John Burr v. Denise Jackson

                                     PUBLISHED

                       UNITED STATES COURT OF APPEALS
                           FOR THE FOURTH CIRCUIT


                                        No. 20-5


JOHN EDWARD BURR,

                    Petitioner - Appellant,

             v.

DENISE JACKSON, Warden, Central Prison, Raleigh, North Carolina,

                    Respondent - Appellee.



Appeal from the United States District Court for the Middle District of North Carolina, at
Greensboro. William L. Osteen, Jr., District Judge. (1:01-cv-00393-WO-JEP)


Argued: September 24, 2021                                 Decided: November 30, 2021


Before WILKINSON, WYNN, and HARRIS, Circuit Judges.


Affirmed by published opinion. Judge Wynn wrote the opinion, in which Judge Wilkinson
and Judge Harris joined.


ARGUED: James P. Cooney, III, WOMBLE BOND DICKINSON (US) LLP, Charlotte,
North Carolina, for Appellant. Kimberly Nicole Callahan, NORTH CAROLINA
DEPARTMENT OF JUSTICE, Raleigh, North Carolina, for Appellee. ON BRIEF:
Ernest Lee Conner, Jr., Greenville, North Carolina, for Appellant. Joshua H. Stein,
Attorney General, L. Michael Dodd, Special Deputy Attorney General, NORTH
CAROLINA DEPARTMENT OF JUSTICE, Raleigh, North Carolina, for Appellee.
WYNN, Circuit Judge:

       Petitioner John Edward Burr was convicted after a 1993 trial for the 1991 murder

of infant Tarissa Sue O’Daniel, known to her family as “Susie.” He was sentenced to death.

In the decades since, Burr has pursued habeas remedies before the state and federal courts,

including this Court. But this appeal concerns a narrow question: whether the district court

erred in declining to grant habeas relief on the basis of claims under Brady v. Maryland

and Napue v. Illinois related to transcripts of interviews with two witnesses, Susie’s mother

and brother.

       Our standard of review is highly deferential to the conclusions of the state post-

conviction relief court. Under that standard of review, we agree with the district court that

Burr is not entitled to habeas relief. Accordingly, we affirm.

                                              I.

       Several courts, including this one, have previously laid out the relevant facts and

procedural history of this case in some detail. 1 We do not repeat that full history here, but

instead report only those facts relevant to the issues before us.

                                             A.

       Susie was born to Lisa Bridges and her then-husband on April 1, 1991. Shortly

thereafter, Bridges began having an affair with Burr. In late June, Bridges and her children




       1
        E.g., Burr v. Jackson, No. 1:01CV393, 

, at *1–6 (M.D.N.C.
Mar. 26, 2020); Burr v. Lassiter, 

, 329–39 (4th Cir. 2013) (per curiam)
(unpublished but orally argued); State v. Burr, 

, 606–11 (N.C. 1995).

                                              2
moved with Burr into a trailer next door to another trailer owned by Bridges’s stepbrother.

Burr quickly became physically abusive toward Bridges.

       Around 6:00 P.M. on August 24, 1991, Bridges’s eight-year-old son, Scott, tripped

over a cord while carrying Susie and fell on a gravel-covered driveway. Bridges and Burr

examined Susie after the fall and found her to be uninjured. But early reports about the

mechanics of the fall were inconsistent. Bridges and Scott both reported in the weeks

afterward that Scott fell on Susie, dropped her, or both. However, at other times during

those same early weeks, Bridges and Scott each reported that Scott did not let go of Susie

and instead cradled her gently as he fell. One of Scott’s early reports was made to a social

worker on August 27, who “assured him that nothing he had done hurt [Susie].” J.A. 1226. 2

       Late on the night of August 24 or in the early hours of the next morning, Bridges

went next door to wash dishes at her stepbrother’s home. She left Susie in her crib in

Bridges’s bedroom. Scott and his younger brother Tony were asleep in another bedroom.

Burr also remained at Bridges’s trailer.

       When Bridges returned forty-five minutes later, she found Susie in her swing in the

living room. Burr claimed he had moved Susie from her baby bed to the swing after she

woke up. It rapidly became clear that Susie was badly injured—she was covered in bruises;

her eyes were unblinking and rolling; and she was unresponsive. Bridges and Burr brought

Susie to the county hospital, where they arrived just before 3:00 A.M. on August 25.




       2
           Citations to the “J.A.” refer to the Joint Appendix filed by the parties in this appeal.

                                                 3
       The emergency room physician found Susie to be unconscious, with a weak pulse

and wandering eyes; “intermittently seizing or having seizures”; and presenting with a

bulging fontanel, “indicat[ing] some swelling inside the head.” J.A. 2812. In addition to

the bruising all over her body, X-rays revealed that both of her arms and both of her thigh

bones were broken. She also appeared to be suffering from a “closed head injury,” though

X-rays did not show a skull fracture. J.A. 2819. The physician immediately formed a “high

suspicion of abuse,” which he asked Bridges about. J.A. 2818. He also contacted the

sheriff’s department and social services.

       Due to the severity of her injuries, Susie was transferred to North Carolina Memorial

Hospital at the University of North Carolina at Chapel Hill. There, she was examined by a

trauma team that included the chief of pediatric surgery. The chief of pediatric radiology

reviewed Susie’s X-rays and CT scan and concluded that Susie had a “depressed skull

fracture” that was mere hours old. J.A. 2713. By contrast, the fractures in both of Susie’s

thigh bones showed evidence of early healing. The radiologist estimated those fractures

were eight to nine days old, with a range of three days on either side of the estimate. Susie’s

arm fractures did not show signs of healing, and the radiologist testified that they could

have happened at the same time as the head injury or up to five days previously.

       A pediatric neurologist also reviewed Susie’s CT scan and agreed that she had a

“depressed skull fracture.” J.A. 2840. He noted that there was no external wound on the

scalp. And he found that Susie exhibited symptoms, such as bilateral retinal hemorrhages,

that were indicative of “shaken baby syndrome,” “a specific kind of injury where the baby

has a whiplash kind of injury from being shaken back and forth.” 

                                              4
       Susie died from her injuries on August 27, 1991. The medical examiner concluded

that Susie’s cause of death was a closed head injury, and that the manner of death was

homicide. Burr was arrested the next day.

                                              B.

       Burr was tried before a jury in a guilt phase that took place over the course of twelve

trial days between March 29 and April 16, 1993. The trial evidence was extensive and

included testimony from Burr, Bridges, Scott, other relatives, investigating officers,

examining doctors, a social worker, and the pathologist who performed Susie’s autopsy.

       Scott testified at trial that he cradled Susie in his arms as he fell with her on August

24, such that she never hit the ground. Bridges and another witness who saw the fall, an

eleven-year-old relative named Jonas, confirmed Scott’s rendition. And Burr himself

testified that Scott held onto Susie when they fell. That said, the trial evidence also included

Burr’s testimony that he saw Scott “laying on top of Susie.” J.A. 3063. And it included

testimony that, in the early hours of August 25, Bridges told a deputy sheriff that Scott “fell

on” Susie, and that she similarly told a social worker on August 25 that Scott had “dropped”

Susie and “fell on her.” J.A. 3617, 3779. This contradicted her trial testimony that Scott

“didn’t drop his sister,” nor did he “fall on [her].” J.A. 2013. In other words, the jury had

before it competing testimony regarding the fall. In any event, Bridges, Burr, and several

other witnesses testified that they examined Susie after the fall and that she was not injured

at that time.

       Five doctors testified at trial: the emergency room physician at the county hospital

who saw Susie around 3:00 A.M. on August 25; the chief of pediatric surgery at North


                                               5
Carolina Memorial Hospital, who was part of the trauma team that evaluated Susie when

she arrived there around 6:00 A.M. on August 25; the chief of pediatric radiology at the

second hospital, who reviewed Susie’s X-rays on August 25; the pediatric neurologist at

the second hospital who examined Susie in the pediatric intensive care unit on August 25

and reviewed her CT scan; and the pathologist who performed Susie’s autopsy on August

28. Each doctor was qualified as an expert witness in their field of medical practice.

       The five testifying doctors unanimously agreed that the fall with Scott around 6:00

P.M. on August 24 could not have caused Susie’s lethal injuries. They confirmed that Susie

had no cuts, scratches, or other abrasions on her, such as one might expect if she fell with

any force on gravel. And they made clear that her head injury—whether or not her skull

was actually fractured—would have required a concentrated blow from a blunt object that

would almost certainly not have resulted from a fall, even if Susie’s head had struck the

gravel surface. For example, the pediatric surgeon testified that he had “seen situations

where somebody lands on top of a child . . . and they can end up with bruising of the liver

or even [a] ruptured liver,” but Susie’s injuries “don’t occur with that type of fall.” J.A.

2915. The pediatric radiologist testified that the skull fracture was “a very unusual fracture

in a very unusual place” that would “take a relatively confined direct blow to that area”

with “a great deal of force” to produce, because where the fracture occurred was “in a

portion of the skull” that is “somewhat protected because it’s a little depressed in,” so “the

rest of the skull would hit first.” J.A. 2726–27, 2738. The pediatric neurologist testified

that Susie had a “depressed skull fracture” caused by significant force from a blunt object,

akin to Susie having been “thrown against something” in a car accident, not from “a simple


                                              6
bump or fall.” J.A. 2840, 2847–48. He further testified that a fall from a height of roughly

three feet would not create “this kind of depressed skull fracture” where the “whole thing

[is] caved in.” J.A. 2849. The creation of such an injury, he explained, instead required

striking by a smaller, blunt object. Accordingly, he concluded that her injuries were not

accidental.

