Amicus Curiae Brief — Darrell Hemphill, Petitioner v. New York

No. 22-488

In the Supreme Court of the United States

__________

DARRELL HEMPHILL, PETITIONER

v.

STATE OF NEW YORK

__________

ON PETITION FOR A WRIT OF CERTIORARI

TO THE COURT OF APPEALS OF NEW YORK

__________

BRIEF FOR NATIONAL ASSOCIATION OF

CRIMINAL DEFENSE LAWYERS AS AMICUS

CURIAE IN SUPPORT OF PETITIONER

__________

MICHAEL MCCONNELL

Wilson Sonsini

Goodrich & Rosati, PC

650 Page Mill Road

Palo Alto, CA 94304

(650) 493-9300

JEFFREY T. GREEN

National Association of

Criminal Defense

Lawyers

1660 L Street NW

Washington, DC 20036

STEFFEN N. JOHNSON

Counsel of Record

PAUL N. HAROLD

GARRETT R. ATHERTON

Wilson Sonsini

Goodrich & Rosati, PC

1700 K Street, NW

Washington, DC 20006

(202) 973-8800

sjohnson@wsgr.com

Counsel for Amicus Curiae

i

QUESTION PRESENTED

Whether an appellate court may excuse constitutional error as “harmless” on the basis of its judgment

that the prosecution’s evidence was “overwhelming”

without engaging in the broader inquiry whether the

error could have affected the jury’s verdict.

ii

TABLE OF CONTENTS

Page

QUESTION PRESENTED......................................... i

TABLE OF AUTHORITIES..................................... iii

INTEREST OF AMICUS CURIAE .......................... 1

STATEMENT ............................................................ 1

SUMMARY OF ARGUMENT ................................... 5

REASONS FOR GRANTING THE PETITION ....... 6

I.

Lower courts are intractably divided over

how to conduct harmless-error review. .............. 7

A. The federal circuits are hopelessly

divided over how to conduct harmlesserror review. ................................................ 7

B. The state appellate courts are equally

divided over how to conduct harmless-error

analysis. ..................................................... 11

II. The decision below conflicts with this

Court’s precedents. ........................................... 14

III. Proper harmless-error analysis is critical to

protecting the rights of criminal defendants

in a large number of cases. ............................... 19

IV. This case is an ideal vehicle for the Court to

clarify the harmless error doctrine. ................. 22

CONCLUSION ........................................................ 23

iii

TABLE OF AUTHORITIES

Page(s)

