Richard Knight v. Florida Department of Corrections

Case: 18-13390 Date Filed: 08/30/2019 Page: 1 of 28

[PUBLISH]

IN THE UNITED STATES COURT OF APPEALS

FOR THE ELEVENTH CIRCUIT
________________________

No. 18-13390
________________________

D.C. Docket No. 0:17-cv-61921-RNS

RICHARD KNIGHT,

Petitioner-Appellant,

versus

FLORIDA DEPARTMENT OF CORRECTIONS,

Respondent-Appellee.
________________________

Appeal from the United States District Court
for the Southern District of Florida
________________________

(August 30, 2019)

Before TJOFLAT, JORDAN, and GRANT, Circuit Judges.
GRANT, Circuit Judge:
Richard Knight, a Florida prisoner sentenced to death for the murders of

Odessia Stephens and her daughter, Hanessia Mullings, appeals the district court’s
denial of his federal habeas corpus petition. At this stage—almost 20 years after
 Case: 18-13390 Date Filed: 08/30/2019 Page: 2 of 28

the crimes were committed and more than a decade after a Florida jury found
Knight guilty of the murders and recommended a death sentence—Knight’s claims

have been winnowed down to two: first, that his death sentence is invalid under
Hurst v. Florida, 136 S. Ct. 616 (2016), and second, that he received ineffective
assistance of counsel at trial. Because Hurst does not apply retroactively to

Knight, any challenge to his death sentence on that basis is beyond our reach on
federal habeas review. Nor can Knight find success in his other challenge; the
Florida Supreme Court’s rejection of his ineffective-assistance claim was not an

unreasonable application of Strickland v. Washington, 466 U.S. 668 (1984). We
therefore affirm.
I.

A.
According to evidence introduced at his murder trial, Knight lived in an
apartment with his cousin, Hans Mullings, and Hans’s girlfriend, Odessia. Hans
and Odessia’s four-year-old daughter, Hanessia, also lived with them in the
apartment. Odessia was tired of supporting Knight and one evening while Hans
was out she argued with Knight, insisting that he move out the next day. After the
argument got heated, Knight left the house to walk around. But as he later
confessed to another inmate, instead of getting less angry with Odessia once he got
some air, Knight became increasingly irate. He returned to the apartment and after
exchanging more words with Odessia, he got a knife from the kitchen. When he
went back to the master bedroom, he found Odessia and her little girl in the bed.
He began stabbing Odessia and continued his attack until she stopped resisting and

2
 Case: 18-13390 Date Filed: 08/30/2019 Page: 3 of 28

curled up on the bedroom floor. He then moved on to little Hanessia, stabbing her
until his knife broke and cutting his hand in the process. As he was leaving the

bedroom, he heard “popping noises” from where Hanessia lay on the floor, and he
thought that the little girl was “drowning in her own blood.” Apparently not
considering his revenge complete, he retrieved a second knife from the kitchen and

returned to continue his attack on Odessia. In the meantime, Odessia had crawled
from the bedroom to the living room, where she had collapsed. Knight turned her
over, saw that she was still alive, and started stabbing her again.

Both Odessia and Hanessia died that night. In total, Odessia had 21 stab
wounds, including 14 in the neck, 24 puncture or scratch wounds, bruising and
ligature marks consistent with having been hit and strangled with a belt, defensive

wounds, and bruises from being hit or punched in the mouth and head. Little
Hanessia had four stab wounds in her upper body and neck, a deep defensive
wound on her hand, bruises on her neck consistent with manual strangulation, and
bruises on her arms consistent with having been grabbed.
Knight showered and changed after completing his brutal acts, then headed
to the living room with a rag to wipe off the knives. Interrupted by a knock on the
front door—it was police responding to a neighbor’s 911 call—Knight ran to his
room and climbed out the window.
Shortly after they arrived, police encountered Knight near the apartment. He
told them that he lived there, but that he did not have a key. This was odd; the
officers had already found that all the doors to the apartment were locked. Knight
was also visibly wet—but it was not raining. Knight explained to police that he

3
 Case: 18-13390 Date Filed: 08/30/2019 Page: 4 of 28

had been jogging, a remarkable contention from a person who was wearing long
pants and dress shoes. He did not appear to be sweating, in any event. And

Knight’s personal appearance subsequently revealed even more clues—he had
blood on the back of his shirt, scratches on his chest and midsection, a scrape on
his shoulder, and fresh cuts on his hand.

Knight was arrested and indicted for two counts of first-degree murder. A
Florida jury found him guilty as charged. That same jury heard evidence and
argument at the penalty phase and unanimously recommended two death

sentences—one for each murder. Consistent with Florida’s then-current death
penalty sentencing procedure, the judge held an additional hearing, made his own
findings regarding aggravating and mitigating circumstances, and sentenced

Knight to death. The Florida Supreme Court affirmed Knight’s convictions and
sentences on direct appeal. Knight v. State, 76 So. 3d 879, 890 (Fla. 2011). The
United States Supreme Court denied his petition for certiorari. Knight v. Florida,
566 U.S. 998, 998 (2012).
B.
Knight filed motions for state collateral relief raising the two claims at issue
here, as well as others that have already been resolved. Specifically, he argued that
the state court should vacate his death sentence in light of Hurst v. Florida, in
which the Supreme Court held—four years after Knight’s conviction was final—
that Florida’s death penalty sentencing scheme violated the Sixth Amendment.
136 S. Ct. at 622. The problem identified by the Supreme Court in Hurst, and
argued by Knight in his post-conviction pleadings, was that the jury’s role in

4
 Case: 18-13390 Date Filed: 08/30/2019 Page: 5 of 28

sentencing was to make a non-binding recommendation; the judge alone made the
ultimate findings of fact necessary to impose the death penalty. Id. at 619, 621–22.

