The harmless error doctrine is beset with problems, both theoretical and practical. In Harmless Error and Substantial Rights, recently published in the Harvard Law Review, Professor Daniel Epps proposes a reconceptualization of constitutional criminal procedure rights that is designed to address these problems. Epps argues that those constitu­tional criminal procedure rights that are capable of being violated by pros­ecutors and judges in nonharmful ways should be redefined to include a require­ment that their violation causes the right holder harm. In other words, what we now regard as a nonharmful violation of a con­stitutional crimi­nal procedure right would not amount to a constitutional violation at all.

This Response argues that, while harmless error doctrine should indeed be reformed, acceptance of Epps’s proposal would create more prob­lems than it would solve. Specifically, the narrower constitutional precedent that would result from implementing the proposal would cause mischief when translated into other adjudicatory and lawmaking con­texts. The Response thus defends the conventional understanding of harmless error review as a remedial inquiry. It does so by summarizing Epps’s argument, laying out concerns about certain transcontexual effects if it were to be accepted, and proposing some alternative pathways to reform.


In Harmless Errors and Substantial Rights, just out in the Harvard Law Review, Professor Daniel Epps proposes a reconceptualization of constitu­tional criminal procedure rights that would pave the way for a reform of harmless error review. 1 Daniel Epps, Harmless Errors and Substantial Rights, 131 Harv. L. Rev. 2117, 2121 (2018). Epps contends that those constitutional criminal procedure rights that are capable of being violated by prosecutors and judges in nonharmful ways should be redefined to include a requirement that their violation causes the right holder harm. 2 See id. at 2158–63. Thus, for example, an accused’s Sixth Amendment right “to be confronted with the witnesses against him” 3 U.S. Const. amend. VI. really should be understood as a right to be confronted by those witnesses whose testimony cannot be dismissed as immaterial to the jury’s later decision to convict. Under Epps’s proposal, harmless error would no longer be an amalgam of remedial doctrines informing whether reviewing courts should reverse or vacate judgments of convic­tion as a consequence of constitutional rights violations at or in connec­tion with a criminal trial. Rather, the harm (if any) caused by the putative invasion of a right would constitute a metric informing whether there has been a violation of the right.

I am a fan of doctrinal-reform scholarship 4 See, e.g., John M. Greabe, A Better Path for Constitutional Tort Law, 25 Const. Comment. 189, 190–91 (2008) (proposing a reinterpretation of 42 U.S.C. § 1983 that would facilitate a clarifying reform of constitutional tort law by doing away with the quali­fied immunity defense). and greatly enjoyed read­ing Epps’s characteristically well-written and provocative paper. I whole­heartedly agree that harmless error doctrine is in dire need of reform. Indeed, I recently wrote an article spelling out my own views of how such reform should proceed. 5 John M. Greabe, The Riddle of Harmless Error Revisited, 54 Hous. L. Rev. 59 (2016) [hereinafter Greabe, Harmless Error Revisited]. And yet, I believe it would cause more problems than it would solve to view harmless error review as part and parcel of some criminal procedure rights. Specifically, the narrower constitutional precedent that would result from implementing the proposal would cause mischief when translated into other adjudicatory and lawmaking contexts. This Response thus defends the conventional understanding of harmless error review as a remedial inquiry. Part I summarizes Epps’s argu­ment. Part II lays out concerns about certain transcontextual effects if it were to be accepted. Part III sketches some alternative pathways to reform harmless error review without narrowing the scope of constitu­tional criminal procedure rights.

I. Epps’s Argument

In Chapman v. California, 6 386 U.S. 18 (1967). the Supreme Court issued two important holdings that appellate courts throughout the country—federal and state—regularly apply. First, appellate courts may conclude that a consti­tutional error committed during the course of judicial proceedings lead­ing to a criminal conviction was harmless and therefore did not warrant reversal or vacatur. 7 Id. at 21–22. Second, the law requires reversal or vacatur if the government fails to persuade the appellate court that the error was “harm­less beyond a reasonable doubt.” 8 Id. at 24.