Cases

Arizona v. Fulminante,

499 U.S. 279 (1991) ............................................ 17

Bartley v. People,

817 P.2d 1029 (Colo. 1991) ................................ 12

Blakely v. Washington,

542 U.S. 296 (2004) ............................................ 20

Bollenbach v. United States,

326 U.S. 607 (1946) ............................................ 20

Brecht v. Abrahamson,

507 U.S. 619 (1993) ............................................ 18

Chapman v. California,

386 U.S. 18 (1967) ......................... 5, 14, 15, 17, 18

Crawford v. Washington,

541 U.S. 36 (2004) ............................................ 3, 4

Delaware v. Van Arsdall,

475 U.S. 673 (1986) ....................... 5, 17, 18, 19, 22

Gov’t of Virgin Islands v. Davis,

561 F.3d 159 (3d Cir. 2009) ................................. 8

Harrington v. California,

395 U.S. 250 (1969) ............................................ 16

Hemphill v. New York,

142 S. Ct. 681 (2022) ............... 1, 2, 3, 4, 19, 22, 23

Higginbotham v. State,

807 S.W.2d 732 (Tex. Crim. App. 1991) ............ 11

iv

Kotteakos v. United States,

328 U.S. 750 (1946) .................................14, 15, 17

Mitchell v. State,

120 P.3d 1196 (Okla. Ct. Crim. App. 2005)..21, 22

Neder v. United States,

527 U.S. 1 (1999) ...........................................15, 16

People v. Hardy,

824 N.E.2d 953 (N.Y. 2005) ............................... 13

People v. Reid,

971 N.E.2d 353 (N.Y. 2012) ................................. 3

Ramos v. Louisiana,

140 S. Ct. 1390 (2020) ........................................ 20

Reiner v. Woods,

955 F.3d 549 (6th Cir. 2020) .......................7, 9, 13

Ring v. Arizona,

536 U.S. 584 (2002) ............................................ 22

Schneble v. Florida,

405 U.S. 427 (1972) ............................................ 16

State v. Lui,

315 P.3d 493 (Wash. 2014) ................................ 12

State v. Mercier,

479 P.3d 967 (Mont. 2021) ................................. 11

State v. Peterson,

652 S.E.2d 216 (N.C. 2007) ................................ 12

State v. Shifflett,

508 A.2d 748 (Conn. 1986) ................................. 12

State v. Tollardo,

275 P.3d 110 (N.M. 2011) .............................11, 12

v

State v. Wall,

910 A.2d 1253 (N.H. 2006)................................. 12

Sullivan v. Louisiana,

508 U.S. 275 (1993) ...........................14, 15, 16, 20

United States v. Bailey,

696 F.3d 794 (9th Cir. 2012) ................................ 7

United States v. Baptiste,

935 F.3d 1304 (11th Cir. 2019) .......................... 11

United States v. Carrasco,

540 F.3d 43 (1st Cir. 2008) .......................9, 12, 13

United States v. Caruto,

532 F.3d 822 (9th Cir. 2008) ............................ 7, 8

United States v. Cudlitz,

72 F.3d 992 (1st Cir. 1996) .................................. 7

United States v. Cunningham,

145 F.3d 1385 (D.C. Cir. 1998) ........................ 7, 9

United States v. Davila,

569 U.S. 597 (2013) ............................................ 16

United States v. Elliott,

89 F.3d 1360 (8th Cir. 1996) .............................. 10

United States v. Erickson,

610 F.3d 1049 (8th Cir. 2010) ............................ 10

United States v. Garcia-Lagunas,

835 F.3d 479 (4th Cir. 2016) .............................. 10

United States v. Hasting,

461 U.S. 499 (1983) .......................................15, 16

United States v. Haymond,

139 S. Ct. 2369 (2019) ........................................ 20

vi

United States v. Henry,

852 F.3d 1204 (10th Cir. 2017) .......................... 21

United States v. Ibisevic,

675 F.3d 342 (4th Cir. 2012) ................................ 7

United States v. Jean-Baptiste,

166 F.3d 102 (2d Cir. 1999) ............................8, 12

United States v. Johnson,

617 F.3d 286 (4th Cir. 2010) ................................ 7

United States v. Makkar,

810 F.3d 1139 (10th Cir. 2015) .....................7, 8, 9

United States v. Morris,

988 F.2d 1335 (4th Cir. 1993) .............................. 9

United States v. Nash,

482 F.3d 1209 (10th Cir. 2007) .......................... 10

United States v. Ofray-Campos,

534 F.3d 1 (1st Cir. 2008) .................................... 7

United States v. Pon,

963 F.3d 1207 (11th Cir. 2020) .......................... 10

United States v. Ramos-Rodriguez,

809 F.3d 817 (5th Cir. 2016) .............................. 10

United States v. Skilling,

638 F.3d 480 (5th Cir. 2011) .............................. 10

United States v. Staggers,

961 F.3d 745 (5th Cir. 2020) ...........................9, 10

United States v. Stewart,

907 F.3d 677 (2d Cir. 2018) ................................. 8

United States v. Taylor,

813 F.3d 1139 (8th Cir. 2016) ............................ 10

vii

Vasquez v. United States,

565 U.S. 1057 (2011) ............................................ 7

Vasquez v. United States,

566 U.S. 376 (2012) .............................................. 7

Ventura v. State,

29 So. 3d 1086 (Fla. 2010) ................................ 11

Virgin Islands v. Martinez,

620 F.3d 321 (3d Cir. 2010) ................................. 7

Weiler v. United States,

323 U.S. 606 (1945) ............................................ 19

Wooden v. United States,

142 S. Ct. 1063 (2022) ........................................ 14

Wray v. Johnson,

202 F.3d 515 (2d Cir. 2000) ............................. 7, 8

Other Authorities

Alice M. Batchelder, Some Brief

Reflections of a Circuit Judge, 54

OHIO ST. L.J. 1453 (1993) .................................. 21

Stephanos Bibas, The Psychology of

Hindsight and After-the-Fact

Review of Ineffective Assistance of

Counsel, 2004 UTAH L. REV. 1 ........................... 21

Harry T. Edwards, To Err Is Human,

But Not Always Harmless: When

Should Legal Error Be Tolerated?,

70 N.Y.U. L. REV. 1167 (1995) ........................... 21

Daniel Epps, The Right Approach to

Harmless Error, 120 COL. L. REV. F.

1 (2020) ............................................................... 22

viii

William M. Landes, et al., Harmless

Error, 30 J. LEGAL STUD.161 (2001) .................. 19

Gregory Mitchell, Against “Overwhelming” Appellate Activism:

Constraining Harmless Error

Review, 82 CAL. L. REV. 1335 (1994) ................. 20

Anne Bowen Poulin, Tests for Harm in

Criminal Cases: A Fix for Blurred

Lines, 17 U. PA. J. CONST. L. 991

(2015) .................................................................. 14

Roger Traynor, The Riddle of Harmless

Error (1970) ........................................................ 20

INTEREST OF AMICUS CURIAE*

The National Association of Criminal Defense

Lawyers (NACDL) is a nonprofit, voluntary professional bar association that works on behalf of criminal

defense attorneys to ensure justice and due process

for those accused of criminal misconduct. Founded in

1958, NACDL has a nationwide membership of many

thousand direct members. NACDL is the only nationwide professional bar association for both public defenders and private criminal-defense lawyers, and its

members include not only lawyers serving in those

roles, but also private criminal-defense lawyers, military defense counsel, law professors, and judges. Consistent with NACDL’s mission of advancing the

proper, efficient, and fair administration of justice,

NACDL files numerous amicus briefs each year in

this Court and other federal and state courts—all to

the end of providing assistance in cases that present

issues of broad importance to criminal defendants,

criminal defense lawyers, and the criminal justice system as a whole.

STATEMENT

The Court will recall the facts of this case from its

previous decision in Hemphill v. New York, 142 S. Ct.

681 (2022), where the Court reversed the defendant’s

criminal conviction on Confrontation Clause grounds

and remanded the case for harmless error analysis.

In brief, a 2006 brawl on a Bronx street turned deadly.

No counsel for any party authored this brief in whole

or in part, and no person other than the amicus, its members, or its counsel financially contributed to its preparation or submission. Pursuant to Rule 37.2, amicus gave

notice of its intent to file the brief.

*

2

Somebody fired a 9-millimeter pistol, killing a child in

a nearby car. The central question at trial was who

fired the gun. Id. at 686-687.

The police searched the home of a suspect, Nicholas Morris, and found 9mm .357-caliber ammunition.

Ibid. When the police arrested Morris, they observed

bruising on his knuckles consistent with fist fighting.

Ibid. Meanwhile, Morris’s friend, Ronnell Gilliam,

who took part in the altercation, surrendered to police,

naming Morris as the gunman. Ibid. Moreover, three

other eyewitnesses identified Morris as the shooter

out of a lineup. Ibid. Yet Gilliam soon changed his

story and named his cousin, Darrell Hemphill, as the

shooter. Ibid. Choosing not to credit Gilliam’s volteface, the State charged Morris with the murder. Ibid.