Knight also argued that his guilt-phase counsel was constitutionally ineffective for
failing to call an available DNA expert.
The Florida Supreme Court rejected his postconviction claims on the merits.

Knight v. State, 225 So. 3d 661, 668 (Fla. 2017) (per curiam). A plurality of the
court agreed with Knight that the sentencing procedure used in his case violated
the Sixth Amendment under Hurst, but also concluded that the Hurst error was

harmless. Id. at 682. The plurality explained that under the facts of Knight’s case
the penalty-phase jury had necessarily made the factual findings necessary to
impose the death penalty—that “sufficient aggravators existed” and that “the

aggravation outweighed the mitigation”—when it returned a unanimous vote
recommending death.1 Id. at 682–83 (citation omitted). As for his ineffective-
assistance claim, the court held that Knight had failed to meet his burden under
Strickland because he had not shown that his attorney’s decision not to call his
DNA expert constituted deficient performance, or that there was any reasonable
probability that that decision negatively affected the outcome of his trial. Id. at
673–74.

1
Three out of seven justices joined the opinion on Knight’s Hurst claim. Two additional justices
concurred in the result only. Knight, 225 So. 3d at 684.
5
 Case: 18-13390 Date Filed: 08/30/2019 Page: 6 of 28

C.
Knight filed a petition for federal habeas review in the Southern District of

Florida, pursuant to 28 U.S.C. § 2254. The district court denied relief but granted
a certificate of appealability on the two claims now before us.
II.

A.
Federal courts may grant habeas corpus relief to prisoners who are being
detained “in violation of the Constitution or laws or treaties of the United States.”

28 U.S.C. §§ 2241(c)(3); 2254(a). But our authority to award this kind of relief to
state prisoners is limited—by both statute and Supreme Court precedent.
First, the Antiterrorism and Effective Death Penalty Act (AEDPA) limits

our authority to award habeas relief. A federal court may not grant a state
prisoner’s habeas petition on any issue that was decided on the merits by the state
court unless the state court’s ruling “(1) resulted in a decision that was contrary to,
or involved an unreasonable application of, clearly established Federal law, as
determined by the Supreme Court of the United States; or (2) resulted in a decision
that was based on an unreasonable determination of the facts in light of the
evidence presented in the State court proceeding.” 28 U.S.C. § 2254(d). And as
the Supreme Court has explained, “clearly established” federal law means “the
holdings, as opposed to the dicta” from its controlling precedents at the time of the
relevant state court decision. Lockyer v. Andrade, 538 U.S. 63, 71 (2003) (quoting
Williams v. Taylor, 529 U.S. 362, 412 (2000)).

6
 Case: 18-13390 Date Filed: 08/30/2019 Page: 7 of 28

A decision is “contrary to” clearly established federal law “if the state court
arrives at a conclusion opposite to that reached by [the Supreme] Court on a

question of law or if the state court decides a case differently than [the Supreme]
Court has on a set of materially indistinguishable facts.” Williams, 529 U.S. at
413. A state court decision involves an unreasonable application of federal law “if

the state court identifies the correct governing legal principle from [the Supreme]
Court’s decisions but unreasonably applies that principle to the facts of the
prisoner’s case.” Id. To justify issuance of the writ under the “unreasonable

application” clause, the state court’s application of Supreme Court precedent must
be more than just wrong in the eyes of the federal court; it “must be ‘objectively
unreasonable.’” Virginia v. LeBlanc, 137 S. Ct. 1726, 1728 (2017) (quoting

Woods v. Donald, 135 S. Ct. 1372, 1376 (2015)); see also Bell v. Cone, 535 U.S.
685, 694 (2002) (explaining that “an unreasonable application is different from an
incorrect one.”).
Second, Supreme Court precedent demands that in any federal habeas
proceeding—including collateral proceedings in capital cases—where the
petitioner seeks the benefit of a “new” rule of constitutional law, we must first
determine whether the rule actually qualifies as new, and then whether that rule
applies retroactively to the case. See Teague v. Lane, 489 U.S. 288, 300–01 (1989)
(plurality opinion); Penry v. Lynaugh, 492 U.S. 302, 313–14 (1989) (stating that
the retroactivity approach from Teague applies in capital cases), abrogated on
other grounds by Atkins v. Virginia, 536 U.S. 304, 312–16, 321 (2002). In most
cases, we cannot disturb a state conviction based on a constitutional rule

7
 Case: 18-13390 Date Filed: 08/30/2019 Page: 8 of 28

announced after a conviction became final. Teague, 489 U.S. at 310. Only two
narrow exceptions pierce this general principle of nonretroactivity: new rules that

are “substantive rather than procedural,” and “watershed rules of criminal
procedure implicating the fundamental fairness and accuracy of the criminal
proceeding.” Schriro v. Summerlin, 542 U.S. 348, 352–53 (2004) (internal

quotation marks and citations omitted). In all other cases the rule applies only
prospectively.
What this means in plain English is that, in the vast majority of cases,

prisoners will not be able to secure federal habeas relief based on a new
constitutional rule—even when that rule runs in their favor. “This is but a
recognition that the purpose of federal habeas corpus is to ensure that state

convictions comply with the federal law in existence at the time the conviction
became final, and not to provide a mechanism for the continuing reexamination of
final judgments based upon later emerging legal doctrine.” Sawyer v. Smith, 497
U.S. 227, 234 (1990).
Though these two constraints—the rule of nonretroactivity set out in Teague
and the deference to state court decisions mandated by AEDPA—are similar in
some respects, they are nonetheless “quite separate” in their operation, and a state
prisoner seeking federal habeas relief must clear both hurdles to succeed. Greene
v. Fisher, 565 U.S. 34, 39 (2011). Accordingly—and the Supreme Court has made
this clear—“in addition to performing any analysis required by AEDPA, a federal
court considering a habeas petition must conduct a threshold Teague analysis when

8
 Case: 18-13390 Date Filed: 08/30/2019 Page: 9 of 28

the issue is properly raised by the state.” Horn v. Banks (Banks I), 536 U.S. 266,
272 (2002) (per curiam).