In Harmless Errors and Substantial Rights, Epps aptly notes that despite their practical importance and courts’ familiarity with them, the doc­trines governing harmless constitutional errors remain surprisingly mys­terious and unsatisfactory. 9 Epps, supra note 1, at 2120. Controversy abounds over fundamental ques­tions such as which constitutional errors should be subject to harmless error review, how harmless error review should proceed with respect to those errors, and what Chapman’s “beyond a reasonable doubt” rule really is—that is, what source of law justifies its use and empowers the Supreme Court to insist that both federal and state appellate courts apply it during direct review of criminal convictions. 10 See id. at 2120–21. Is it a constitutional prin­ciple derivable from constitutional criminal procedure rights or an appel­late court’s obligation to provide due process? 11 See, e.g., Stephen H. Goldberg, Harmless Error: Constitutional Sneak Thief, 71 J. Crim. L. & Criminology 421, 424 n.31 (1980) (describing Chapman’s harmless beyond a reasonable doubt principle as a “constitutional judgment”); Richard M. Re, The Due Process Exclusionary Rule, 127 Harv. L. Rev. 1885, 1912–18 (2014) (arguing for the due process view). Or is it an example of a controversial set of doctrines that Professor Henry P. Monaghan famously labeled “constitutional common law”—“a substructure of sub­stantive, procedural, and remedial rules drawing their inspiration and authority from, but not required by, various constitutional provisions”? 12 Henry P. Monaghan, The Supreme Court, 1974 Term—Foreword: Constitutional Common Law, 89 Harv. L. Rev. 1, 2–3 (1975); see also Henry P. Monaghan, Harmless Error and the Valid Rule Requirement, 1989 Sup. Ct. Rev. 195, 200 n.30 (characterizing the Chapman harmless error rule as constitutional common law). A number of other commen­tators have agreed with Monaghan’s description of harmless error doctrine as constitu­tional common law. See, e.g., Daniel J. Meltzer, Harmless Error and Constitutional Remedies, 61 U. Chi. L. Rev. 1, 26 –29 (1994); Craig Goldblatt, Comment, Harmless Error as Constitutional Common Law: Congress’s Power to Reverse Arizona v. Fulminante, 60 U. Chi. L. Rev. 985, 1009–12 (1993). But the legitimacy of constitutional common law—and in particular, the propriety of imposing it on the states—is hotly contested by many com­mentators, including (quite recently) Justice Thomas. See, e.g., Collins v. Virginia, 138 S. Ct. 1663, 1675 –80 (2018) (Thomas, J., concurring) (arguing that it is illegitimate to require the states to apply the Fourth Amendment’s exclusionary rule, which Justice Thomas characterizes as constitutional common law); Greabe, Harmless Error Revisited, supra note 5, at 116 n.308 (collecting authority critical of recognizing the legitimacy of constitutional common law). And how, if at all, does it relate to 28 U.S.C. § 2111, a federal statute the Supreme Court has largely ignored that directs reviewing courts to with­hold remedies for errors that have not affected the substantial rights of the parties? 13 The statute states: “On the hearing of any appeal or writ of certiorari in any case, the court shall give judgment after an examination of the record without regard to errors or defects which do not affect the substantial rights of the parties.” 28 U.S.C. § 2111 (2012). See also Fed. R. Crim. P. 52(a) and Fed. R. Civ. P. 61, both of which direct courts to withhold remedies for errors that do not affect the “substantial rights” of a party.

Epps has a novel idea for clearing up these mysteries. Invoking Professor Daryl J. Levinson’s warning that courts and constitutional scholars remain vigilant against mindless “rights essentialism”—that is, broadly and abstractedly conceiving of rights without regard to the often narrow ways in which they are operationalized in real-world contexts 14 See Daryl J. Levinson, Rights Essentialism and Remedial Equilibration, 99 Colum. L. Rev. 857, 858 (1999). —he proposes treating the Chapman harmless beyond a reasonable doubt principle not as a remedial doctrine (as it is conventionally understood) but rather as a metric used to define the substance of constitutional crim­inal procedure rights. 15 See Epps, supra note 1, at 2158–63. Thus, as noted in the Introduction, the Sixth Amendment confrontation right should be understood as extending only to testimony that a reviewing court cannot deem immaterial to a later decision to convict beyond a reasonable doubt. 16 See supra text accompanying note 3.