After the trial began, Morris and the State reached a

plea bargain: Morris pled guilty to possessing a .357,

and the State dropped the murder charge. Ibid.

Five years passed, at which point the State discovered that Hemphill’s DNA matched DNA taken from

a blue sweater that police had found in Gilliam’s

apartment shortly after the crime. Ibid. Witnesses

had described the shooter as wearing a blue shirt or

sweater. Ibid. Yet two more years went by before, in

2013, Hemphill was arrested and indicted for the

murder. Ibid.

At trial, Hemphill contended that the State was

right the first time: Morris was the real shooter. Ibid.

After the jury heard evidence that police recovered

both 9mm and .357 bullets from Morris’s home, and

that eyewitnesses had identified Morris as the shooter,

the State sought to introduce evidence that Morris

had ultimately pled guilty to possessing a .357. Morris was not available to take the stand, so in lieu of his

3

live testimony, the State proffered the transcript from

his plea allocution. Hemphill objected. He argued

that the transcript was hearsay and that he was being

denied the opportunity to cross-examine Morris, in violation of Crawford v. Washington, 541 U.S. 36 (2004).

See 142 S. Ct. at 687-688.

Citing People v. Reid, 971 N.E.2d 353 (N.Y.

2012)—a New York Court of Appeals case holding

that a criminal defendant can “open the door” to evidence otherwise in violation of the Confrontation

Clause if the evidence is “reasonably necessary to correct a misleading impression” made by the defense (id.

at 357 (cleaned up))—the court admitted the transcript.

As the trial court explained, although

Hemphill’s third-party culpability defense was a “necessary” argument under the circumstances, it nevertheless opened to door to unconfronted testimony.

Ibid.

At closing arguments, Hemphill reiterated his theory that Morris was the shooter. The prosecutor then

used the fact that Morris had pled guilty to the possession charge as affirmative evidence that he did not

commit the murder, stating:

And in the end, against the defense attorney’s

advice, Morris admits to possessing the 357 that

day.

Do you know why?

Because that’s the crime he actually committed

on April 16, 2006. * * * He didn’t have to plead

guilty. He took responsibility for the crime he committed.

J.A. 356. The jury found Hemphill guilty, and the

court sentenced him to 25 years to life in prison. Ibid.

4

Hemphill appealed. Both the New York Supreme

Court, Appellate Division and the New York Court of

Appeals affirmed. Hemphill then successfully sought

certiorari from this Court, which reversed on Confrontation Clause grounds. The Court accepted the trial

court’s determination that Morris’s allocution was

“reasonably necessary” to “rebut [petitioner]’s theory

that Morris committed the murder.” 142 S. Ct. at 686

(citation omitted). But regardless of how “misleading”

of an impression the defendant created or how “necessary” the unconfronted evidence was to rebut that impression, the Court held, the trial court violated the

Confrontation Clause in admitting evidence against a

criminal defendant without “testing in the crucible of

cross-examination.” Id. at 692-693 (quoting Crawford,

541 U.S. at 61).

Rather than address the State’s harmless error argument, the Court remanded, in keeping with its

“general custom of allowing state courts initially to assess the effect of erroneously admitted evidence.” Id.

at 694 n.5. On remand, the New York Court of Appeals held the Crawford error harmless. Pet. App. 1a.

As the court put it: because the “evidence of

[Hemphill]’s guilt was overwhelming,” “there is no

reasonable possibility that the erroneously admitted

plea allocution might have contributed to defendant's

conviction.” Pet. App. 4a (quotation omitted).

Hemphill now returns to this Court to protect his

initial victory and remedy the violation of his Sixth

Amendment right of confrontation. The Court should

grant Hemphill’s petition and reverse.

5

SUMMARY OF ARGUMENT

The state and lower federal courts are hopelessly

divided over how to conduct harmless error review.

The court below, and many others, focus on whether

the evidence—minus whatever portion was wrongly

admitted—supports the conviction beyond a reasonable doubt, usually by finding that evidence “overwhelming” (and thus that the error was harmless). A

larger number of courts focus on the error, asking

whether it likely affected the verdict. This conflict is

entrenched and calls out for this Court’s resolution.

Amicus submits that the “overwhelming evidence”

approach to harmless error analysis usurps the function of the jury, effectively making the appellate court

—which sees only a cold record—the ultimate arbiter

of guilt. Asking whether the error likely affected the

verdict, by contrast, ensures that the appellate role is

properly limited to reviewing the trial error and remedying the error whenever it might have led the jury

to a different outcome. As the Court held in Chapman

v. California, 386 U.S. 18, 22 (1967), the purpose of

the harmless error doctrine is not to sideline the jury,

but to avoid “setting aside convictions for small errors

or defects that have little, if any, likelihood of having

changed the result of the trial.”

The decision below falls on the wrong side of this

divide. It fails to understand the process for assessing

the harmfulness of Confrontation Clause violations

that this Court laid out in Delaware v. Van Arsdall,

475 U.S. 673 (1986). On finding the evidence against

Hemphill “overwhelming”—a bizarre conclusion on

these facts, but not the issue on certiorari—the court

gave only perfunctory treatment to the factors that

6

other courts consider in assessing whether the constitutional error had an effect on the verdict. That error

is particularly egregious in the context of the Confrontation Clause, which as this Court reaffirmed in this

very case is not a matter for balancing, but a fundamental constitutional rule that must be obeyed. It

would be paradoxical for this Court to give the Clause

such an uncompromising reading in Hemphill I, only

to allow an appellate court to excuse the error in

Hemphill II, on the flimsy ground that the court

thought the evidence strong enough anyway. Such a

rule would incentivize prosecutors to risk Confrontation Clause violations secure in the knowledge that,

even if held unconstitutional, they will not result in

reversal of the conviction.