B.
Before conducting that analysis here, we pause to explain why we cannot
simply accept the Florida Supreme Court’s decision to apply Hurst retroactively to

Knight and review only its harmless-error analysis, as Knight urges us to do.
Because the Florida Supreme Court had already decided to give him the benefit of
Hurst, Knight says, the Teague retroactivity analysis no longer has any bearing in

his case. He is wrong. While states may fashion their own retroactivity doctrines
as a matter of state law, those doctrines cannot displace Teague on the federal
stage. Our ability to consider whether Florida applied Hurst correctly depends

entirely on whether we can apply Hurst ourselves. So far, neither the Supreme
Court nor this Circuit has answered that question by analyzing Hurst’s retroactivity
under Teague.2
Florida, on the other hand, has its own retroactivity standard—and is free to
give broader retroactive effect to new constitutional rules in state court proceedings

2
We have noted in passing that Hurst would not apply retroactively to a petitioner whose
convictions became final long before the Supreme Court decided Ring v. Arizona, 536 U.S. 584
(2002), on which Hurst relied, and even before Apprendi v. New Jersey, 530 U.S. 466 (2000),
which formed the basis for Ring. See Lambrix v. Sec’y, Fla. Dep’t of Corr., 851 F.3d 1158, 1165
n.2 (11th Cir. 2017). And we have concluded in the context of an improvidently granted
certificate of appealability that the Florida Supreme Court’s decision not to apply Hurst
retroactively to the same petitioner as a matter of state law was not contrary to or an
unreasonable application of existing Supreme Court precedents. See Lambrix v. Sec’y, Dep’t of
Corr., 872 F.3d 1170, 1182 (11th Cir. 2017) (per curiam). But the question of Hurst’s
retroactivity under Teague to a petitioner like Knight—whose convictions became final after
Ring but before Hurst—was not squarely presented in either case, so we did not conduct that
analysis.
9
 Case: 18-13390 Date Filed: 08/30/2019 Page: 10 of 28

than Teague allows in federal cases. See Danforth v. Minnesota, 552 U.S. 264,
282 (2008). That is because the Teague bar “was intended to limit the authority of

federal courts to overturn state convictions—not to limit a state court’s authority to
grant relief for violations of new rules of constitutional law when reviewing its
own State’s convictions.” Id. at 280–81. So when states choose to apply new rules

of constitutional procedure that are not retroactive under Teague in federal courts,
they “do not do so by misconstruing the federal Teague standard. Rather, they
have developed state law to govern retroactivity in state postconviction

proceedings.” Id. at 288–89 (emphasis in original).
In deciding to apply Hurst retroactively to certain state habeas cases, the
Florida Supreme Court did just that. See Mosley v. State, 209 So. 3d 1248, 1274–

83 (Fla. 2016) (per curiam). Florida’s retroactivity doctrine is unique to Florida—
it applies to a limited class of constitutional decisions that announce changes “of
fundamental significance,” which for procedural rules requires consideration of
three factors: “(a) the purpose to be served by the rule, (b) the extent of reliance on
the prior rule, and (c) the effect that retroactive application of the new rule would
have on the administration of justice.” Asay v. State, 210 So. 3d 1, 16–17 (Fla.
2016) (per curiam) (citation omitted). For new constitutional rules involving the
death penalty, Florida courts also consider on a case-by-case basis whether
fundamental fairness requires retroactive application of the rule. Mosley, 209 So.
3d at 1274 (citing James v. State, 615 So. 2d 668, 668 (Fla. 1993)).3 If that sounds

3
The Florida Supreme Court is now considering whether it should recede from the retroactivity
analysis employed in Asay, Mosley, and James. See Owen v. State, No. SC18-810, April 24,

10
 Case: 18-13390 Date Filed: 08/30/2019 Page: 11 of 28

broader than Teague, it is for good reason—the Florida Supreme Court has itself
acknowledged that this retroactivity standard is “more expansive” than the federal

rule. Johnson v. State, 904 So. 2d 400, 409 (Fla. 2005) (per curiam), abrogated on
unrelated grounds by Asay, 210 So. 3d at 15–16.
All that to say, Florida may make its own choice about the retroactivity of a

given case as a matter of state law. And for Hurst, it has done so. Using its own
standard, the Florida Supreme Court decided that Hurst would apply retroactively
in state collateral review proceedings for petitioners whose convictions had not yet

become final when the U.S. Supreme Court decided Ring v. Arizona; Ring held in
2002 that the Sixth Amendment requires a jury (rather than a judge) to find the
aggravating factors necessary to impose the death penalty. See Mosley, 209 So. 3d

at 1283 (citing Ring v. Arizona, 536 U.S. 584 (2002)). The Florida Supreme Court
reasoned that because Hurst struck down Florida’s capital sentencing scheme
based on Ring, prisoners whose cases were still pending on direct appeal when
Ring was decided “should not suffer due to the United States Supreme Court’s
fourteen-year delay in applying Ring to Florida.” Id. Following that rule, because
Knight’s conviction became final ten years after Ring was decided, the Florida
Supreme Court applied Hurst retroactively in his postconviction proceeding.
Knight, 225 So. 3d at 682.