       The pediatric neurologist also testified that retinal hemorrhages, such as those seen

in the back of both of Susie’s eyes, would not be caused by a fall, but rather “would require

really very violent shaking.” J.A. 2876. He further testified that, while Susie’s condition

might have worsened in the hours following her injury, she would have been “significantly

ill, and obviously in trouble from the very beginning.” J.A. 2851. That is, “whatever the

injury was[,] from that point on the child should have been obviously not right,” even to a

layperson, with loss of consciousness occurring “within minutes to an hour or so.” J.A.

2851, 2854. Other doctors agreed. Finally, the pathologist testified that Susie was covered

with bruises across her body that were consistent with strikes from a blunt object, and that

she had bruising on her neck that was “consistent with marks that could be caused by a

handprint.” J.A. 2964.

       In the face of this evidence, defense counsel’s trial strategy was not to suggest that

Scott’s fall with Susie had been the cause of her injuries. In fact, counsel explicitly and

vigorously disclaimed that view in both their opening and closing statements, and elicited

agreement from Burr that the fall was “highly unlikely” to be the cause of the injuries Susie

received. J.A. 3111. Instead, counsel sought to suggest that someone else—most notably

Bridges—could have caused Susie’s injuries. This included arguing that Bridges’s


                                             7
testimony that she did not know Susie had fractured limbs lacked credibility; noting that

Bridges had much more access to Susie than Burr did in the weeks leading up to her death;

and introducing testimony from a witness who claimed to have once seen Bridges slap

Susie so hard that she fell off a couch.

       For its part, the State relied on the timeline of events; testimony from Scott, who at

the time of trial was ten years old; testimony to undermine that of Burr’s witness regarding

the couch-slapping incident; and character evidence showing Burr’s physically abusive

side, including testimony that he could be rough with his own toddler son.

       In his testimony, Scott told the jury that, after his mother left to go wash the dishes

on the night of August 24, he was awoken by “hammer noises.” J.A. 2769. He then heard

Susie crying and Burr “mumbling” before Susie’s crying ceased. 

 Scott testified that he

“[j]ust went back to bed” after this incident because he was “scared” to go check on Susie.

J.A. 2770. Scott also testified that he had seen Burr surreptitiously “shaking” Susie on

multiple occasions.

       The jury convicted Burr of first-degree murder, felonious child abuse, and assault

on a female (for abuse of Bridges). The court sentenced him to death at the jury’s

recommendation. The Supreme Court of North Carolina affirmed on direct appeal, and the

U.S. Supreme Court denied certiorari. See State v. Burr, 

 (N.C. 1995);

Burr v. North Carolina, 

 (1996).




                                              8
                                              C.

       Burr filed a motion for appropriate relief (“MAR”) in state court in 1996. He

included with his MAR several affidavits from non-treating doctors who claimed that

Susie’s injuries could have resulted from the fall with Scott. 3

       The MAR was denied the following year. But in July 1998, the Supreme Court of

North Carolina remanded the case for reconsideration in light of a new statutory

requirement that the State produce “the complete files of all law enforcement and

prosecutorial agencies involved” in investigations leading to death sentences. N.C. Gen.

Stat. § 15A-1415(f) (1996); see State v. Burr, 

 (N.C. 1998) (citing State

v. McHone, 

 (N.C. 1998); State v. Bates, 

 (N.C. 1998)).

       Under the remand order, the State produced tape recordings of interviews with Scott

and Bridges that had been conducted in February 1993, shortly before trial, but were never

given to defense counsel. Burr filed an amended MAR, alleging that the interviews

represented exculpatory evidence that should have been turned over pursuant to Brady v.

Maryland, 

 (1963), and that they showed that the State relied on testimony it

knew would leave the jury with a materially false impression, in violation of Napue v.

Illinois, 

 (1959). The state court rejected the amended MAR in 2000. The

Supreme Court of North Carolina denied certiorari.

       Burr then turned to the federal courts, filing a habeas petition in the Middle District

of North Carolina in 2001. The parties conducted extensive discovery, including


       3
         The medical license of one of the doctors who provided Burr with a supporting
affidavit has since been revoked. See Burr, 

 n.4.

                                              9
introducing new medical evidence. In 2009, the magistrate judge recommended that the

district court grant habeas relief because competent counsel would have secured an

independent medical expert and “presented evidence that the actual mechanism of [Susie]’s

death was an accidental fall.” Burr v. Branker, No. 1:01CV393, 

, at *7

(M.D.N.C. May 6, 2009).

       The district court did not file its order addressing the magistrate judge’s

recommendation until 2012, at which point the Supreme Court had issued its decision in

Cullen v. Pinholster, 

 (2011). Pinholster held that federal review pursuant to

(d)(1) “is limited to the record that was before the state court that

adjudicated the claim on the merits.” Pinholster, 

. The district court obliged,

confining its review to the record that was before the MAR court and disregarding the

parties’ discovery from the federal proceedings. Burr v. Branker, No. 1:01CV393, 

, at *1 (M.D.N.C. May 30, 2012).

       Nevertheless, the district court agreed with the magistrate judge and granted habeas

relief on the basis of ineffective assistance of counsel. Id. at *9. But, applying our highly

deferential standard of review under 

, this Court reversed. Burr v. Lassiter,

 (4th Cir. 2013) (per curiam) (unpublished but orally argued). The

case returned to the district court for evaluation of Burr’s remaining claims.

       In 2015, the State uncovered and disclosed an additional recording of a conversation

between Bridges and investigators in December 1992, a few months before trial. Only

some, but not all, of the 1992 recording had previously been disclosed to Burr. Burr moved

to amend the record under Rule 7 of the Rules Governing Section 2254 Cases to include


                                             10
the new transcript of the 1992 conversation. 4 The State did not object, and the district court

agreed.

       The district court held argument on the grounds remaining in Burr’s petition, after

which it denied the petition and declined to issue a certificate of appealability. Burr v.

Jackson, No. 1:01CV393, 

, at *29 (M.D.N.C. Mar. 26, 2020). Burr

appealed only as to his Brady and Napue claims, and we granted a certificate of

appealability.

                                              II.

       We review the district court’s decision de novo. Valentino v. Clarke, 

 (4th Cir. 2020). But our review of the state MAR court’s decision is highly deferential.

       The Antiterrorism and Effective Death Penalty Act of 1996 (“AEDPA”) empowers

federal courts to “entertain” applications for writs of habeas corpus filed by convicted state

prisoners “on the ground that [they are] in custody in violation of the Constitution or laws

or treaties of the United States.” 

(a). “But the manner in which the federal

courts may entertain such an application depends considerably on how the state court treats

a petitioner’s claims” and, in particular, whether its decision qualifies as an “adjudication

on the merits.” Valentino, 

; see 

(d). If it does, then our

review is severely circumscribed: we may only disturb the state court’s ruling if it (1)

“resulted in a decision that was contrary to, or involved an unreasonable application of,


       4
        Rule 7 provides that “the judge may direct the parties to expand the record by
submitting additional materials relating to the petition,” such as “letters predating the filing
of the petition, documents, exhibits, and answers under oath to written interrogatories
propounded by the judge,” as well as affidavits. 

 Rule 7(a), (b).

                                              11
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.” 

(d).

       A state court decision involves an “unreasonable application” of clearly established

federal law “if the state court identifies the correct governing legal rule” but “unreasonably

applies it to the facts.” Williams v. Taylor, 

 (2000). For such an

“unreasonable application” to exist, the state court’s decision must have been “so lacking

in justification that there was an error well understood and comprehended in existing law

beyond any possibility for fairminded disagreement.” White v. Woodall, 

(2014) (emphasis added) (quoting Harrington v. Richter, 

 (2011)).

       Similarly, a state court decision is “based on an unreasonable determination of the

facts” only where the “factual determination [is] ‘sufficiently against the weight of the

evidence that it is objectively unreasonable,’” which means “‘it must be more than merely

incorrect or erroneous.’” Williams v. Stirling, 

 (4th Cir. 2019), as

amended (Feb. 5, 2019) (quoting Winston v. Kelly (Winston I), 

 (4th Cir.

2010)). Further, the state court’s determination of factual issues is “presumed to be correct”

and may only be overturned by “clear and convincing evidence.” 

(e)(1).

       Accordingly, the § 2254(d) standard results in “a formidable barrier to federal

habeas relief.” Burt v. Titlow, 

 (2013). However, one aspect of the case at

bar gives us pause. Burr contends that the state MAR court’s 1997 and 2000 orders

“adopted verbatim the findings proposed by the State in [its] proposed Order[s].” Opening

Br. at 21 (discussing 2000 order); see also 

 (same regarding 1997 order). And in


                                             12
some cases, the verbatim adoption of a proposed order can heighten the standard of review,

loosening the amount of deference the reviewing court gives to the order under review. 5

See Jefferson v. Upton, 

 (2010) (per curiam). So the question before us

is whether we should afford § 2254 deference to state court orders in a capital habeas

proceeding where the orders largely track proposed orders filed by the State. We need not

determine whether such deference is always applicable in such cases, however, because we

conclude that, here, § 2254 deference applies.