This case is a compelling vehicle for addressing

harmless error review, which affects a host of cases

annually. The Court is already familiar with the facts;

the decision on remand perfectly tees up the most

pressing questions about harmless-error review; and

the logical integrity of this Court’s prior opinion depends on correcting the decision below. Certiorari

should be granted.

REASONS FOR GRANTING THE PETITION

The decision below both deepens an entrenched

split over how appellate courts should conduct harmless-error review and conflicts with this Court’s precedents. Moreover, this case provides an outstanding

opportunity to clarify that courts conducting such review—which disposes of more criminal cases than any

other doctrine—must consider not only whether the

error tainted the government’s “overwhelming evidence of guilt,” but also its likely effect on how the jury

viewed the defense.

7

I. Lower courts are intractably divided over

how to conduct harmless-error review.

In 2011, this Court granted certiorari in Vasquez

v. United States, 565 U.S. 1057 (2011), to address a

circuit split over the correct harmless-error test.

Vasquez was dismissed as improvidently granted, 566

U.S. 376 (2012), but the split has only deepened.

A. The federal circuits are hopelessly divided over how to conduct harmless-error

review.

1. Eight circuits apply an effect-on-the-verdict test

that considers not simply “the strength of the prosecution’s case,” but whether the error “had a substantial and injurious effect.” Wray v. Johnson, 202 F.3d

515, 526 (2d Cir. 2000); see United States v. Cudlitz,

72 F.3d 992, 999 (1st Cir. 1996); Virgin Islands v.

Martinez, 620 F.3d 321, 338 (3d Cir. 2010); United

States v. Ibisevic, 675 F.3d 342, 350 (4th Cir. 2012);

Reiner v. Woods, 955 F.3d 549, 557 (6th Cir. 2020);

United States v. Caruto, 532 F.3d 822, 831 (9th Cir.

2008); United States v. Makkar, 810 F.3d 1139, 1148

(10th Cir. 2015); United States v. Cunningham, 145

F.3d 1385, 1394 (D.C. Cir. 1998).

Under this approach, courts assess “the closeness

of the case,” yet “the strength of the Government’s

case does not, in itself, resolve the ‘closeness’ question.”

Ibisevic, 675 F.3d at 354. Courts ask whether the case

was “close” in light of the “government’s case” and the

“evidence supporting [the] defense.” United States v.

Bailey, 696 F.3d 794, 805 (9th Cir. 2012). This entails

considering the defendant’s “testimony on his own behalf.” United States v. Ofray-Campos, 534 F.3d 1, 28

(1st Cir. 2008). In United States v. Johnson, 617 F.3d

286, 298 (4th Cir. 2010), for example, the court held

8

an error not harmless after assessing both “the government’s case” and “the fact that [the defendant] testified in his own defense as well as called numerous

witnesses to support his innocence.”

The circuits that take this approach recognize that

an error was likely harmful if it “‘bears on an issue

that is plainly critical to the jury’s decision,’ such as

the defendant’s credibility” if “central to her defense.”

United States v. Jean-Baptiste, 166 F.3d 102, 108 (2d

Cir. 1999) (citation omitted); see also Caruto, 532 F.3d

at 832 (“[A] constitutional error that goes directly to

the defendant’s credibility usually is not harmless

where the defendant’s theory of the case is plausible,

even if it is not particularly compelling.”). Likewise,

an error was likely harmful if “material to the establishment of [a] critical fact,” “emphasized in arguments to the jury,” or not “corroborated and cumulative.” Wray, 202 F.3d at 526.

Indeed, many circuits examine the error’s harmfulness in light of key disputed issues. As then-Judge

Gorsuch wrote for the Tenth Circuit, “[w]hen an error

deprives a defendant of ‘important evidence relevant

to a sharply controverted question going to the heart

of [the] defense, * * * substantial rights [are] affected.’”

Makkar, 810 F.3d at 1148 (citation omitted). Errors

are also important if they “directly undermin[e] the

plausibility of [the] defense.” Gov’t of Virgin Islands

v. Davis, 561 F.3d 159, 167 (3d Cir. 2009). In assessing the error’s harmfulness, whether “overwhelming” evidence establishes “several elements” of an offense matters little when “the central point of contention” is disputed. United States v. Stewart, 907 F.3d

677, 691 (2d Cir. 2018); see id. at 688 (considering “the

9

presence or absence of evidence corroborating or contradicting the government’s case on the factual questions at issue”).

These circuits likewise consider other telltale signs

that the error may have affected the verdict, including

whether “[t]he district court recognized” the issue’s

importance (United States v. Morris, 988 F.2d 1335,

1341–1342 (4th Cir. 1993)), whether the prosecution

acknowledged the evidence’s significance (Makkar,

810 F.3d at 1148), or whether the evidence was emphasized “during closing argument” or the “jury’s conduct” (Reiner, 955 F.3d at 557).

At bottom, these circuits “‘demand[] a panoramic,

case-specific inquiry considering, among other things,

the centrality of the tainted material, its uniqueness,

its prejudicial impact, the uses to which it was put

during the trial, the relative strengths of the parties’

cases, and any telltales that furnish clues to the likelihood that the error affected the factfinder’s resolution of a material issue.’” United States v. Carrasco,

540 F.3d 43, 55 (1st Cir. 2008) (citation omitted). As

the D.C. Circuit has explained, any other approach is

“‘inconsistent with’” defendants’ “‘right to have juries,

not appellate courts, render judgments of guilt or innocence.’” Cunningham, 145 F.3d at 1394 (citation

omitted).