2019 order (directing parties to brief the issue). The uncertain fate of Florida’s current
retroactivity doctrine offers another reason that we cannot simply rely on a state retroactivity
decision as a basis for federal habeas relief. If the state doctrine were to change during our
review, would we then be faced with the question of whether to apply the state’s new
retroactivity doctrine—retroactively?
11
 Case: 18-13390 Date Filed: 08/30/2019 Page: 12 of 28

But that state-law retroactivity determination has no significance in federal
court. Unlike state courts, lower federal courts are not free to create our own rules

of retroactivity—if the government raises the issue, a Teague analysis is
mandatory. Banks I, 536 U.S. at 271–72. As we have said, “States may exercise
their collateral review power without regard to the Teague doctrine. Their doing

so has no effect on later federal review.” Glock v. Singletary, 65 F.3d 878, 890–91
(11th Cir. 1995) (en banc). So we are bound to follow Teague’s retroactivity
principles whether or not the state court chose to apply the new rule in its own

collateral proceeding. See, e.g., Banks I, 536 U.S. at 271–72.
Here, then, we must conduct our own retroactivity analysis, using the
standards articulated in Teague. And to repeat: Teague retroactivity is a “threshold

question in every habeas case.” Caspari v. Bohlen, 510 U.S. 383, 389 (1994).
When “issues of both retroactivity and application of constitutional doctrine are
raised,” we must decide the retroactivity issue first. Bowen v. United States, 422
U.S. 916, 920 (1975). Where the State raises the issue, therefore, “federal habeas
corpus courts must apply Teague before considering the merits” of the petitioner’s
claims. Beard v. Banks (Banks II), 542 U.S. 406, 412 (2004) (emphasis in
original) (internal quotation marks and citation omitted).
What’s more, if a constitutional claim is Teague-barred, we do not reach its
merits. See, e.g., id. at 410 n.2. That is because the Supreme Court’s
“jurisprudence concerning the ‘retroactivity’ of ‘new rules’ of constitutional law is
primarily concerned, not with the question whether a constitutional violation
occurred, but with the availability or nonavailability of remedies.” Danforth, 552

12
 Case: 18-13390 Date Filed: 08/30/2019 Page: 13 of 28

U.S. at 290–91. If the holding relied on qualifies as a new rule and does not meet
Teague’s strict requirements for retroactivity, then the claim is not redressable

here—the “nonretroactivity principle prevents a federal court from granting habeas
corpus relief to a state prisoner based on a rule announced after his conviction and
sentence became final.” Caspari, 510 U.S. at 389 (emphasis in original). And as

we have said before, if “the court cannot relieve the harm of which a plaintiff
complains, the court should not take the case; in the absence of an effective remedy
its decision can amount to nothing more than an advisory opinion.” Wymbs v.

Republican State Exec. Comm., 719 F.2d 1072, 1085 (11th Cir. 1983).
Our authority to overturn state convictions is limited, and the retroactivity
principles articulated in Teague are tailored to those limitations. See Danforth, 552

U.S. at 277. The fact that state courts do not face the same constraints on collateral
review of their own criminal proceedings as we do does not relieve us of the
obligation to apply federal retroactivity standards. See id. at 280–81; Banks I, 536
U.S. at 271. In fact, our narrow authority as federal courts to disrupt final state-
court convictions reflects our recognition of the states’ own sovereignty. So
Florida may design and apply its retroactivity principles as generously as it wishes.
But notwithstanding Florida’s decision to apply Hurst—or any future decision—
retroactively as a matter of state law, as a federal court we are required to perform
the Teague analysis to determine whether prisoners can receive retroactive relief
under federal law.
In sum, if Hurst announced a new rule of constitutional law, but one that
does not fall into one of the exceptions to Teague’s bar on retroactivity, Knight

13
 Case: 18-13390 Date Filed: 08/30/2019 Page: 14 of 28

cannot obtain federal habeas relief for any Hurst error in his sentence—regardless
of what Florida may choose to do under state law. And if Knight cannot obtain

federal habeas relief for his Hurst claim in any event, we may not offer an advisory
opinion on whether the claim could have merit.
III.

A.
Turning to Teague, our analysis has three steps. First, we determine the date
when the petitioner’s conviction became final. Banks II, 542 U.S. at 411. This

happens when the United States Supreme Court “affirms a conviction on the merits
on direct review or denies a petition for a writ of certiorari, or when the time for
filing a certiorari petition expires.” Clay v. United States, 537 U.S. 522, 527

(2003). Here, the Court denied Knight’s petition for a writ of certiorari on May 14,
2012—more than three years before Hurst was decided. Knight, 566 U.S. at 998.
Second, if the rule that the petitioner wants to apply had not been announced
by that final-conviction date, we “assay the legal landscape” as it existed at the
time and determine whether existing precedent compelled the rule—that is,
whether the case announced a new rule or applied an old one. Banks II, 542 U.S.
at 413. If—and only if—the holding was “dictated by precedent existing at the
time the defendant’s conviction became final,” then the rule is not new and may be
applied retroactively on federal habeas review (indeed, it must be). Caspari, 510
U.S. at 390 (emphasis in original) (quoting Teague, 489 U.S. at 301). And that is
not a light test—a rule is not dictated by prior precedent “unless it would have been
‘apparent to all reasonable jurists.’” Chaidez v. United States, 568 U.S. 342, 347

