional view, which we seem to have
rejected in Barapind, is that since we are empowered only to decide
cases, not to legislate, only those principles necessary to the decision are
binding law of the circuit.”).

-- 24 of 26 --

FORD V . P EERY 25
rule’s inherent malleability invites inconsistent application
and can be used as a tool to drive results-oriented outcomes.
The evolution of the majority’s irrelevant and erroneous
dicta in this case especially illuminates the rule’s potential
for abuse. When called out for its egregious defiance of
AEDPA deference in its original opinion, the majority was
forced to retreat from its original position. Not content to
simply retreat, the majority decided to try to take a hostage
or two on the way. Its new opinion denied habeas relief but
baked in clearly unnecessary dicta, in an apparent attempt to
create a new rule for future AEDPA cases regarding the
propriety of a prosecutor’s presumption-of-innocence
remarks. To be sure, the rule should not apply given the
dicta’s obvious lack of “reasoned consideration,” McAdory,
935 F.3d at 843, but the majority buffered it with just enough
faux analysis to allow a future sympathetic panel to seize and
use it as dispositive to its own AEDPA case. And because
the majority slipped in this dicta while now ultimately
denying habeas relief, it could safely bet that no one would
object—least of all the government, which was now the
prevailing party. But in a future case where this issue is
actually presented, a panel that is sympathetic to this panel
majority’s dicta here could simply rely on that dicta as
“binding,” and not have to provide its own well-reasoned
defense of the rule it applies. The law should not be made
this way, but the majority’s insistence on its dicta’s inclusion
in its new opinion illuminates that the binding dicta rule
creates the possibility for exactly that type of judicial
mischief.
III.
All of this could have easily been avoided if the majority
had voluntarily removed its dicta during the first en banc
proceedings. See Marino v. Ocwen Loan Servicing LLC,

-- 25 of 26 --

26 FORD V . P EERY
978 F.3d 669, 679 (9th Cir. 2020) (Bea, J., concurring)
(“[R]ather than foment claims and arguments as to whether
the majority’s dicta gave reasoned consideration to an issue,
or whether the posited hypothetical was germane to eventual
resolution of this case, it would be better to edit out that
hypothetical.” (internal quotation marks omitted)). The
panel majority’s refusal to do so is powerful evidence that it
wanted to promote an advisory “rule” that it hoped would
nonetheless somehow be deemed binding in the future. If
panel majorities won’t discipline themselves, our court as a
whole should encourage them to do so. It would foster
respect for our court and allow the Supreme Court to focus
on business more important than cleaning up our unforced
errors. See Circuit Scorecard, SCOTUSBLOG,
https://www.scotusblog.com/statistics/ (last visited Aug. 4,
2021) (showing our court was reversed 15 out of 16 times in
the October 2020 term, three times more than any other
circuit).

-- 26 of 26 --