       The Supreme Court stated in Anderson v. City of Bessemer City, an employment-

discrimination case, that “even when the trial judge adopts proposed findings verbatim, the

findings are those of the court and may be reversed only if clearly erroneous.” Anderson v.

City of Bessemer City, 

 (1985). Anderson reversed a decision from this

Court in which we applied a heightened level of scrutiny to the record where the district

court “directed [the] plaintiff’s counsel to submit proposed findings of fact, conclusions of

law, and an appropriate judgment”; allowed the defendant to respond; and adopted “[t]he

substance of [the] plaintiff’s submission . . . as the final opinion in the case.” Anderson v.

City of Bessemer City, 

 (4th Cir. 1983), rev’d, 

; see 

 at


       5
          To be sure, Burr does not argue that the verbatim adoption of the State’s proposed
orders changes our standard of review. And normally, “[a] party waives an argument by
failing to present it in its opening brief or by failing to develop its argument—even if its
brief takes a passing shot at the issue.” Grayson O Co. v. Agadir Int’l LLC, 

 (4th Cir. 2017) (internal quotation marks and alterations omitted). Nevertheless, “[o]ur
case law is clear that ‘parties cannot waive the proper standard of review by failing to argue
it,’” including in habeas cases. Richardson v. Kornegay, 

, 701 n.9 (4th Cir.
2021) (quoting United States v. Venable, 

 (4th Cir. 2019)). Rather, we
“must independently assure ourselves” of the appropriate standard of review. Sierra Club
v. U.S. Dep’t of the Interior, 

 (4th Cir. 2018).

                                             13
156. The Supreme Court held that the district court did “not appear to have uncritically

accepted findings prepared without judicial guidance by the prevailing party,” as it had

issued a “preliminary memorandum” before soliciting the proposed opinion and “the

findings it ultimately issued . . . var[ied] considerably in organization and content from

those submitted by [the plaintiff]’s counsel.” Anderson, 470 U.S. at 572–73.

       After Anderson, “we have taken a more lenient approach to district court opinions

that closely mirror a party’s submissions.” Alig v. Quicken Loans Inc., 

, 790

n.8 (4th Cir. 2021). However, we and other courts have continued to look to the facts of

each case to determine whether Anderson applies. 6 For example, applying pre-AEDPA

federal habeas law, the Supreme Court suggested in Jefferson v. Upton that verbatim

adoption of proposed findings in habeas cases might be problematic in some circumstances.

Jefferson, 

 (vacating and remanding “for the lower courts to

determine . . . whether the state court’s factual findings warrant a presumption of

correctness”).

       But Jefferson involved a particularly extreme example in which the petitioner

alleged that the state court asked the State in an ex parte conversation “to draft the opinion


       6
         See Alig, 

 n.8 (applying Anderson where “[t]he district court
engaged extensively with the issues over several years” and where its opinion “included
substantial sections the court wrote itself—as well as language adopted from [the
appellants’] briefs”); Aiken Cnty. v. BSP Div. of Envirotech Corp., 

, 676–77
(4th Cir. 1989) (applying Anderson despite the court’s near-verbatim adoption of an ex
parte proposed order where the opposing party had the opportunity to air its views fully
and the court appeared to have exercised independent judgment); Bright v. Westmoreland
Cnty., 

, 732 (3d Cir. 2004) (reversing and remanding a § 1983 case for further
consideration where the issue was not merely “findings of fact” but instead “a District Court
opinion that is essentially a verbatim copy of the appellees’ proposed opinion”).

                                             14
of the court.” Id. at 287. And some of our sister circuits have applied Anderson—and

therefore have not altered their deferential standards of review—in less extreme post-

AEDPA habeas cases, including after Jefferson. See Barksdale v. Att’y Gen. Ala., No. 20-

10993-P, 

, at *18 (11th Cir. June 29, 2020) (unpublished) (“We have

held that a state court’s verbatim adoption of the prosecution’s proposed order is entitled

to AEDPA deference as long as (1) both parties ‘had the opportunity to present the state

habeas court with their version of the facts’ and (2) the adopted findings of fact are not

‘clearly erroneous.’” (quoting Rhode v. Hall, 

 (11th Cir. 2009) (per

curiam))); Green v. Thaler, 

 (5th Cir. 2012) (applying Anderson in a

capital habeas case even though “the state court requested [the proposed order] ex parte,

and signed [it] verbatim,” where “the state court had already rendered a judgment from the

bench” and the petitioner “apparently had an opportunity to object to the findings in a

motion to strike he filed in the state court”); see also Nichols v. Scott, 

(5th Cir. 1995) (applying Anderson in a capital habeas case where the state court “adopted

verbatim the [S]tate’s proposed findings of fact and conclusions of law”).

       We need not lay down any blanket perimeters for how Anderson applies in capital

habeas cases generally, and we decline to do so where the parties have not briefed this

issue. Under the circumstances of this case, we conclude that Anderson applies, and

therefore the MAR court’s opinions are entitled to full § 2254 deference.

       Burr filed his initial MAR in 1996, after which the State filed a combined response

to and motion for summary denial of the MAR in March 1997. The State attached a

proposed 114-page order to that response. The record does not indicate that Burr filed a


                                             15
response to the motion for summary denial, or a reply related to the MAR. 7 Instead, six

months later, Burr filed an amended MAR. The state court denied the MAR in October

1997 in a 116-page order. The state court’s filed order copied the State’s proposal nearly

verbatim, with the exception of a handful of minor modifications and the addition of a

three-page section at the end to address the amended MAR.

       Following the Supreme Court of North Carolina’s 1998 remand of the case, Burr

filed his second amended MAR in February 1999. The State again filed a combined

response to and motion for summary dismissal of the second amended MAR in May 1999,

to which it attached a 43-page proposed order and memorandum opinion. Burr apparently

did not reply, instead filing a third amended MAR in October 1999, to which the State filed

a response in November. The state MAR court ultimately denied the MAR in June 2000.

At 68 pages long, its filed order was largely based on the proposed order but included

substantial additions by the court as well.


       7
         State law may have required Burr to seek permission from the court to file a reply.
See State v. Riley, 

 (N.C. Ct. App. 2000); N.C. Gen. Stat. § 15A-
1420(b1) (specifically noting, in both its 1997 and current versions, that the judge could
direct the State to file a response to the defendant’s MAR, but not mentioning an
opportunity for the defendant to reply); cf. State v. Vinh Nguyen, 

 (N.C. Ct.
App. 2018) (unpublished table decision) (noting that the State moved to strike the
defendant’s reply as improper, though dismissing the motion as moot). But see State v.
Howard, 

 (N.C. Ct. App. 2016) (noting that the defendant filed a reply
without noting that he first sought permission to do so); State v. Chekanow, No. 14 CRS
50306 & 50307, 2019 N.C. Super. LEXIS 478, at *1 (May 7, 2019) (same); State v. Lynch,
No. 08CRS58929, 58934, 2014 N.C. Super. LEXIS 237, at *1 (June 11, 2014) (same); cf.
State v. Lane, No. 14 CRS 50314-15, 2019 N.C. Super. LEXIS 450, at *7 (Jan. 9, 2019)
(noting that the defendant filed a response in opposition to the State’s motion to dismiss
his MAR without noting that he first sought permission to do so). Regardless, even if Burr
was required to seek such permission, the record does not show that he did so. Nor did he
attach his own proposed order to his amended MAR.

                                              16
       The MAR court’s 2000 order is primarily at issue in this appeal, as it contains the

Brady and Napue analyses. But the 1997 order is also relevant because of its discussion of

the cause of Susie’s death, and the 2000 order “reconsider[s]” and then incorporates the

1997 order. J.A. 1786. Accordingly, the question is whether Anderson applies to the 1997

and 2000 orders.

       We conclude that it does. We have carefully compared both filed orders with the

proposed orders and are convinced that the filed orders represent the state court’s own

work. Certainly, the 1997 filed order includes only minor changes from the proposed order.

But the 2000 order, which “reconsidered” the 1997 one, shows more input from the court.

 The 2000 order still adopted nearly all of the State’s proposed order, but Burr is

incorrect to state unequivocally that the adoption was “verbatim.” Opening Br. at 21. The

court made both major and minor changes throughout, including adding substantial

sections of text. For example, in discussing ineffective assistance of counsel, the court’s

order expands a one-sentence paragraph into a section spanning several pages. Compare

J.A. 1726, with J.A. 1792–98. As another example, and of relevance to the Brady and

Napue claims, the MAR court noted that “[Burr]’s transcriptions of the prosecutors’

discussion with [Bridges] demonstrate that the prosecutors were most concerned about

assuring that she testified truthfully.” J.A. 1830. That statement is absent from the

equivalent point in the State’s proposed order.

       To summarize, in this case, Burr had an opportunity to contest the proposed orders.