14
 Case: 18-13390 Date Filed: 08/30/2019 Page: 15 of 28

(2013) (quoting Lambrix v. Singletary, 520 U.S. 518, 528 (1997)). The fact that a
decision is “within the logical compass of” or even “controlled by” prior precedent

is not conclusive in the Teague analysis. Butler v. McKellar, 494 U.S. 407, 415
(1990) (internal quotation marks and citation omitted). To “‘dictate’ a result, prior
precedent must be specific; it is not enough that it name the general principle from

which the assertedly new rule sprang.” Glock, 65 F.3d at 884.
Knight argues that Ring v. Arizona dictated the rule in Hurst. In Ring, the
Supreme Court held that Arizona’s capital sentencing scheme, which required

judges alone to hear penalty-phase evidence and make factual findings relevant to
the imposition of the death penalty, violated the defendant’s Sixth Amendment
right to a jury trial. 536 U.S. at 588–89. In doing so, the Court explicitly

overruled its precedent upholding Arizona’s death penalty sentencing scheme, “to
the extent that it allows a sentencing judge, sitting without a jury, to find an
aggravating circumstance necessary for imposition of the death penalty.” Id. at
609 (emphasis added) (overruling Walton v. Arizona, 497 U.S. 639, 647–49
(1990)).
Knight wants us to find that the new Hurst rule is actually the old Ring rule
for an obvious reason—if the rule is not new, and instead was binding on lower
courts at the time of Knight’s conviction, then he is entitled to the benefit of the
rule on federal habeas review. See Yates v. Aiken, 484 U.S. 211, 216–17 & n.3
(1988). But Ring did not dictate the Supreme Court’s later invalidation of
Florida’s death penalty sentencing scheme in Hurst. In fact, the Ring Court
specifically acknowledged that Florida’s capital sentencing procedure differed

15
 Case: 18-13390 Date Filed: 08/30/2019 Page: 16 of 28

from the Arizona scheme that it rejected. See Ring, 536 U.S. at 607–08 & n.6
(categorizing state capital sentencing schemes according to jury involvement in

sentencing). In Arizona, the judge alone made the factual findings necessary to
impose the death penalty and imposed that penalty entirely apart from the jury.
See id. at 588. Florida’s scheme, in contrast, incorporated an advisory jury that

considered penalty-phase evidence and recommended a sentence of life or death to
the court. Only after the jury’s recommendation did the judge impose a sentence.
See id. at 608 n.6; Fla. Stat. Ann. § 921.141 (2001).

Hurst’s conclusion that Florida’s “hybrid” death penalty sentencing system
violated the Sixth Amendment was not “apparent to all reasonable jurists” when
Knight’s convictions became final in 2012. We venture to count ourselves and our

colleagues on this Court as members of that distinguished group. And in Evans v.
Secretary, Florida Department of Corrections, we held that Ring did not invalidate
Florida’s capital sentencing scheme. 699 F.3d 1249, 1264–65 (11th Cir. 2012).
We found it significant that Florida’s statute—unlike the Arizona law at issue in
Ring—required the penalty phase jury to find that sufficient aggravating
circumstances existed before it could recommend a death sentence, and directed
the sentencing court to give “great weight” to the jury’s advisory verdict. Id. at
1261 (citation omitted). We also noted that the Supreme Court had taken obvious
pains in Ring to distinguish hybrid systems like Florida’s from the “judge-only”
sentencing scheme in Arizona and concluded that such distinctions would not have
been necessary if the Court had intended to strike down both systems. Id. at 1262.

16
 Case: 18-13390 Date Filed: 08/30/2019 Page: 17 of 28

And we were not the only ones. In fact, Justice Alito wrote along those
same lines in his dissenting opinion in Hurst: “Although the Court suggests that

today’s holding follows ineluctably from Ring, the Arizona sentencing scheme at
issue in that case was much different from the Florida procedure now before us.”
136 S. Ct. at 625 (Alito, J., dissenting). After describing the “critically important

role” of the advisory jury in Florida’s death penalty sentencing scheme, Justice
Alito concluded that the “decision in Ring did not decide whether this procedure
violates the Sixth Amendment, and I would not extend Ring to cover the Florida

system.” Id. at 626. Clearly, reasonable jurists could—and did—disagree that
Ring compelled the outcome in Hurst. The Alito dissent and this Court’s pre-
Hurst holding strongly indicate that Hurst announced a new constitutional rule

rather than applying an old one. See Banks II, 542 U.S. at 415.
That conclusion is only strengthened by the fact that Ring did not
specifically address the continued validity of the Supreme Court’s precedents
upholding Florida’s death-penalty sentencing system—Spaziano v. Florida, 468
U.S. 447 (1984), and Hildwin v. Florida, 490 U.S. 638 (1989) (per curiam). And
the Court has repeatedly instructed us to follow its precedents, even if later
decisions appear to undermine them, unless and until the Court itself sets them
aside. See, e.g., Bosse v. Oklahoma, 137 S. Ct. 1, 2 (2016) (per curiam) (“Our
decisions remain binding precedent until we see fit to reconsider them, regardless
of whether subsequent cases have raised doubts about their continuing vitality.”
(citation omitted)); Rodriguez de Quijas v. Shearson/Am. Express, Inc., 490 U.S.
477, 484 (1989). So Spaziano and Hildwin remained good law until the Court

17
 Case: 18-13390 Date Filed: 08/30/2019 Page: 18 of 28

explicitly overruled them in Hurst. See Saffle v. Parks, 494 U.S. 484, 488 (1990)
(explaining that the “explicit overruling of an earlier holding no doubt creates a

new rule”); see also Hurst, 136 S. Ct. at 623. It is hard to imagine that the
Supreme Court overruled those cases in Ring but forgot to say so until Hurst.
Because all these factors show that Hurst was not dictated by prior

precedent—and in fact explicitly overruled existing precedent upholding Florida’s
death penalty sentencing scheme—we can see that the rule in Hurst, which led to a
conclusion that the Florida scheme was unconstitutional, was new.