2. By contrast, other courts conducting harmlesserror analysis rely on the strength of the prosecution’s

case to the exclusion of other important factors. While

courts in every circuit have considered the strength of

the government’s case in finding errors harmless, the

Fifth, Eighth, and Eleventh Circuits, among others,

frequently rely solely on “overwhelming evidence” of

guilt. E.g., United States v. Staggers, 961 F.3d 745,

10

760 (5th Cir. 2020) (finding “error was harmless, because there was overwhelming evidence of [defendant’s] guilt”); United States v. Ramos-Rodriguez, 809

F.3d 817, 824 (5th Cir. 2016); United States v. Erickson, 610 F.3d 1049, 1054 (8th Cir. 2010); see also, e.g.,

United States v. Garcia-Lagunas, 835 F.3d 479, 501–

506 & n.5 (4th Cir. 2016) (Davis, J., dissenting) (criticizing majority for “fundamental error” of relying on

“‘overwhelming evidence of guilt alone’” (citation

omitted)); United States v. Nash, 482 F.3d 1209, 1222

(10th Cir. 2007) (McKay, J., dissenting) (criticizing

majority for relying on “its laundry list of properly admitted evidence—much of it contested[—]to establish

Defendant’s guilt,” and for ignoring “the most significant part of the standard,” the error’s likely “prejudicial effect * * * upon the jury”).

In the Eighth Circuit, for example, “[e]vidence erroneously admitted in violation of the Confrontation

Clause is harmless beyond a reasonable doubt as long

as the remaining evidence is overwhelming.” United

States v. Taylor, 813 F.3d 1139, 1149–1150 (8th Cir.

2016). These circuits have also relied on overwhelming evidence in “declin[ing],” for example, “to give * * *

any weight” to a defendant’s testimony deemed “selfserving.” United States v. Skilling, 638 F.3d 480, 484

(5th Cir. 2011); see United States v. Elliott, 89 F.3d

1360, 1369 (8th Cir. 1996) (relying solely on “the

weight of the government’s massive case”).

Similarly, Eleventh Circuit precedent “is thick

with decisions” finding errors harmless based on

“overwhelming evidence.” United States v. Pon, 963

F.3d 1207, 1240 (11th Cir. 2020) (collecting cases). In

one telling case, that court relied on its “long line of

precedent” holding that “overwhelming evidence of

guilt suffices to demonstrate” harmless error in

11

brushing off an error admitting evidence that a “key

[defense] witness[]” had accepted a bribe from the defendant to give perjured testimony. United States v.

Baptiste, 935 F.3d 1304, 1308, 1314 (11th Cir. 2019).

In so holding, the court considered neither the prejudice of undermining a “key [defense] witness” nor the

inference of guilt that a jury might draw from evidence that a defendant suborned perjury. Ibid.

B. The state appellate courts are equally divided over how to conduct harmless-error

analysis.

1. State courts too are divided. Many state high

courts expressly reject reliance on overwhelming evidence of guilt at the expense of other factors informing

the verdict’s reliability. In Florida, for example,

“‘[o]verwhelming evidence of guilt does not negate the

fact that an error that constituted a substantial part

of the prosecution’s case may have played a substantial part in the jury’s deliberation and thus contributed to the actual verdict.’” Ventura v. State, 29 So.

3d 1086, 1089 (Fla. 2010) (per curiam). The Texas

Court of Criminal Appeals likewise “reject[s]” the

“‘overwhelming evidence of guilt’ test,” directing

courts to consider “the source of the error, the nature

of the error, whether or to what extent it was emphasized by the State,” “its probable collateral implications,” “how much weight a juror would probably place

upon the error,” and “whether declaring the error

harmless would encourage the State to repeat it with

impunity.” Higginbotham v. State, 807 S.W.2d 732,

734–735 (Tex. Crim. App. 1991) (citation and quotation omitted). Montana and New Mexico’s high courts

agree. See State v. Mercier, 479 P.3d 967, 977 (Mont.

2021) (“‘overwhelming evidence’ absent the tainted

evidence in favor of guilt will not alone suffice”); State

12

v. Tollardo, 275 P.3d 110, 122 (N.M. 2011) (“improper

for ‘overwhelming evidence” of guilt “to serve as the

main determinant of whether an error was harmless”).

2. Other state courts, in contrast, rely on the presence of overwhelming evidence. Under Washington’s

“‘overwhelming untainted evidence’ test,” courts “look

to the untainted evidence to determine if it was so

overwhelming that it necessarily leads to a finding of

guilt.” State v. Lui, 315 P.3d 493, 511 (Wash. 2014).

In New Hampshire, if “properly admitted evidence” is

“of overwhelming weight,” the government has “met

its burden to prove” constitutional errors harmless.

State v. Wall, 910 A.2d 1253, 1262 (N.H. 2006). Likewise, in North Carolina, error is harmless if “the independent non-tainted evidence is ‘overwhelming.’”

State v. Peterson, 652 S.E.2d 216, 222 (N.C. 2007).

Colorado and Connecticut law are similar. Bartley v.

People, 817 P.2d 1029, 1034 (Colo. 1991); State v.

Shifflett, 508 A.2d 748, 752 (Conn. 1986).

3. As exemplified by the decision below, the New

York courts may be the most confused and inconsistent of all. While aspects of its analysis feint in the

direction of an effect-on-the-verdict test, the Court of

Appeals emphasized the State’s overwhelming evidence to the exclusion of any other serious consideration. The court recognized that the question of who

fired the 9-millimeter was the “primary disputed issue” at trial (Pet. App. 2a; cf. Jean-Baptiste, 166 F.3d

at 108). It considered whether Morris’s testimony

that he possessed a .357 was cumulative of Gilliam’s

statement of the same fact1 (cf. Carrasco, 540 F.3d at

1 Conspicuously

failing to note that Gilliam had initially testified to the opposite.

13

55) and took note of the degree to which the prosecutor relied on the erroneously admitted evidence (cf.