They were not filed ex parte. Nor did the state court simply rubber-stamp the proposed

orders—a comparison of the proposed and final 2000 orders reveals that the court carefully


                                            17
considered the issues and made modifications where appropriate, and the 2000 order makes

clear that the court had also reconsidered the closer-to-verbatim 1997 order. Those factors

combined convince us that the MAR court’s orders represent its own work and are thus

entitled to our deference.

       The preparation of proposed orders by parties in capital habeas cases appears to

persist as a practice in North Carolina. E.g., State v. Allen, 

, 280 (N.C. 2021)

(“[T]he MAR court sent the parties a Memorandum of Ruling asking the parties to draft

proposed orders disposing of [the petitioner’s] MAR . . . claims.”); cf. N.C. State Bar v.

Sutton, 

 (N.C. Ct. App. 2016) (noting, in an attorney discipline case,

that “[i]t is the accepted practice in North Carolina for the prevailing party to draft and

submit a proposed order that the decision-making body may then issue as its own—with

or without amendments”). To be clear, though we are sympathetic about the substantial

caseloads facing state trial judges, there are serious problems with this practice, as we and

other courts have noted previously. E.g., Aiken Cnty. v. BSP Div. of Envirotech Corp., 

 (4th Cir. 1989) (labeling the “near-verbatim adoption” of proposed findings

of fact and conclusions of law “less than ideal”); Anderson, 

 (noting that

the Court had “criticized” the “verbatim adoption of findings of fact prepared by prevailing

parties”); Jefferson, 560 U.S. at 293–94 (same). 8 Those concerns are particularly


       8
         See also, e.g., In re Equifax Inc. Customer Data Sec. Breach Litig., 

 (11th Cir. 2021) (noting the Eleventh Circuit’s “sharp[] critique[s]” of this practice);
Flying J Inc. v. Comdata Network, Inc., 

, 829–30 (10th Cir. 2005) (stating
that the near-verbatim adoption of proposed findings of fact and conclusions of law is
“[r]egrettabl[e]” and that it “provides little aid on appellate review”); Basso v. Stephens,


                                             18
pronounced when the state court adopts the State’s proposed order in a capital case, where

the need for an adversarial process and a neutral arbiter is at its zenith.

       But, in light of evidence that Burr was on notice of the proposed order and that the

state court judge exercised judicial discretion in adopting the State’s proposed order, we

conclude that it is appropriate to apply Anderson under these circumstances even as we

continue to “strongly criticize” the practice of verbatim (or close-to-verbatim) adoption of

proposed opinions. Hamm v. Comm’r, Ala. Dep’t of Corr., 

, 756 n.3 (11th

Cir. 2015) (per curiam). Accordingly, we apply § 2254 deference to Burr’s claims in this

appeal.

                                              III.

       Turning to the merits, we begin with the unreasonable determination of fact Burr

alleges pursuant to § 2254(d)(2). We then evaluate the MAR court’s discussion of his

Brady and Napue claims under § 2254(d)(1). We close by analyzing whether we can

consider the suppressed transcript that was not turned over until 2015, and if so, whether it

makes a difference to Burr’s claims. 9 We hold that the district court correctly rejected

Burr’s habeas petition.

 (5th Cir. 2014) (noting the practice is “troubling”); Philbrook v.
Ansonia Bd. of Educ., 

, 53 (2d Cir. 1991) (collecting cases on this point).
       9
         In addition to this transcript, Burr also seeks to rely on appeal on medical evidence
first introduced during the federal proceedings. Pinholster squarely precludes our
consideration of this evidence. Pinholster, 563 U.S. at 180–81.

                                              19
                                              A.

       Burr only alleges one factual error on appeal: he claims that the MAR court erred in

concluding “that Susie died of a depressed skull fracture.” Opening Br. at 30. To satisfy

§ 2254, he would have to show that “the state [MAR] court based its decision[] ‘on an

objectively unreasonable factual determination in view of the evidence before it, bearing

in mind that factual determinations by state courts are presumed correct absent clear and

convincing evidence to the contrary.’” Elmore v. Ozmint, 

 (4th Cir. 2011)

(quoting Baum v. Rushton, 

 (4th Cir. 2009)) (citing § 2254(e)(1), (d)(2)).

An “unreasonable determination of the facts,” as the phrase is used in § 2254(d)(2), “is not

merely an incorrect determination, but one ‘sufficiently against the weight of the evidence

that it is objectively unreasonable.’” Gray v. Zook, 

 (4th Cir. 2015)

(emphasis added) (quoting Winston I, 

).

       Burr cannot satisfy this hefty burden. As an initial matter, the MAR court did not

find that Susie died of a depressed skull fracture; rather, it is clear the court was persuaded

by the medical testimony at trial that Susie’s cause of death was a brain injury resulting

from child abuse. However, the court did find that she had a depressed skull fracture. We

will assume that is the factual finding Burr takes issue with. In light of the conflicting

testimony on that point, the MAR court’s finding was not “objectively unreasonable.” 

       At trial, the chief of pediatric radiology and a pediatric neurologist each testified

that Susie had a skull fracture. The radiologist took the jury through the CT scans and

showed them where he saw “a fracture to the skull.” J.A. 2712. He explained that there was

“a notch, as if something had hit the skull and pushed this portion of the skull into the inner


                                              20
table of the skull itself,” creating a “little v-shaped depression in the skull.” J.A. 2713–14.

The neurologist reviewed the same CT scan and testified that “the skull was caved in in

that area.” J.A. 2847. Perhaps for that reason, Burr’s counsel referred to a “skull fracture”

in argument to the court. J.A. 4219.

       Certainly, other experts who viewed the skull using other methods did not see a

fracture. The initial treating physicians in the emergency room performed an X-ray that

“revealed no obvious [skull] fracture,” J.A. 2829, and the autopsy report indicates

“SKULL: No fractures,” J.A. 1012. The chief of pediatric surgery testified that Susie “had

one area that looked like a fracture on the [X]-ray and the CT scan,” but that an area of the

skull can be pushed in like a dent on a ping-pong ball, without the bone actually breaking.

J.A. 2906 (emphasis added). The pathologist who performed the autopsy agreed, testifying

that this is particularly possible for infants, because their bones “are not completely

calcified, so they are more likely to be deformed [meaning dented in or depressed] by an

injury rather than broken.” J.A. 2979.

       In other words, the disagreement was not between, say, some experts saying Susie

had a “linear skull fracture[] or crack in the skull” and others saying she had a depressed

injury, J.A. 2859, or between some experts saying her skull was fractured in one spot and

others saying it was dented in a different spot, or between some experts saying she had a

head injury and others disputing that point altogether. The experts agreed that Susie’s skull

had an indentation on the left side with associated brain injuries. They disagreed only as to

whether the bone was actually fractured at the site.




                                              21
       Given the competing medical expert testimony on the latter fact, it was not

“sufficiently against the weight of the evidence” so as to be “objectively unreasonable” for

the MAR court to conclude that the chief of pediatric radiology and pediatric neurologist

correctly interpreted the CT scan as demonstrating a fracture of the skull. Winston I, 

. “To the extent multiple interpretations of the facts may exist, the . . . state

court’s determination of the facts . . . is not [objectively] unreasonable.” Duke v. Allen, 

 (11th Cir. 2011). After all, “[i]f reasonable minds reviewing the record

might disagree about the finding in question, on habeas review that does not suffice to

supersede the trial court’s determination.” Brumfield v. Cain, 

 (2015)

(internal quotation marks and alterations omitted); cf. Wainwright v. Goode, 

 (1983) (holding, under pre-AEDPA law, that the lower court “erred in substituting its

view of the facts for that of the [state court]” where “the record [was] ambiguous” and

therefore both views “[found] fair support in the record”).

       We further note that, even if Burr is correct that the MAR court’s finding that there

was a skull fracture was erroneous, “we are at a loss to see much critical significance”

regarding “the existence or nonexistence of an actual fracture to the skull itself.” Burr, 

 n.6. As we noted the last time this case was before us, “[a]ll of the treating

physicians and the medical examiner agreed that the cause of Susie’s death was blunt force

head trauma, and its resulting swelling and pressure in the brain, and that significant force

was necessary to cause this trauma. Burr presented no evidence to the state MAR court that

the treating physicians would have changed their opinions regarding child abuse vis-à-vis

accident based upon the difference in the radiographic evidence and the autopsy report.”


                                              22


 It is hard to see how the MAR court’s decision could have been “based on” a fact that,

in the context of all the other evidence in this case, represented a minor discrepancy. 

(d)(2) (emphasis added); cf. DelValle v. Armstrong, 

 (2d

Cir. 2002) (rejecting a § 2254(d)(2) claim where the state supreme court’s misstatement of

fact was “irrelevant”); Green v. Travis, 

 (2d Cir. 2005) (Sotomayor, J.)

(distinguishing DelValle because, in Green, the precise facts “were not tangential” to the

claim, “but central to it,” which “distinguishe[d] [Green] from those situations[, like

DelValle,] in which a state court’s misunderstanding of the facts of a case had little bearing

on the state court’s ultimate resolution of the claim”). So, we conclude that Burr has not

demonstrated that the MAR court’s decision was “based on an unreasonable determination

of the facts.” 

(d)(2).

                                             B.