Having determined that Hurst announced a new rule of constitutional law,
we proceed to the final step in the Teague analysis—whether Hurst “falls within
either of the two exceptions to nonretroactivity.” Banks II, 542 U.S. at 411. Those

exceptions, again, include (1) holdings that create substantive (not procedural)
rules that place “certain kinds of primary, private individual conduct beyond the
power of the criminal law-making authority to proscribe,” and (2) holdings that
constitute “watershed rules of criminal procedure.” Teague, 489 U.S. at 311–13
(citation omitted).
The Hurst rule does not fit within either exception. To begin, substantive
rules include decisions that change “the range of conduct or the class of persons
that the law punishes.” Schriro, 542 U.S. at 353. Procedural rules, on the other
hand, “regulate only the manner of determining the defendant’s culpability.” Id.
(emphasis in original). In considering which category the Hurst rule falls into, we
have a head start because the Supreme Court has already held that Ring represented
a “prototypical procedural rule[].” Id. And that makes sense: Ring changed the

18
 Case: 18-13390 Date Filed: 08/30/2019 Page: 19 of 28

permissible procedure for sentencing in a capital case when it required “that a jury
rather than a judge find the essential facts” necessary to impose the death penalty.

See id. Because Hurst’s holding—that an advisory “jury’s mere recommendation
is not enough” to satisfy this procedural requirement—is an extension of the rule
from Ring, we have no trouble concluding that Hurst also announced a procedural

rule, and not a substantive one. Hurst, 136 S. Ct. at 619.
The second exception is for “watershed rules of criminal procedure.” Banks
II, 542 U.S. at 417. This exception is extremely limited in scope—it applies “only

to a small core of rules” so fundamental to our criminal process that it is “unlikely
that many such components of basic due process have yet to emerge.” Id.
(citations omitted). Indeed, the watershed exception remains somewhat theoretical

at this point; in the years following Teague, the Supreme Court has never found a
rule that fits. See id. And in “providing guidance as to what might fall within this
exception,” the Court has “repeatedly referred to the rule of Gideon v. Wainwright,
372 U. S. 335 (1963) (right to counsel), and only to this rule.” Id. Knight does not
contend that Hurst announced a new watershed rule that compares to Gideon, and
we do not see how it could have either. In short, Hurst meets neither exception,
and therefore is not retroactive.
B.
Because the Hurst rule is not retroactive, Knight cannot receive federal
habeas relief on his Hurst claim. That is as it must be—we are conscientious about
the fact that “Teague’s nonretroactivity principle acts as a limitation on the power
of federal courts to grant habeas corpus relief.” Id. at 412 (internal quotation

19
 Case: 18-13390 Date Filed: 08/30/2019 Page: 20 of 28

marks and citation omitted). And because we would have no lawful remedy to
offer even if we could identify an error, we must decline to consider whether any

Hurst error exists. That means, of course, that we also do not consider whether the
un-found and un-remediable error could be harmless.
Knight disputes this path. He argues that, even conceding the lack of an

available remedy for any Hurst error, whether that un-remediable error was
harmless is itself a separate question of federal law under Brecht v. Abrahamson,
507 U.S. 619, 637–38 (1993). He thus urges us to review the state court’s

harmless-error analysis, regardless of whether we reach the merits of his Hurst
claim. We will not. It strains the imagination—as well as our constitutional and
institutional respect for state courts—to suppose that we cannot remedy an error,

but that we can somehow remedy an erroneous state-court conclusion that the error
was harmless.
And Brecht does not stand for the proposition that Knight asserts in any
event. While Brecht established the harmless-error standard for collateral review
of constitutional trial errors, it did not create a stand-alone federal claim severable
from the question of whether remediable error existed in the first place.4 See
Williams v. Singletary, 114 F.3d 177, 180 (11th Cir. 1997) (per curiam), as
amended on denial of reh’g (July 29, 1997). Some underlying violation of federal
law that we can address is a necessary predicate to federal habeas relief—unless

4
As an aside, what Brecht really decided was that federal courts evaluating constitutional trial
error on collateral review would apply a more relaxed harmless-error standard—whether the
error “had substantial and injurious effect or influence in determining the jury’s verdict”—rather
than the harmless-beyond-a-reasonable-doubt standard used on direct review. Brecht, 507 U.S.
at 638 (quoting Kotteakos v. United States, 328 U.S. 750, 776 (1946)).
20
 Case: 18-13390 Date Filed: 08/30/2019 Page: 21 of 28

we agree that an error has occurred, it makes no difference whether the purported
error was harmless. “We have consistently applied the Brecht harmless error

standard only after determining that there was an error.” Id.; cf. Wilson v.
Corcoran, 562 U.S. 1, 6 (2010) (per curiam) (“It is not enough to note that a
habeas petitioner asserts the existence of a constitutional violation; unless the

federal court agrees with that assertion, it may not grant relief.” (emphasis in
original)). There is no free-floating federal constitutional right to infallible
application of harmless-error principles.