Reiner, 955 F.3d at 557). Pet. App. 4a. But despite

acknowledging the centrality of the issue, it concluded

that the erroneous admission of testimony bearing on

that same issue was harmless, based primarily on

“other, overwhelming evidence of [Hemphill’s] guilt.”

Pet. App. 4a.

Rationally speaking, the court below may be right

that evidence that Morris pled guilty to a lesser offense should have had little bearing on whether he

had in fact pulled the trigger. See Pet. App. 4a (“[t]he

plea allocution neither exculpated Morris nor inculpated [Hemphill].”) But the prosecutor did not think

so, and in light of his misleading use of that evidence

to prove that Morris’ actual crime was mere possession, it cannot be presumed that the jury saw the flaw

in the prosecutor’s argument. That is why this is a

textbook case of why the two approaches to harmless

error can make a difference. Focusing solely on the

unconstitutionally admitted evidence, the admission

of Morris’s allocution does not really count for much.

But when one considers how the prosecution used that

evidence, and its probable effect on the jury, the

wrongly admitted hearsay might well have been decisive.

In other cases, by contrast, the New York Court of

Appeal has cautioned that “‘however overwhelming

may be the quantum and nature of the other proof,

the error is not harmless * * * if ‘there is a reasonable

possibility that the * * * [error] might have contributed to the conviction.’’” People v. Hardy, 824 N.E.2d

953, 957–958 (N.Y. 2005) (citations and quotations

omitted). One simply does not know, from one case to

the next, how the New York courts will act.

14

That confusion directly affects the fair administration of justice. “When courts are confused, the defendant may be deprived of protection.” Anne Bowen Poulin, Tests for Harm in Criminal Cases: A Fix for

Blurred Lines, 17 U. PA. J. CONST. L. 991, 1016 (2015).

Where criminal defendants’ “liberties hang in the balance,” this Court should not allow “confusion in the

lower courts and conflicting results” to prevail.

Wooden v. United States, 142 S. Ct. 1063, 1080 (2022)

(Gorsusch, J., concurring in the judgment).

II. The decision below conflicts with this

Court’s precedents.

As the decision below illustrates, state and lower

federal courts have drifted dramatically from this

Court’s harmless-error precedents, which strike a balance between the defendant’s right to a jury trial and

concerns about reversing judgments based on “hypertechnicalit[ies].” Kotteakos v. United States, 328 U.S.

750, 759 n.14 (1946). This Court strikes that balance

by asking whether the error potentially “contribute[d]

to the verdict obtained.” Chapman, 386 U.S. at 24.

Errors that “make [the defendant’s] version of the evidence worthless, can no more be considered harmless

than” a “coerced confession.” Id. at 25–26.

A proper analysis considers all relevant factors—

not just the perceived strength of the government’s

case—giving deference to the jury’s role. And where

appellate courts fail to consider whether errors might

have undermined the jury’s belief in the defense’s case,

they are reduced to “appellate speculation about a hypothetical jury’s action.” Sullivan v. Louisiana, 508

U.S. 275, 280 (1993). “The Sixth Amendment requires

more.” Ibid.

15

A

The “most important element” of the jury trial

right is “the right to have the jury, rather than the

judge, reach the requisite finding of ‘guilty.’” Id. at

277. Thus, proper harmless-error analysis must ask

“what effect [the error] had upon the guilty verdict.”

Id. at 279. Such harmless-error review targets “the

basis on which ‘the jury actually rested its verdict.’”

Ibid. (citation omitted).

“The inquiry,” therefore, “is not whether, in a trial

that occurred without the error, a guilty verdict would

surely have been rendered, but whether the guilty

verdict actually rendered in this trial was surely unattributable to the error.” Ibid. Rather than sitting

as “a second jury,” courts must “ask[] whether the record contains evidence that could rationally lead to a

contrary finding.” Neder v. United States, 527 U.S. 1,

19 (1999). “The inquiry cannot be merely whether

there was enough to support the result, apart from the

phase affected by the error.” Kotteakos, 328 U.S. at

765.

Time and again, this Court and individual Justices

have rejected “overemphasis” on “overwhelming evidence.” Chapman, 386 U.S. at 23. As the Court explained in Kotteakos, a reviewing court’s assessment

of the defendant’s “guilt or innocence” cannot be the

“sole criteria for reversal or affirmance”—“[t]hose

judgments are exclusively for the jury.” 328 U.S. at

763. The question is whether “the error complained

of did not contribute to the verdict obtained.” Chapman, 386 U.S. at 24. Simply put, a court “should not

find harmless error merely because it believes that

the other evidence is ‘overwhelming.’” United States

16

v. Hasting, 461 U.S. 499, 516 (1983) (Stevens, J., concurring in the judgment). But a host of lower courts

have not gotten the memo.

The requirement that courts consider more than

the strength of the government’s case stems from the

very nature of harmless-error review itself. Those

trial errors susceptible to review for harmlessness differ from structural errors precisely because they can

“be quantitatively assessed in the context of other evidence presented.” Sullivan, 508 U.S. at 281 (emphasis

added). Reliance solely on the government’s case does

not account for the full context of all the other evidence presented. Cf. United States v. Davila, 569 U.S.

597, 612 (2013) (harmless- and plain-error analysis

must be conducted in the “light of the full record”).

When this Court has relied on “overwhelming evidence” to find an error harmless, therefore, it has typically done so based on “uncontested” evidence. Neder,

527 U.S. at 17. In Hasting, it was deemed harmless

to permit a prejudicial prosecutorial remark where

the government’s case was “unanswered.” 461 U.S. at

512. In Harrington v. California, 395 U.S. 250 (1969),

the trial court committed a Bruton violation by admitting the testimony of co-defendants who placed themselves and the defendant at the scene of the crime.