       Burr’s § 2254(d)(1) claims fare no better. He contends that the MAR court’s

decision “involved an unreasonable application of” Brady, Napue, and their progeny. Id.

§ 2254(d)(1). We cannot agree.

                                              i.

       We begin with the applicable legal principles. In Brady v. Maryland, the Supreme

Court held that “the suppression by the prosecution of evidence favorable to an accused

upon request violates due process where the evidence is material either to guilt or to

punishment, irrespective of the good faith or bad faith of the prosecution.” Brady, 

. “To establish a Brady violation, the accused must demonstrate that the evidence was”

(1) “suppressed by the prosecution;” (2) “favorable to the defendant, either because it [was]


                                             23
exculpatory or impeaching;” and (3) “material.” Horner v. Nines, 

 (4th

Cir. 2021). There is no dispute that the evidence here was suppressed, but the parties

disagree on the other two prongs. Because the parties and the MAR court focused chiefly

on materiality, and because we can affirm the district court’s denial of the petition on that

basis, we train our analysis primarily on that prong.

       “Favorable ‘evidence is material only if there is a reasonable probability that, had

the evidence been disclosed to the defense, the result of the proceeding would have been

different. A “reasonable probability” is a probability sufficient to undermine confidence in

the outcome.’” 

 at 206 (quoting United States v. Bagley, 

 (1985)).

Thus, “Brady does not ‘automatically require a new trial whenever a combing of the

prosecutors’ files after the trial has disclosed evidence possibly useful to the defense but

not likely to have changed the verdict.’” 

 (quoting Bagley, 

).

       “[U]nder [the Supreme Court’s decision in] Napue [v. Illinois], the government

‘may not knowingly use false evidence, including false testimony, to obtain a tainted

conviction’ or ‘allow[] it to go uncorrected when it appears.’” United States v. Chavez, 

 (4th Cir. 2018) (quoting Napue, 

). False testimony includes

both perjury and evidence that, “though not itself factually inaccurate, . . . creates a false

impression of facts which are known not to be true.” Hamric v. Bailey, 

(4th Cir. 1967). Convictions “obtained by the knowing use of perjured testimony [are]

fundamentally unfair, and must be set aside if there is any reasonable likelihood that the




                                             24
false testimony could have affected the judgment of the jury.” Chavez, 

(quoting United States v. Agurs, 

 (1976)).

                                               ii.

       Burr makes several interpretive moves to position Bridges and Scott’s testimony as

vital and therefore his Brady and Napue claims based on their testimony as viable. Pointing

to medical evidence introduced before the MAR court and the conflicting testimony

regarding whether Susie had a skull fracture, Burr argues that Susie died from a closed

head injury but did not have a fractured skull. If Susie did not have a fractured skull, Burr

contends, it is possible that her head trauma resulted from being dropped—such as in the

fall with Scott—rather than from a concentrated blow. Therefore, he argues, by the time of

the MAR proceedings, “the only basis for the State’s theory that the fall [with Scott] did

not cause the fatal injury, was the description of that fall at trial by” Bridges and Scott.

Opening Br. at 34. Thus, Burr posits that Bridges and Scott’s credibility was absolutely

essential to the case, because “[e]vidence about whether the fall could have caused the

injury . . . was literally the entire case.” Id. at 32. And therefore, Burr’s theory goes, he can

satisfy the materiality element of the Brady and Napue claims because anything that

undermines the credibility of such key witnesses necessarily calls into question the validity

of the jury’s verdict.

       The central problem with Burr’s argument is that the State’s case was not so entirely

reliant on Bridges and Scott, or on their descriptions of the fall, as he suggests. We have

already noted “the overwhelming medical evidence that Susie was a victim of child abuse.”

Burr, 

. While we made that comment in the different context of


                                               25
analyzing the effectiveness of the assistance of counsel, the record evidence remains the

same. The conclusion that abuse, rather than the fall with Scott, caused Susie’s death was

supported by: (1) significant medical testimony from treating physicians and the

pathologist, including regarding the type of head injury, the limb fractures, the retinal

hemorrhages, and the bruising all over Susie’s body—such as a handprint-shaped bruise

on her neck—all of which was consistent with abuse rather than an accidental fall; (2)

eyewitness testimony, albeit somewhat mixed, from those who saw the fall, which included

Jonas in addition to Bridges and Scott; (3) testimony from several individuals, including

Burr, who examined Susie after the fall and found her, in their lay opinions, to be uninjured;

and (4) medical expert testimony that Susie would have been so clearly unwell immediately

after sustaining the head injury that ultimately killed her that even a layperson would have

recognized she was in danger, such that the head injury must have occurred later than the

time of the fall (around 6:00 P.M.) if Susie did not show symptoms until after midnight.

       With this background and our deferential standard of review in mind, we can easily

dispense with Burr’s Brady and Napue claims.

                                             iii.

       Burr argues that the MAR court unreasonably applied Brady when it concluded that

“any inconsistencies between the trial testimony of [Bridges and Scott] and their pre-trial

comments to the prosecutors are of de minimis significance.” J.A. 1827. He contends that

the undisclosed tapes would have enabled trial counsel to materially undermine Bridges

and Scott’s testimony on crucial points related to Scott’s fall with Susie and the incidents

when Scott had previously seen Burr shake Susie. Therefore, he argues, the tapes are


                                             26
material because it is reasonably likely that “had the [tapes] been disclosed to the defense,

the result of the [trial] would have been different.” Horner, 

. We disagree

as to Bridges, Scott, and the evidence considered in the aggregate.

       We begin with Burr’s arguments related to the suppressed 1993 interview of

Bridges. Burr cherry-picks statements from that interview in an attempt to demonstrate that

the interview provides new, damning information. Much of what he picks out is made up

of questions from the prosecutors about how Bridges could not have known that Susie had

at least some fractured limbs for days before she sustained her head injury. But Bridges’s

credibility as to Susie’s demeanor before the night in question was thoroughly aired at trial.

It had to have been apparent to the jury that any testimony that Susie was happy and did

not cry too much before August 24, 1991, was questionable given that she had at least some

fractured limbs for days beforehand. But that dubious testimony did not only come from

Bridges’s family; Burr also testified that Susie seemed “fine” earlier in the evening of

August 24. J.A. 3190–91. Moreover, the evidence about Susie’s demeanor was

inconsistent, as Bridges and her family also testified that Susie had been crying a lot over

the preceding weeks, which they chalked up to her throat infection, a result of oral thrush.

Bridges explained that Susie “cried constantly” because of the pain in her throat and

contended that she assumed this was related to oral thrush rather than realizing that Susie

had broken bones. J.A. 2209. Accordingly, the conversation between Bridges and

prosecutors about whether and to what degree Susie was in pain before the night of August

24 is cumulative evidence to that provided at trial. Such cumulative evidence “is generally




                                             27
not considered material for Brady purposes.” Juniper v. Zook, 

 (4th Cir.

2017) (quoting Johnson v. Folino, 

 (3d Cir. 2013)).

       Burr points to another statement in the transcript that was also aired at trial. In the

suppressed transcript, Bridges said that Burr “did not act like he would hurt [Susie].” J.A.

1571. Burr contends that this contradicts Bridges’s trial testimony, which “listed a litany

of Burr’s abusive behavior suggesting he could hurt a child.” Opening Br. at 17. But at

trial, Bridges was also asked why it had not occurred to her immediately that Burr could

have caused Susie’s injuries. She explained this was because she does not hurt children,

“and you can’t see it in someone else. . . . [P]eople hurt people, but they don’t hurt a child.”

J.A. 2076; see also J.A. 2183–84 (showing Bridges being challenged at trial with

statements she made to a social worker soon after Susie was injured, in which she said she

could not identify who hurt Susie).

       Burr further contends that Bridges “admit[ted]” in the transcript “that she asked her

family to lie for her.” Opening Br. at 40. He overstates the limited nature of Bridges’s

concession, in which she acknowledged that she had sought to prevent one of her sisters

from saying “that the baby cried all the time” because she was worried about “look[ing]

bad.” J.A. 1587. Still, this statement could certainly have undermined Bridges’s credibility

had the defense known about it and been able to impeach Bridges’s testimony with it. But

again, substantial contradictory evidence at trial allowed the jury to weigh whether Bridges

and her family were being truthful regarding Susie’s condition. The MAR court’s

conclusion that this undisclosed statement would not have changed the outcome of the trial

was not unreasonable.


                                              28
       Finally, Burr argues that the prosecutors coached Bridges to provide a believable

story about why she did not realize that Burr was abusing Susie. For example, the

prosecutor suggested that Bridges’s rationale could have been, “I was so in love with this

guy that I didn’t want my sister to know, or say anything to him if she saw them getting

spanked. . . . I was love blind or something.” Opening Br. at 40–41 (quoting J.A. 1586).

But the prosecutor went on to add, “[o]r whatever it was.” 

 at 41 (quoting J.A. 1586).

That is, the prosecution asked Bridges to provide whatever story was the accurate one.

Further, the MAR court concluded that “the prosecutors were most concerned about

assuring that [Bridges] testified truthfully.” J.A. 1830. That was not an objectively

unreasonable conclusion. The transcript reveals that prosecutors repeatedly urged Bridges

to tell “[e]very little [shred] of” the truth, “no matter how bad it ma[de] [her] look.” J.A.