And again, where our retroactivity doctrine forecloses the possibility of
federal habeas relief for a constitutional error, we are constrained to stop looking.
See Bowen, 422 U.S. at 920. The Supreme Court “consistently has declined to

address unsettled questions regarding the scope of decisions establishing new
constitutional doctrine in cases in which it holds those decisions nonretroactive,”
and has instructed us to do the same. Id. at 920–21. This directive carries
constitutional weight; as an Article III court, we are “without power to decide
questions that cannot affect the rights of litigants in the case” at hand. North
Carolina v. Rice, 404 U.S. 244, 246 (1971) (per curiam). So even if we found
Hurst error in Knight’s sentencing, we would still be prohibited from issuing a writ
of habeas corpus on that ground because Hurst is not retroactively applicable to
Knight under Teague. After all, Teague’s nonretroactivity command is a limitation
on our power, not a polite suggestion. Banks II, 542 U.S. at 412. So our opinion—
whatever it might be—on Knight’s Hurst claim would be purely advisory. “And it
is quite clear that the oldest and most consistent thread in the federal law of

21
 Case: 18-13390 Date Filed: 08/30/2019 Page: 22 of 28

justiciability is that the federal courts will not give advisory opinions.” Flast v.
Cohen, 392 U.S. 83, 96 (1968) (internal quotation marks and citation omitted).

Where, as here, Teague bars relief before we reach the preliminary question
of whether constitutional error occurred at all, consideration of the secondary
question of whether any such error was harmless would be a prohibited and

pointless exercise for both the petitioner and this Court. We therefore cannot grant
Knight relief on his Hurst claim, whether or not it is cloaked in the garb of
harmless error.

IV.
We now turn to Knight’s other claim. The State presented extensive DNA
evidence against him during the guilt phase of his trial. Kevin Noppinger, a

serologist with the Broward County Sheriff’s Office crime lab, testified that Knight
had Odessia’s blood on his hand and her DNA on his shirt when he was arrested.
Fingernail scrapings from Odessia’s body showed that she, in turn, had Knight’s
DNA under her fingernails. Noppinger also tested samples from the bloody
clothes (boxer shorts, a shirt, and jean shorts) found under the bathroom sink. He
found Knight’s blood mixed with Hanessia’s blood on the boxer shorts, Odessia’s
and Hanessia’s blood elsewhere on the boxer shorts and on the jean shorts, and
Odessia’s blood on the shirt.
One of the State’s other experts was Kevin McElfresh of Bode Technology
Group, whose DNA analysts had conducted additional testing on different samples
from the same items of clothing. In particular, McElfresh’s group analyzed DNA
samples from an unstained area of the waistband of the boxer shorts in an attempt

22
 Case: 18-13390 Date Filed: 08/30/2019 Page: 23 of 28

to determine who owned them. Although Bode’s initial report stated that Knight’s
DNA was not on the waistband sample, McElfresh testified at trial that he had

conducted some additional analysis and determined that some of the DNA on the
waistband could have been Knight’s.
Knight’s guilt-phase counsel consulted with DNA expert Dr. Norah Rudin,

who was listed as a potential trial witness by the defense. Dr. Rudin informed
counsel that, although some of Noppinger’s sample labeling practices were sloppy,
she generally agreed with his conclusions about the sources of the DNA samples

he analyzed. She was much more critical of McElfresh’s analysis: she called his
methods “fundamentally incorrect and inherently biased” and his testimony
“incomplete and misleading.” In her opinion, the DNA test results for the

waistband samples were “inconclusive.” But Dr. Rudin did not stop there—she
also considered McElfresh’s testimony “relatively inconsequential when viewed in
the context of the biological evidence as a whole.” She told Knight’s counsel that
she did not believe that she could help his case, and that she would not call herself
as an expert if she were in his shoes. Counsel, it appears, agreed with her
perspective and did not call her at trial.
Testimony indicates that one additional fact persuaded counsel that calling
Dr. Rubin to quibble with McElfresh’s methods would not be good trial strategy.
At the time of Knight’s trial, Florida permitted a defendant who did not put up any
evidence at trial to give both initial and rebuttal closing arguments—an advantage
ordinarily offered to the prosecution. Dr. Rudin would have been Knight’s only
witness, so calling her to testify would have meant giving up the opportunity to

23
 Case: 18-13390 Date Filed: 08/30/2019 Page: 24 of 28

have the first word and the last at closing. Given that fact, she would have largely
corroborated Noppinger’s testimony while also costing Knight an advantage at

closing arguments.
Nonetheless, on state collateral review, Knight contended that his counsel
provided constitutionally ineffective assistance by failing to call Dr. Rudin as an

expert. He argued that Dr. Rudin’s criticisms of McElfresh’s methods and the
crime lab’s labeling practices would have cast doubt on all of the State’s DNA
evidence and significantly damaged the State’s case. The Florida Supreme Court

disagreed, finding that Knight had not met his burden on either prong of the two-
part test for ineffective assistance of counsel claims set out in Strickland. See
Knight, 225 So. 3d at 672–74. Knight, however, argues that the state court’s

decision “was contrary to, or involved an unreasonable application of, clearly
established Federal law, as determined by the Supreme Court of the United States.”
28 U.S.C. § 2254(d)(1).
To begin, the Florida Supreme Court correctly identified the governing
standard. Strickland is the relevant “clearly established” Supreme Court precedent
for purposes of an ineffective-assistance claim. See Premo v. Moore, 562 U.S.
115, 118 (2011). Under Strickland, to succeed on a claim of ineffective assistance
of counsel, a defendant must show both that his attorney’s performance was
deficient—that is, “that counsel’s representation fell below an objective standard of
reasonableness”—and that he was prejudiced by the inadequate performance.
Strickland, 466 U.S. at 687–88. In applying the first prong, courts “must indulge a
strong presumption that counsel’s conduct falls within the wide range of

24
 Case: 18-13390 Date Filed: 08/30/2019 Page: 25 of 28

reasonable professional assistance.” Id. at 689. And to show prejudice, the
“defendant must show that there is a reasonable probability that, but for counsel’s

unprofessional errors, the result of the proceeding would have been different. A
reasonable probability is a probability sufficient to undermine confidence in the
outcome.” Id. at 694.