This was held harmless when the defendant admitted

he was there. Id. at 253-254. Similarly, in Schneble

v. Florida, 405 U.S. 427 (1972), the erroneously admitted testimony of a co-defendant placing the defendant at the scene of the murder was harmless

when the defendant had confessed to the murder. Id.

at 429-430. And even when the Court has relied on

“‘over-whelming evidence’ of guilt,” it has admonished

“against giving [it] too much emphasis” (Harrington,

395 U.S. at 254 (citing Chapman, 386 U.S. at 23))—

17

and it has reversed harmless-error findings based on

supposedly overwhelming evidence. Arizona v. Fulminante, 499 U.S. 279, 297 (1991); Chapman, 386 U.S.

at 25.

The evidence here is anything but uncontested.

All of the evidence that the government marshaled in

support of its original prosecution of Morris counts as

evidence against its later prosecution of Hemphill for

the same murder. Gilliam’s testimony that Hemphill

pulled the trigger is contradicted by Gilliam’s earlier

testimony that Morris did so. The discovery of

Hemphill’s DNA on a blue sweater in Gilliam’s possession could have multiple innocent explanations. In

short, this is not a case in which uncontested evidence

led to an ineluctable conclusion of guilt.

Because courts may not simply focus on the prosecution’s “overwhelming evidence,” proper harmlesserror analysis requires considering “a host of factors.”

Van Arsdall, 475 U.S. at 684. In analyzing Confrontation Clause violations, “[t]he correct inquiry is

whether, assuming that the damaging potential of the

cross-examination were fully realized, a reviewing

court might nonetheless say that the error was harmless beyond a reasonable doubt.” Ibid. In undertaking that inquiry, the Court considers not only “the

overall strength of the prosecution’s case,” but also

“the importance of the witness’ testimony in the prosecution’s case, whether the testimony was cumulative,

the presence or absence of evidence corroborating or

contradicting the testimony of the witness on material

points, [and] the extent of cross-examination otherwise permitted.” Ibid. “The crucial thing is the impact of the thing done wrong on the [jurors’] minds” in

“the total setting.” Kotteakos, 328 U.S. at 764. This

18

requires “consider[ing] all the ways that error can infect the [trial].” Brecht v. Abrahamson, 507 U.S. 619,

642 (1993) (Stevens, J., concurring).

B

No such analysis occurred below. The New York

Court of Appeals began and ended its opinion with references to the “overwhelming” evidence of Hemphill’s

guilt. Pet. App. 1a, 4a. The court’s approach is reflected in its own synopsis: “For the following reasons,

we hold that the evidence of defendant’s guilt was

overwhelming and that the error was harmless.” Pet.

App. 1a. By the court’s own account, its focus was on

whether Hemphill was guilty (the business of the jury)

rather than on whether the constitutional error had

an impact on the verdict.

Although the Court of Appeals did recite some Van

Arsdall factors, its treatment of them was perfunctory.

Pet. App. 4a. In particular, the court did not explore

the possibility—in amicus’s view, the likelihood—that

an unsophisticated jury may have viewed the revelation that Morris had been convicted only of possession

as a “gotcha” moment, discrediting Hemphill’s selfserving attempt to turn Morris into the gunman.

Coming as it did toward the end of the defense’s case,

and amplified by the prosecutor’s arguments, it is difficult to fathom the conclusion that introducing this

hearsay was a “small error[] or defect[] that ha[d] little, if any, likelihood of having changed the result of

the trial.” Chapman, 386 U.S. at 22. Nor did the court

pay attention to the trial judge’s observation regarding the centrality of this issue to the defense case. See

19

J.A. 1852 (calling the defense strategy of pinning the

murder on Morris “necessary” to the defense).

To be sure, from the abstract perch of the appellate

bench, it may seem obvious that a plea bargain for a

lesser charge in no way exonerated Morris, and thus

should have had little impact on the balance of the evidence. But that is precisely not the proper inquiry.

The question should not have been whether the appellate judges remained persuaded of Hemphill’s guilt,

but whether the Confrontation Clause error may have

contributed to the jury’s conclusion that Hemphill was

guilty.

The New York Court of Appeals’ “overwhelming

evidence” test thus caused it to neglect the “host of

factors” informing whether the error “‘contribute[d] to

the verdict.’” Van Arsdall, 475 U.S. at 684, 680 (citation omitted). Reviewing courts may not “look at the

printed record, resolve conflicting evidence, and reach

the conclusion that the error was harmless because

[they] think the defendant was guilty. * * * [J]uries

alone” bear “that responsibility.” Weiler v. United

States, 323 U.S. 606, 611 (1945).

III. Proper harmless-error analysis is critical to

protecting the rights of criminal defendants

in a large number of cases.

As “probably the most cited rule in modern criminal appeals,” the harmless-error standard is critical.

William M. Landes, et al., Harmless Error, 30 J. LEGAL STUD.161, 161 (2001).

2 Citations

to the J.A. are to the Joint Appendix in

Hemphill v. New York, 142 S. Ct. 681 (2022) (No. 20-637).

20

Relying on appellate judges’ view of supposedly

overwhelming evidence risks usurping the jury’s role

and the defendant’s jury-trial right. Judges “may not

direct a verdict for the State, no matter how overwhelming the evidence.” Sullivan, 508 U.S. at 277.

As this Court has observed, the “promise” that “[o]nly

a jury, acting on proof beyond a reasonable doubt, may

take a person's liberty * * * stands as one of the Constitution's most vital protections against arbitrary

government.” United States v. Haymond, 139 S. Ct.

2369, 2373 (2019).

The jury trial right is thus “a fundamental reservation of power in our constitutional structure”—one

“meant to ensure [juries’] control in the judiciary.”