1585. So, in our view, Burr’s “coaching” argument cannot provide the requisite showing

of materiality.

       Burr’s arguments regarding the suppressed interview with Scott are equally

unpersuasive. To be sure, Scott comes across in the interview transcript as, in the magistrate

judge’s words, “confused, scared[,] and easily susceptible to suggestion.” Burr, 

, at *18. So, Burr contends, “[i]n the hands of competent counsel, Scott’s repeated

contradictions, his embellishments, and his mistakes could have been used to prove to the

[j]ury that his memory of [the fall and the shaking incidents] was not trustworthy.” Opening

Br. at 45. But the jury could assess for themselves how much weight to give Scott’s

testimony. At the time of trial, Scott was only ten years old and was testifying about

traumatic events that occurred when he was eight. His testimony at trial was also littered


                                             29
with “I don’t remember” statements. E.g., J.A. 2774–89. So even without the suppressed

interview transcript, it is apparent from the trial evidence alone that Scott was an imperfect

witness, a point that defense counsel could and did drive home to the jury. E.g., J.A. 4054–

55 (defense counsel pointing in closing to how Scott’s story regarding the fall had changed

over time). The interview transcript would have been cumulative evidence on that point.

       Moreover, Scott’s trial testimony that he “cradled” Susie during his fall such that

she never hit the ground was echoed by testimony from Bridges and Jonas and was

supported by the lack of abrasions on Susie’s skin, “[e]ven over the area on the left part of

the skull” where her head was most badly injured. J.A. 2916. And, to the extent Scott’s

description of the fall can be disputed by statements he made in the suppressed transcript,

it was already able to be disputed by similar evidence provided to Burr before trial. As

noted, in the weeks following Susie’s death, both Scott and Bridges at times described the

fall as Scott “dropp[ing]” Susie or “fall[ing] on top of her.” J.A. 2262, 3398; see also J.A.

3415, 3425–26, 3779. Finally, Burr contends that prosecutors planted in Scott’s mind the

idea that he did not hurt Susie, and that this led him to change his description of the fall.

But Scott’s descriptions had shifted over time even before that point, and a social worker

was the first person in the record to urge him that he had not hurt Susie—just days after

she sustained her fatal head injury.

       Burr also argues that the suppressed interview is revealing regarding Scott’s

description of the incidents in which he saw Burr “shake” Susie. Specifically, he alleges

that the transcript demonstrates that Scott’s story about those incidents shifted in response

to leading questions. But this argument fails for at least three reasons.


                                             30
       First, Burr does not directly explain why the MAR court’s reading of the record—

which rejected Burr’s view of the transcript—was objectively unreasonable.

       Second, he fails to explain how the transcript would have allowed him to impeach

Scott’s testimony. The MAR court found that any inconsistencies in Scott’s testimony were

“of de minimis significance.” J.A. 1842. This was not unreasonable, particularly because

Burr had an opportunity to cross-examine Scott at trial on the question of whether

prosecutors had planted the idea of Burr shaking Susie in Scott’s mind. At the time of trial,

Burr was aware of Scott’s September 5, 1991, statement that he had never seen Burr “do

anything” to Susie, such as whipping her—a statement that contradicted his trial testimony

that he had seen Burr shake Susie multiple times. J.A. 1655. Defense counsel questioned

Scott about that statement at trial. Additionally, Scott conceded at trial that he did not

mention the shaking incidents to anyone else before he told the prosecutors about them. He

explained that this was because he was scared of Burr and thought that, if he told Bridges

about Burr shaking Susie, Burr might kill Bridges. Burr has not clarified how having access

to the suppressed transcript would have assisted his trial strategy rather than just being

cumulative to the evidence he was already aware of.

       Third and finally, Burr cannot sustain his burden to show as objectively

unreasonable the MAR court’s conclusion that, had Burr had the opportunity to impeach

Scott’s testimony using this transcript, the result of the trial would not have been different.

There is no way to know if the jury believed Scott’s story about the shaking incidents based

on the evidence before them. It is possible they did not, as he was a child whose credibility

could be questioned given that he was frightened of Burr and believed Burr had killed his


                                              31
baby sister. But even if the jury believed Scott was being truthful about Burr shaking Susie,

and even if the impeachment value of the suppressed transcripts would have been the

evidence to tip them against finding Scott credible on that point, there was plenty of other

evidence from which they could nevertheless infer that Burr was Susie’s assailant in the

assault that led to her death.

       For example, as the MAR court noted, the evidence “demonstrate[d] that [Burr]

abused Susie on [an]other occasion[]”: Bridges testified to a prior incident in the weeks

before Susie’s death in which Bridges had awoken around 4:00 A.M. to the sound of Susie

screaming loudly and found Burr in another room holding her. J.A. 1426. The MAR court

further found that the jury “had the opportunity to carefully evaluate [Burr]’s [own]

extended testimony and demeanor on the witness stand” and “obviously conclud[ed] that

[Burr] was not being truthful with them” about the events of the night in question. J.A.

1832. To name just some of the other circumstantial evidence from which the jury could

have inferred guilt, there was testimony that Burr was abusive toward Bridges and his own

toddler son; numerous witnesses testified that Susie was fine before she was left alone with

Burr on the night in question; Bridges’s niece testified that she went to Bridges’s trailer

during the forty-five-minute window in which Bridges was washing dishes next door, and

that at that time she leaned over to kiss Susie and saw no bruises or markings on her;

medical experts testified that Susie’s seizure-like symptoms, which she exhibited when

Bridges returned to her trailer, would have occurred very soon after the head injury; and

medical experts testified that the retinal hemorrhages Susie displayed “would require really




                                             32
very violent shaking,” J.A. 2876, and that her head injury would require a “great deal of

force,” J.A. 2739, 2847.

       Our view of the suppressed transcripts of conversations with Bridges and Scott does

not change when we consider its possible impeachment value in the aggregate. For the

reasons described above, many of the aspects of the transcripts Burr points to are

cumulative to testimony that was presented, or impeachment opportunities that already

existed, at trial. When the cumulative evidence is put aside, what remains is too

insignificant to pose a realistic possibility of altering the trial outcome had Burr’s counsel

been aware of it before trial. See J.A. 1827 (MAR court finding that “any inconsistencies

between the trial testimony of [Bridges and Scott] and their pre-trial comments to the

prosecutors are of de minimis significance”). The MAR court concluded that, “[a]t the

bottom line,” Burr “ha[d] not presented anything that undermine[d] the Court’s confidence

in the outcome of the trial proceeding.” J.A. 1842. We cannot say that this was objectively

unreasonable.

       In short, the MAR court’s conclusion that Burr had not shown “a reasonable

probability that[,] had [the tapes] been disclosed[,] the result of the trial would have been

different,” J.A. 1827, did not amount to an “unreasonable application” of Brady, 

(d)(1). 10


       10
           Burr also briefly argues that the MAR court “diminish[ed] the value of
impeachment for Brady purposes” and that this led it to “engage[] in an interpretation that
[was] contrary to Brady and established law.” Opening Br. at 39; see 

(d)(1) (relief may be granted where the state court’s decision was “contrary to . . .
clearly established Federal law”). He does not explain how the MAR court “diminished the


                                             33
                                              iv.

       We reach the same conclusion regarding Burr’s Napue claim. As noted, Napue

requires courts to set aside a conviction as violative of due process if there is “any

reasonable likelihood that the false testimony could have affected the judgment of the

jury.” Chavez, 

. Further, “[t]he government does not have to solicit the false

evidence; it is enough if the government allows the evidence to go uncorrected when it

surfaces.” United States v. Griley, 

 (4th Cir. 1987).

       Burr’s Napue claim rests on the same undisclosed evidence that forms the basis of

his Brady claim. He does not contend that the State knowingly relied on perjured testimony,

but he does allege that the State denied him due process by creating false impressions of

critical facts at trial. He points to three key ways in which he claims the State did this. But

we are not persuaded.

       First, Burr contends that the prosecutors allowed Bridges and Scott to present a

“sanitized” rendition of Scott’s fall with Susie that was “simply not true.” Opening Br. at

49. Specifically, the suppressed 1993 interview tape with Bridges “show[s] that as late as

several weeks before trial, Bridges and the State continued to refer to Scott dropping Susie,

or falling with her.” 

 at 48–49. Thus, the theory goes, the trial testimony that Scott




value of impeachment for Brady purposes,” and our review of the MAR court’s order
reveals that the MAR court explicitly recognized that evidence can be favorable for Brady
purposes “either because it is exculpatory, or because it is impeaching.” J.A. 1822 (quoting
Strickler v. Greene, 

 (1999)). The MAR court applied the correct
standard; it simply found the evidence to be immaterial.

                                              34
“cradled” Susie “created a false impression of the fall—a ‘sanitized’ version which

[prosecutors] knew to be sanitized.” Id. at 49.