The Strickland standard is “highly deferential,” as is the review of a state-
court decision under AEDPA; “when the two apply in tandem, review is ‘doubly’
so.” Harrington v. Richter, 562 U.S. 86, 105 (2011) (citations omitted). In

reviewing a state court’s application of Strickland, therefore, a federal habeas court
cannot conduct a de novo review and reverse simply because it strongly disagrees
with the state court’s conclusion. Id. at 102. Instead, for Knight to succeed on his

ineffective-assistance claim, we must conclude that the Florida Supreme Court’s
decision “was an objectively unreasonable application of the Strickland standard.”
Allen v. Sec’y, Fla. Dep’t of Corr., 611 F.3d 740, 751 (11th Cir. 2010).
The Florida Supreme Court determined that counsel’s decision not to call
Dr. Rudin at trial was reasonable trial strategy. Knight, 225 So. 3d at 674.
Whether the decision was actually a matter of strategy is a question of fact; thus,
the state court’s finding on that issue is presumed to be correct. Provenzano v.
Singletary, 148 F.3d 1327, 1330 (11th Cir. 1998). We must accept all factual
findings made by the state court unless the petitioner rebuts the presumption of
correctness “by clear and convincing evidence.” 28 U.S.C. § 2254(e)(1); see also
Wood v. Allen, 542 F.3d 1281, 1285 (11th Cir. 2008). Knight has not presented

25
 Case: 18-13390 Date Filed: 08/30/2019 Page: 26 of 28

any evidence to suggest that counsel’s decision was anything other than a matter of
strategy, and we accept the state court’s finding on that point.

Whether counsel’s strategic decision not to call Dr. Rudin was reasonable is
a question of law, which we review through the lens of AEDPA deference. See
Ferrell v. Hall, 640 F.3d 1199, 1223–24 (11th Cir. 2011). In assessing an

attorney’s performance under Strickland, “strategic choices made after thorough
investigation of law and facts relevant to plausible options are virtually
unchallengeable.” Strickland, 466 U.S. at 690. “Which witnesses, if any, to call,

and when to call them, is the epitome of a strategic decision, and it is one that we
will seldom, if ever, second guess.” Waters v. Thomas, 46 F.3d 1506, 1512 (11th
Cir. 1995) (en banc). We have no reason to doubt the Florida Supreme Court’s

conclusion that counsel’s decision not to call an equivocal expert, in part to
preserve an advantage at closing, was well within the wide range of reasonably
competent performance.
The Florida Supreme Court also determined that Knight had not met his
burden of showing prejudice under Strickland because there was no reasonable
probability that Dr. Rudin’s testimony would have made a difference in the
outcome of the trial, given the weight of the evidence against him. See Knight, 225
So. 3d at 674. That conclusion was also reasonable.
Dr. Rudin generally agreed with Noppinger’s conclusions regarding the
DNA evidence. With or without Dr. Rudin’s testimony, therefore, there was no
dispute that Knight had Odessia’s blood on his hand and her DNA on his shirt
when he was arrested, or that Knight’s DNA was found under Odessia’s

26
 Case: 18-13390 Date Filed: 08/30/2019 Page: 27 of 28

fingernails. Knight’s blood—as well as Odessia’s blood and Hanessia’s—was on
the clothes discarded at the crime scene. And even if the jury had heard Dr.

Rudin’s criticism of McElfresh’s testimony that Knight could not be ruled out as a
DNA contributor for the waistband of the boxer shorts, other evidence showed that
the clothes were his: the boxer shorts were the same brand and size as the ones that

Knight was wearing when he was arrested, and Knight’s cousin testified that the
bloody shirt was one that Knight wore often.
And the evidence did not stop there. Knight was known to be home with

Hanessia less than an hour before the murders, and the jury heard evidence of
ongoing tension with Odessia. Police found him near the scene of the murders
shortly after they arrived, and his answers to their questions were inconsistent with

his appearance. All of that is in addition to the testimony of an inmate housed with
Knight at the Broward County jail, who testified that Knight confessed the crime to
him, providing a detailed description of events and a floorplan of the apartment. In
short, even if the jury had entirely discounted McElfresh’s testimony as a result of
Dr. Rudin’s criticism of his testing methods, the remaining evidence against
Knight was so strong that the chance of a not-guilty verdict still would have been
remote, to say the least.
The Florida Supreme Court’s conclusion that Knight failed to make the
required showings of deficient performance and prejudice was not an unreasonable
application of Strickland.

27
 Case: 18-13390 Date Filed: 08/30/2019 Page: 28 of 28

***
In sum, we are prohibited from considering Knight’s Hurst claim by

Teague’s rule of nonretroactivity, which eliminates any possibility of relief
regardless of whether there was an error and regardless of whether any error was
harmless. And our review of Knight’s ineffective-assistance claim shows that the

state court’s ruling on that issue was not an unreasonable application of governing
Supreme Court precedent. We therefore affirm the district court’s denial of
Knight’s federal habeas petition.

AFFIRMED.

28