In making this argument, Epps points out that two constitutional criminal procedure rights—the right of criminal defendants to discovery of exculpatory evidence recognized in Brady v. Maryland and its progeny 17 373 U.S. 83, 87 (1963) (holding that “suppression by the prosecution of evidence favorable to an accused upon request violates due process where the evidence is material either to guilt or punishment”). and the right to effective assistance of counsel recognized in Strickland v. Washington 18 466 U.S. 668, 687 (1984) (holding that to establish constitutionally ineffective assistance of counsel “the defendant must show that the deficient performance prejudiced the defense”). —have built-in harm requirements of the type he envisions. 19 See Epps, supra note 1, at 2160 & nn.273–75. Epps proposes extending this approach to the other con­stitutional criminal procedure rights that, under current doctrine, are capable of being infringed in nonharmful ways. 20 See id.

More generally, Epps says that harmless error doctrine can and should be likened to an implementing rule such as the strict scrutiny test that informs the First Amendment right to free speech. 21 See id. at 2123 (citing Richard H. Fallon, Jr., The Supreme Court, 1996 Term—Foreword: Implementing the Constitution, 111 Harv. L. Rev. 54, 56 –57 (1997)); see also id. at 2159–64 (elaborating on this argument). Thus, in the same way that the right to free speech is really just a right to free speech in circumstances in which the government does not have a compelling interest in restricting speech through narrowly tailored means, 22 See, e.g., Reed v. Town of Gilbert, 135 S. Ct. 2218, 2226 (2015). the Sixth Amendment confrontation right (to stay with the same example) encompasses only testimony that the jury might have relied upon in deciding to convict. As Epps puts it:

When a court engages in harmless error analysis, then, it is applying a doctrinal rule that is designed to get at whether a conviction violated the defendant’s constitutional rights. Once that rights question is answered, the actual remedies question is easy: if the defendant’s conviction is unconstitutional, the appel­late court must reverse; if the conviction isn’t, the court need not. 23 Epps, supra note 1, at 2123.

Epps argues that, if courts were to reconceive of constitutional crim­inal procedure rights in this way, they could solve a number of enduring mysteries about harmless error review. First, Chapman’s beyond a reason­able doubt principle becomes a constitutional rule rather than mere (and more controversial) constitutional common law. 24 See supra notes 10–12 and accompanying text. This explains why it binds state courts as well as federal courts. 25 See Epps, supra note 1, at 2164 –65. Second, this reconceptual­ization clarifies the relationship between Chapman and 28 U.S.C. § 2111: 26 See supra note 13 and accompanying text. The statute is nothing more than a command that courts refrain from overenforcing rights by affording a remedy when no underlying constitutional violation has taken place. 27 See Epps, supra note 1, at 2164 –66.

Third, the inquiry into whether a particular putative invasion of rights is amenable to harmless error review at all becomes more straight­forward. Epps says:

If, properly understood, the right protects against conviction using a particular form of process, a conviction obtained via that disfavored method can never be harmless. If, by contrast, the right is best understood as guarding against conviction based on a particular form of disfavored evidence, or as a result of a particular procedure, then a prima facie violation can be harmless. That is, such a right is not really violated if the defendant’s conviction wasn’t based on, or caused by, the introduction of the disfavored evidence or other procedural violation. 28 Id. at 2167.

Fourth, the inappropriateness of treating harmless error review as an inquiry into whether the defendant truly is guilty instead of an inquiry into the effect of the alleged rights invasion upon the verdict—an incor­rect approach that is applied by reviewing courts with regrettable fre­quency—becomes manifest. He explains: “If the Confrontation Clause is best understood as protecting against convictions based on unconfronted testimonial hearsay, the relevant question should be—as Chapman, but not necessarily some of the later cases, implies—what role the inadmissi­ble testimony had in causing or contributing to the conviction.” 29 Id. at 2168–69.

Finally, and most importantly, Epps contends that such a reconcep­tion of constitutional criminal procedure rights would yield a more real­istic understanding of the right–remedy relationship—one that would make it harder for courts to surreptitiously undermine constitutional values. Epps concludes his paper with the following reflection:

[T]he rights-based approach has at least one significant practi­cal payoff: it would require courts to be clearer about the values at stake when adjudicating claims of harmless error. No longer could a court declare a broad scope of a constitutional right while in the same breath undercutting that right’s effective value through the use of harmless error analysis. That would be a step forward, even if only a small one. 30 Id. at 2186.