       This argument fails for at least three reasons. One, “[m]ere inconsistencies in

testimony by government witnesses do not establish the government’s knowing use of false

testimony.” Griley, 

 (citing Overton v. United States, 

(5th Cir. 1971) (per curiam)). Two, the trial evidence included the contradictory

descriptions of the fall that Scott and Bridges had provided over time, such as testimony

that Bridges told a deputy sheriff that Scott “fell on” Susie and told a social worker that

Scott had “dropped” Susie. J.A. 3617, 3779. So the evidence presented at trial was not a

clean, “sanitized” version of events; it included competing statements, including

contradictory statements from the same witnesses. Three, and relatedly, even without

access to Bridges’s suppressed 1993 statement, Burr had plenty of evidence from which he

could have impeached Scott and Bridges’s “sanitized” accounts of the fall, namely, their

inconsistent statements about it. And he did indeed use these discrepancies to undermine

their credibility. See J.A. 4021 (defense counsel noting in closing that the defense had

questioned witnesses about the fall and stating that those questions “[went] to the

credibility of the witnesses”).

       Burr’s second argument in favor of his Napue claim is similar. He alleges that the

State “did not believe Bridges[] and her family when they said that Susie was uninjured

before the evening of August 24.” Opening Br. at 49. In the suppressed 1993 interview,

prosecutors accused Bridges of having coordinated her children’s and relatives’ stories to

give the impression that Susie was a “normal” and “[h]appy” baby who rarely cried, saying


                                             35
that her family’s testimony had been “like a script.” J.A. 1593. And Bridges indicated that

she had tried to prevent one of her sisters from saying “that the baby cried all the time” out

of concern that she would “look bad.” J.A. 1587. Yet, according to Burr, she presented the

same “script” on the witness stand without a word from the prosecution.

       In that same transcript, however, one of the prosecutors specifically stated that he

was “playing a devil’s advocate”—that he was pressing Bridges’s story for weak points

because he was “looking at [how] the defense attorneys are going to jump on you.” J.A.

1579. So, it is not clear that the prosecutors actually disbelieved Bridges; they may have

simply been preparing her for the hard questions they knew would be coming at trial.

Indeed, the MAR court found that the prosecutor’s “devil’s advocate” statement

“reveal[ed] [their] motive.” J.A. 1829.

       Further, the question of how Susie behaved before the night of August 24 was

thoroughly aired at trial. The jury heard both evidence that Susie was a happy baby who

did not cry much, and evidence that she had been crying quite a bit in the weeks leading

up to the night of August 24 due to pain in her throat. So, again, the jury was not left with

a “sanitized” version of events—they were given competing testimony that they had to

weigh. The jury could judge for themselves how the family’s “script” accorded with the

medical facts and, if it did not, factor that into their credibility determinations.

       Third and finally, Burr alleges that prosecutors were wrong to rely so heavily on

Scott’s testimony when they knew he had trouble remembering facts and was prone to

embellish. But, again, the jury could make its own determination of how much weight to

give the statements of a young child who repeatedly stated that he could not remember key


                                               36
details. Burr has not pointed to any “false impression” about Scott’s testimony that

prosecutors should have been aware of and flagged but that the jury would not also have

been aware of after listening to Scott’s testimony. So, again, the MAR court did not err by

rejecting this claim.

       In sum, we cannot say that the MAR court’s determination that the Napue claim was

without merit was unreasonable “beyond any possibility for fairminded disagreement.”

Harrington, 

.

                                                  v.

       That leaves us with one final question. We have concluded that the MAR court did

not base its opinion on unreasonable determinations of fact or unreasonably apply clearly

established federal law based on the record the court had before it in 2000. But what are

we to make of the suppressed transcript of a 1992 conversation with Bridges that was not

fully turned over to Burr until 2015—decades after trial and fifteen years after the

proceedings before the MAR court wrapped up? May we consider this transcript, even

though it was not part of the record before the MAR court?

       Burr urges us to do so. And this request raises numerous fascinating questions about

how the Supreme Court’s decisions in Brady and Pinholster intersect. We ultimately

conclude, however, that we need not resolve these questions because, even if we consider

the transcript, it does not alter our analysis.

       Some background will help explain why this is a difficult question. Section 2254(d)

prevents federal courts from granting habeas relief on “any claim that was adjudicated on

the merits in State court proceedings” unless the state court’s decision “was contrary to, or


                                                  37
involved an unreasonable application of, clearly established Federal law” (§ 2254(d)(1)) or

“was based on an unreasonable determination of the facts in light of the evidence presented

in the State court proceeding” (§ 2254(d)(2)). 

(d) (emphasis added).

Section § 2254(d)(2), on its face, restricts the federal court’s evaluation of claims that were

adjudicated on the merits by the state court to the evidence that was before the state court.

Pinholster, 

 n.7. And the Supreme Court has instructed that federal habeas

review under § 2254(d)(1) of claims adjudicated on the merits in state court is similarly

restricted to the record before the state post-conviction relief court. Id. at 180–81.

       But “[u]nderpinning the Supreme Court’s discussion in Pinholster is the terse

acknowledgment that the habeas petitioner’s claims [in that case] had been adjudicated on

the merits in state-court proceedings.” Winston v. Pearson (Winston II), 

(4th Cir. 2012) (emphasis added). And we have held, even after Pinholster, that a claim

was not “adjudicat[ed] on the merits for purposes of § 2254(d)” when the state court made

its decision “on a materially incomplete record.” Id. at 496 (quoting Winston I, 592 F.3d at

555–56). “In this rare scenario, the gloves come off: The federal habeas inquiry is more

penetrating, and—if consistent with statute and the Rules Governing § 2254 Cases—we

may hear evidence that would otherwise be immaterial under § 2254(d)’s limited review.”

Valentino, 

. In these circumstances, § 2254(d) deference does not apply,

and federal courts instead review the claim de novo, applying deference only to any factual

findings the state court actually made pursuant to § 2254(e)(1). Winston II, 

,

500–01; see also Valentino, 

.




                                              38
       Winston and related cases provide a narrow exception to Pinholster that we have

held arises where the “state court shuns its primary responsibility for righting wrongful

convictions and refuses to consider claims of error.” Valentino, 

. Thus, we

have applied the exception in scenarios where a state court “unreasonably refuse[d] to

permit further development of the facts of a claim.” Gordon v. Braxton, 

(4th Cir. 2015) (quoting Winston II, 

) (internal quotation marks omitted);

cf. Hurst v. Joyner, 

 (4th Cir. 2014) (petitioner could develop claim under

§ 2254(e)(2) where “the state MAR court unreasonably denied [his] motion for further

evidentiary development”). “In this circumstance, we do not offend the principles of

‘comity, finality, and federalism’ that animate AEDPA deference because the state court

has ‘passed on the opportunity to adjudicate [the] claim on a complete record.’” Gordon,

 (quoting Winston I, 

). Put another way, “[w]hen a state

court forecloses further development of the factual record, it passes up the opportunity that

exhaustion ensures.” Winston II, 

 (quoting Winston I, 

). But

that raises the question: What if the state court made its decision on a materially incomplete

record, not through the fault of the state court itself, but instead because the State

suppressed evidence, in potential violation of Brady?

       That is the situation we face regarding the tape recording of the 1992 interview of

Bridges. The Supreme Court of North Carolina remanded the MAR court’s initial denial

of Burr’s petition in order for the State to comply with N.C. Gen. Stat. § 15A-1415(f). At

the time, that statute required the State, “to the extent allowed by law, [to] make available

to [a] capital defendant’s counsel the complete files of all law enforcement and


                                             39
prosecutorial agencies involved in the investigation of the crimes committed or the

prosecution of the defendant.” 11 N.C. Gen. Stat. § 15A-1415(f) (1996). So, here, the state

courts did not prohibit Burr from obtaining the tape recording in question. Rather, it was

the State that failed to comply by turning over all relevant documents. 12

       The U.S. Supreme Court has contemplated a similar scenario in dicta—dicta that

we partially relied on in Winston II. Pinholster held that, for claims adjudicated on the

merits in state court, the petitioner “must overcome the limitation of § 2254(d)(1) on the

record that was before that state court.” Pinholster, 

. In dissent, Justice

Sotomayor criticized that interpretation as precluding relief for some petitioners with Brady

claims. She contemplated the following scenario, which is not unlike the situation before

us now: A petitioner “diligently attempt[s] in state court to develop the factual basis of a

claim that prosecutors withheld exculpatory witness statements in violation of Brady,” but

“[t]he state court denie[s] relief on the ground that the withheld evidence then known d[oes]

not rise to the level of materiality required under Brady.” 

 (Sotomayor, J.,

dissenting). However, before the deadline for filing a federal habeas petition passes, “a

state court orders the State to disclose additional documents the petitioner had timely

requested under the State’s public records Act. The disclosed documents reveal that the

State withheld other exculpatory witness statements, but state law would not permit the

petitioner to present the new evidence in a successive petition.” 

 at 214–15. Justice


       11
         The same statutory requirement exists today, but it is no longer restricted to capital
defendants.
      12
         Whether that failure was intentional or accidental is irrelevant for Brady purposes.
Brady, 

.

                                              40
Sotomayor noted that “it is unclear how [such a] petitioner can obtain federal habeas relief

after” Pinholster. 

.

       In a responsive footnote, the Pinholster majority “suggested . . . that the prohibition

on new evidence might not always apply.” Hanna v. Ishee, 

 (6th Cir.