Blakely v. Washington, 542 U.S. 296, 305–306 (2004);

see also Ramos v. Louisiana, 140 S. Ct. 1390, 1397

(2020). But over-reliance on the government’s case

usurps juries’ role by inviting appellate courts to

make “independent conclusion[s] of guilt” (Gregory

Mitchell, Against “Over-whelming” Appellate Activism: Constraining Harmless Error Review, 82 CAL. L.

REV. 1335, 1340 (1994)) and risks transforming the

reviewing court into “in effect a second jury to determine whether the defendant is guilty” (Roger Traynor,

The Riddle of Harmless Error 21 (1970)). As this

Court has put it, “the appellate court” must not take

on “the jury’s function of measuring the evidence

[with] legal yardsticks.” Bollenbach v. United States,

326 U.S. 607, 614 (1946).

Commentators across the spectrum agree that appellate courts are institutionally ill-equipped to weigh

trial evidence. “[M]any events of trial pass without

casting so much as a shadow upon the printed transcript,” and appellate courts “cannot watch the demeanor of witnesses, listen to the intonations of their

21

voices, or engage in any of the countless other observations that inhere in an assessment of credibility.”

Harry T. Edwards, To Err Is Human, But Not Always

Harmless: When Should Legal Error Be Tolerated?, 70

N.Y.U. L. REV. 1167, 1193 (1995). “The difference between trying a case on the district level court and

merely reading the briefs on appeal is only a little less

marked than the difference between watching Gone

With the Wind and reading the TV Guide description

of it.” Alice M. Batchelder, Some Brief Reflections of

a Circuit Judge, 54 OHIO ST. L.J. 1453, 1453 (1993).

As then-Judge Gorsuch once noted, harmless-error

doctrine is not “license for rank speculation”; “[w]hen

it comes to the loss of liberty, it is better to know on

remand than guess on appeal.” United States v. Henry,

852 F.3d 1204, 1209 n.3 (10th Cir. 2017).

Moreover, the nature of harmless-error review

means review is always based on a record that has already led to the defendant’s conviction. Poulin, supra,

at 1024. As then-Professor (now Judge) Stephanos Bibas has noted, when “looking back at [that] final result,” well-documented cognitive biases mean “courts

might regard that outcome as inevitable.” Stephanos

Bibas, The Psychology of Hindsight and After-the-Fact

Review of Ineffective Assistance of Counsel, 2004 UTAH

L. REV. 1, 3. Simply put, “confirmatory bias” makes it

“just too hard for judges to screen out the fact of conviction.” Id. at 3, 6. So if appellate courts at their best

guess on harmless-error review, at their worst they

habitually guess in the same direction.

Worse, over-reliance on the government’s evidence

incentivizes prosecutors to use questionable tactics,

knowing they can be excused as harmless. See, e.g.,

Mitchell v. State, 120 P.3d 1196, 1216 (Okla. Ct. Crim.

22

App. 2005) (prosecutor urged court to admit unconfronted hearsay because it would “only be reviewed for

harmless error anyway”). Incentivizing such misconduct undermines the “public respect for the criminal

process” that the harmless-error doctrine seeks to promote. Van Arsdall, 475 U.S. at 681.

IV. This case is an ideal vehicle for the Court to

clarify the harmless error doctrine.

Although this Court often considers whether an error may be reviewed for harmlessness, it rarely undertakes that review itself, “ordinarily leav[ing] it to

lower courts to pass on the harmlessness of error in

the first instance.” Ring v. Arizona, 536 U.S. 584, 609

n.7 (2002); Hemphill, 142 S. Ct. at 693 n.5. Here, that

first instance has occurred. The court below has found

the underlying error harmless, and the issue is ripe

for this Court’s review. Furthermore, this Court is already well-acquainted with the facts of the case.

In addition, the coherence of this Court’s prior decision is on the line. That decision rested on the premise that Morris’ plea allocution was introduced to establish Hemphill’s guilt by suggesting “that possession of a .357 revolver, not murder, was the crime

[Morris] actually committed” (id. at 688 (quotation

omitted)), thereby “rebut[ting] Hemphill’s theory that

Morris committed the murder” (id. at 686). Were that

not so, and Morris’s testimony did nothing to inculpate Hemphill, it would have been inadmissible “under ordinary principles of evidentiary relevance.”

Daniel Epps, The Right Approach to Harmless Error,

120 COL. L. REV. F. 1, 4 (2020). Indeed, this Court

specifically rejected the dissent’s view that no confrontation issue was properly before it. Hemphill, 142

23

S. Ct. at 689 n.2. Under these circumstances, the notion that the “[t]he plea allocution neither exculpated

Morris nor inculpated [Hemphill]” retroactively

leaves this Court’s decision a nullity. Pet. App. 4a.

The Court should not allow the court below to render

its prior opinion advisory.

CONCLUSION

Darrell Hemphill is serving a sentence of 25-yearsto-life on an evidentiary record so thin that the State

initially tried another man for the crime. Nevertheless, the court below thought the evidence of his guilt

was so “overwhelming” that it could excuse a violation

of his Sixth Amendment right to confront his accusers

confirmed by this Court. The absurdity of that result

is the inevitable byproduct of a harmless-error doctrine in dire need of this Court’s attention. Certiorari

should be granted.

Respectfully submitted,

MICHAEL MCCONNELL

Wilson Sonsini

Goodrich & Rosati, PC

650 Page Mill Road

Palo Alto, CA 94304

(650) 493-9300

JEFFREY T. GREEN

National Association of

Criminal Defense

Lawyers

1660 L Street NW

Washington, DC 20036

STEFFEN N. JOHNSON*

PAUL N. HAROLD

GARRETT R. ATHERTON

Wilson Sonsini

Goodrich & Rosati, PC

1700 K Street, NW

Washington, DC 20006

(202) 973-8800

sjohnson@wsgr.com

* Counsel of Record

Counsel for Amicus Curiae

JANUARY 2023