II. Transcontextual Concerns

Epps concedes that, outside of the Brady and Strickland contexts, 31 See supra notes 17–18 and accompanying text. the Supreme Court has not embraced his argument that a harm analysis should be part of determining the scope of the right. 32 See Epps, supra note 1, at 2170. Indeed, as he acknowledges, the Court has explicitly rejected the argument that a defendant’s Sixth Amendment Confrontation Clause rights are only vio­lated upon conviction for the charged crime. 33 Id. His article quotes the fol­lowing passage from Delaware v. Van Arsdall, in which the Court explains why a Sixth Amendment Confrontation Clause violation takes place upon the improper admission of evidence, and not upon conviction:

[T]he focus of the Confrontation Clause is on individual wit­nesses. Accordingly, the focus of the prejudice inquiry in determining whether the confrontation right has been violated must be on the particular witness, not on the outcome of the entire trial. It would be a contradiction in terms to conclude that a defendant denied any opportunity to cross-examine the witnesses against him nonetheless had been afforded his right to “confront[ation]” because use of that right would not have affected the jury’s verdict. We think that a criminal defendant states a violation of the Confrontation Clause by showing that he was prohibited from engaging in otherwise appropriate cross-examination designed to show a prototypical form of bias on the part of the witness . . . . 34 Id. (alterations in original) (quoting Delaware v. Van Arsdall, 475 U.S. 673, 680 (1986)).

Thus, at the threshold, Epps’s argument runs up against the fact that it has been considered and rejected by the Court. But such is the nature of doctrinal-reform scholarship. There is nothing at all wrong with argu­ing that the Court should change course.

But Van Arsdall got it right. In recent years, scholars have called attention to a host of problems that can arise when precedent resulting from constitutional innovation within one adjudicatory context is trans­lated into other contexts. 35 See, e.g., Michael Coenen, Spillover Across Remedies, 98 Minn. L. Rev. 1211, 1244 –47 (2014); Jennifer E. Laurin, Rights Translation and Remedial Disequilibration in Constitutional Criminal Procedure, 110 Colum. L. Rev. 1002, 1013–15, 1057 (2010); Jennifer E. Laurin, Trawling for Herring : Lessons in Doctrinal Borrowing and Convergence, 111 Colum. L. Rev. 670, 703–06 (2011); Leah Litman, Remedial Convergence and Collapse, 106 Calif. L. Rev. (forthcoming 2018). Epps’s argument seeks to reform the manner in which constitutional error is handled within the context of direct review of criminal convictions. But his approach would usher in a regime with a fundamental disconnect between the applicable law of constitu­tional criminal procedure rights—developed at the appellate level—and how we should want trial courts to rule on motions seeking to enforce those rights in criminal trials. Moreover, it would cause serious problems when translated into the contexts of collateral review of criminal convic­tions and congressional lawmaking under section 5 of the Fourteenth Amendment. 36 Constitutional criminal procedure rights protect against the actions and inactions of prosecutors and trial judges. Translating Epps’s proposal into the constitutional tort context would not likely cause problems under current law because prosecutors (acting in a prosecutorial capacity) and trial judges (acting in a judicial capacity) are accorded abso­lute immunity from civil rights damages actions under 42 U.S.C. § 1983 and the cause of action recognized in Bivens v. Six Unknown Named Agents of Fed. Bureau of Narcotics, 403 U.S. 388 (1971). See Imbler v. Pachtman, 424 U.S. 409, 417–20 (1976) (holding that prosecutors are absolutely immune from constitutional tort liability for actions undertaken in a prosecutorial capacity); Pierson v. Ray, 386 U.S. 547, 554 –55 (1967) (holding that judges are absolutely immune from constitutional tort liability for actions undertaken in a judicial capacity). But see infra section II.C (observing that, at least theoretically, Congress could abrogate absolute judicial and prosecutorial immunity under section 5 of the Fourteenth Amendment).