2012). The Court stated that, while it “d[id] not decide where to draw the line between new

claims and claims adjudicated on the merits,” the facts of Justice Sotomayor’s hypothetical

“may well present a new claim.” Pinholster, 

 n.10 (majority opinion)

(emphasis added). The majority made this assertion even though, in the stated hypothetical,

“the new evidence merely bolster[ed] a Brady claim that was adjudicated on the merits in

state court.” 

 (Sotomayor, J., dissenting) (emphasis added). And our decision in

Winston II relied in part on Pinholster’s “tacit acknowledgment that [a] hypothetical

petitioner [presenting a Brady claim] would be free to present new, material evidence.”

Winston II, 

.

       We are left, however, with a plethora of unanswered questions. Most notably for

present purposes, even if we can consider new Brady evidence—as Pinholster suggests

and as we noted in Winston II—what does that mean for our standard of review? 13 Our


       13
          Another question posed by this case, but which we need not answer here, is
whether it matters that the State did not object below to the inclusion of the 2015 transcript
in the record, that the district court opted to include it in the record pursuant to Rule 7 of
the Rules Governing Section 2254 Cases, that the State has not appealed that ruling, or that
the State continues to take the position that we can review the 2015 transcript. Our sister
circuits have suggested the answer is no. See Frazier v. Bouchard, 

 (11th
Cir. 2011) (holding that Pinholster precluded the court from considering “the expanded
record[] presented to the district court” under Rule 7); Moore v. Mitchell, 

 (6th Cir. 2013) (applying Pinholster’s prohibition on new evidence even “when


                                             41
Winston line of cases considers the entire claim, not just the new evidence, de novo, albeit

with the appropriate deference to the state court for factual findings it could actually make

based on the evidence before it. See 

 at 492–93, 496–97, 500–01 (affirming a grant of

habeas relief after having remanded to the district court to reconsider the entire claim de

novo—with deference to “relevant factual findings made by the state court” under

§ 2254(e)(1)—where the petitioner alleged his trial attorneys were ineffective for failing

to raise a claim that “his mental retardation categorically barred imposition of a death

sentence” and where new evidence did not fundamentally alter the claim so as to render it

unexhausted); Valentino, 

 (noting that if the claim was not adjudicated on

the merits by the state court, “we must remand to the district court for de novo review and

a possible evidentiary hearing”). Arguably, the same rule should apply for Brady claims:

although individual pieces of evidence could be analyzed either de novo or with § 2254(d)

deference, depending on whether or not they were before the state court, Brady also asks

us to consider evidence in the aggregate. Juniper, 

. An aggregate review is

more easily performed if all of the evidence is to be reviewed under the same standard.

       That said, our sister circuits considering Brady claims post-Pinholster have taken

approaches that differ both from each other and from our Winston approach. The Sixth

Circuit has evaluated the previously disclosed evidence that was before the state court


the parties jointly move to expand the record,” reasoning that “by agreeing to look at
evidence beyond the state record[,] we would be permitting the parties to declare their own
standard of review”); Champ v. Zavaras, 

 (10th Cir. 2011)
(“Although [Pinholster] dealt with new evidence that the district court admitted in the
context of an evidentiary hearing, this newly articulated rule applies with equal force to
any expansion of the record under Habeas Rule 7.”).

                                             42
under the § 2254 standard, while evaluating evidence discovered during the federal habeas

proceedings de novo to determine whether it supported a Brady claim. Jones v. Bagley,

, 486–87 (6th Cir. 2012); see also Hanna, 

 (reviewing new

evidence de novo). By contrast, the Ninth Circuit has evaluated the new materials only to

determine if, combined with the old materials, they presented a “potentially meritorious”

Brady claim, at which point the court remanded the case to the district court with

instructions to stay federal proceedings so the petitioner could present his claim “in the

first instance to [the] state court.” Gonzalez v. Wong, 

 (9th Cir. 2011).

       We need not, and do not, resolve this question today. Even assuming, purely for the

sake of argument, that we may consider the entirety of Burr’s Brady claim de novo, we

would still affirm the denial of Burr’s petition. See Stokley v. Ryan, 

 (9th

Cir. 2011) (analyzing a claim assuming that Pinholster did and, alternatively, did not apply,

and holding that “[e]ven considering the new evidence, we conclude that [the petitioner]

has not presented a colorable claim of ineffective assistance of counsel”); Hanna, 

 (opting to deny a Brady claim related to newly discovered evidence on the merits,

notwithstanding a failure to exhaust).

       We need not dwell long on the evidence that was before the state MAR court. For

the reasons discussed at length above, Burr has not shown that the suppressed transcripts,

individually or in combination, were material for Brady purposes—that there “is a

reasonable probability that, had the evidence been disclosed to the defense, the result of

the proceeding would have been different.” Horner, 

 (quoting Bagley, 

). Although our analysis above paid due deference to the state MAR court, it


                                             43
would not change if we were to consider the evidence de novo. Burr has not come close to

establishing that the jury would not have found him guilty had the defense been aware of

the suppressed transcripts, which would have provided at most cumulative or tangential

impeachment opportunities.

       That analysis does not change when we consider the 1992 transcript first fully

revealed in 2015, which was not before the state MAR court. Burr’s counsel conceded

below that this “new” evidence was largely duplicative of evidence already in the record.

And several details he focuses on before this Court were actually included in evidence he

had access to during the trial or MAR proceedings. 14

       The only truly new evidence Burr points to is that the full transcript “reveals that

the prosecutors spent considerable time with Bridges showing her autopsy pictures and

attempting to distinguish between the bruising caused by medical treatment and the bruises

caused by the abuse.” Opening Br. at 26. Burr contends that “the prosecutors described the




       14
           Burr cites Bridges’s description of Scott’s fall, but that or a materially
indistinguishable description was included in a version of the transcript that Burr concedes
was formerly provided to him. Another detail, which Burr describes as “a previously
undisclosed accusation of abuse towards Susie,” was in fact discussed at trial. Opening Br.
at 26 (citing J.A. 5528–29 (describing an incident in which Bridges’s niece accidentally
tripped over Susie’s car seat)); see, e.g., J.A. 2349–51 (trial testimony about this incident).
Similarly, Burr argues that the transcript shows that “interviewers further questioned
Bridges about whether she beat Susie and how she could not notice that her daughter’s
arms and legs had been broken for some time.” Opening Br. at 26. But that, too, was
discussed at length at trial. Finally, there is no evidence in the newly disclosed transcript
(or anywhere else in the record) regarding “an incident when Bridges slapped her infant
out of a car seat with such force that she flew across the floor,” as Burr claims. Id. at 29;
see also id. at 43.

                                              44
bruises they believed were caused by medical care [to Bridges] so that her testimony could

be confined to bruises that existed before Susie went to the hospital.” Id.

       But, at trial, the sources of the bruising were exhaustively covered by the

pathologist. In front of the jury, the pathologist drew circles on photographs of Susie’s

body to note which bruises were caused by medical interventions such as the insertion of

intravenous lines. Bridges’s trial testimony regarding which bruises were present on Susie

before her trip to the hospital was cumulative to that of a medical expert aware of the types

of bruising that would be caused by medical intervention, so evidence that would have

impeached Bridges’s testimony on that point is of limited value. Cf. Juniper, 

 (“Suppressed evidence that would be cumulative of other evidence . . . is generally not

considered material for Brady purposes.” (quoting Johnson, 

)).

       Further, the transcript does not directly undermine Bridges’s testimony. At most,

the suppressed transcript would have provided the defense an opportunity to impeach

Bridges’s cumulative testimony on this point by questioning her about alleged coaching by

the prosecution related to the bruises. We conclude that this evidence is not enough to

support a Brady claim. Burr “has not convinced us that there is a reasonable probability

that the jury would have returned a different verdict if [Bridges’s] testimony [about Susie’s

bruises] had been . . . impeached” in this manner. Strickler v. Greene, 

(1999).

       Finally, in the circumstances of this case, where the new material is of the same kind

as, and largely cumulative of, the old material, and relates only to a tangential issue, we

readily conclude that considering the transcript finally revealed in 2015 would not alter our


                                             45
calculation above regarding the “aggregate effect” of the evidence. Cf. Jones, 

 (concluding that the new evidence did not tip the scales of the “cumulative effect”

because there was “otherwise strong circumstantial evidence of [the petitioner’s] guilt”).

       Accordingly, we leave the questions surrounding the Brady exception to Pinholster

for another day when the issue has been more squarely presented and more thoroughly

briefed. Even considering all the suppressed evidence, old and new, de novo, we conclude

that Burr cannot satisfy Brady.

                                             IV.

       This is a deeply serious case, both because of the nature of the crime and because

of the punishment. But our standard of review renders the analysis of the only claims before

us straightforward. Under § 2254(d), the state court’s judgment stands unless it is contrary

to or involves an unreasonable application of clearly established federal law, or unless it is

based on an unreasonable determination of the facts. And here, the MAR court did not base

its decision on an unreasonable determination of the facts, nor did it unreasonably apply

the principles the Supreme Court laid out in Brady and Napue. Even if we consider the

suppressed transcript that was not before the state court, moreover, our analysis does not

change. We affirm the district court’s decision to deny the petition for habeas relief.

                                                                                 AFFIRMED




                                             46