A. The Trial-Administration Problem

As noted above, Epps argues that incorporating harmless error prin­ciples into criminal procedure rights such as the Sixth Amendment right to confrontation can be analogized to how strict scrutiny is incorporated into the First Amendment right to free speech. Thus, again, in the same way that the right to free speech is really just a right to free speech when the government does not have a compelling interest in restricting speech through narrowly tailored means, the Sixth Amendment right to con­frontation is a right to confront any witness whose testimony cannot be deemed (beyond a reasonable doubt) immaterial to a later decision to convict. 37 See supra notes 21–23 and accompanying text. But this argument overlooks a crucial difference between the strict scrutiny and harmless error tests.

When the conditions of strict scrutiny are satisfied, the government’s regulation of speech is unobjectionable. The same cannot be said, how­ever, of circumstances in which unconfronted testimony that did not affect the jury’s subsequent verdict is admitted into evidence. In the First Amendment context, a court reviewing the government’s action would say that there is nothing wrong with what happened; the government simply regulated in highly unusual circumstances when it was permitted to regulate—indeed, when society should wish it to regulate. But in the Sixth Amendment context, a court reviewing the trial judge’s evidentiary ruling should disapprove of the admission of the evidence, even though reversal or vacatur of the underlying criminal judgment would not be an appropriate response to it.

Here, it is important to note that Epps agrees that the desirable path for trial judges to follow is to exclude all unconfronted evidence, even though that is not what the Sixth Amendment requires under his theory. He states:

In my theory, the relevant decision rule suggests that the intro­duction of testimonial hearsay without opportunity for confron­tation violates the defendant’s constitutional rights only if it actually causes his conviction. But it doesn’t follow that we want trial courts themselves to apply that test during the trial. In the midst of a trial, a trial court might make a mistake about the likely impact that a particular piece of evidence or testimony might have on the jury. (Moreover, what purpose is there for admitting the testimony if it isn’t going to be used to support a conviction?) Far better for a trial court to simply refuse to admit evidence that might trigger the constitutional rule in question, even if we can’t know for certain whether that admission will actually violate the Constitution until the jury renders a verdict. In other words, the conduct rule we want trial courts to follow is a blanket prohibition on the admission of testimony that might create a potential Confrontation Clause problem. 38 Epps, supra note 1, at 2171 (footnote omitted) (citing Meir Dan-Cohen, Decision Rules and Conduct Rules: On Acoustic Separation in Criminal Law, 97 Harv. L. Rev. 625, 627 (1984)). As Epps notes, Dan-Cohen distinguished “‘decision rules,’ which are ‘the laws addressed to officials,’ [from] ‘conduct rules,’ which are ‘the laws addressed to the general public.’” Id. at 2170 n.313 (quoting Dan-Cohen, supra, at 627).

But if trial courts should exclude all unconfronted evidence, why should appellate courts more narrowly define the scope of the relevant right? The definition of the relevant constitutional criminal procedure right should track what society wants the regulated government agent (here, the trial judge) to do or to refrain from doing, should it not? Or, put in terms of the distinction Epps draws between “decision” and “con­duct” rules, trial judges should conform their conduct to their under­standing of the applicable “decision rules,” as they do in other contexts of trial administration.

Indeed, by what authority would a trial judge exclude the evidence if the right were redefined to include a harm metric? Trial judges do not make discretionary, a priori decisions about what evidence a jury should consider and what evidence it should not. Rather, the parties decide what evidence to introduce. Trial courts exclude evidence only on objection of a party and only when some principle of law authorizes exclusion. But what principle of law would authorize exclusion of unconfronted but rel­evant evidence not likely to be material to the verdict if the Sixth Amendment right were so narrowed? Certainly not the Sixth Amendment. Perhaps appellate courts, exercising their supervisory authority over trial courts, could develop a constitutional common law evidentiary conduct rule requiring the exclusion of all unconfronted evi­dence in order to protect Sixth Amendment rights as a prophylactic mat­ter. But if they did so, what exactly would be accomplished by narrowing the scope of the right? For the result would be replacement of a contro­versial constitutional common law remedial doctrine—the Chapman regime as it is currently understood 39 Epps agrees that harmless error principles are best regarded as constitutional common law under the conventional understanding of harmless error as a remedial doc­trine. See id. at 2150 (“Understanding harmless constitutional error [as a remedial doc­trine] as a form of constitutional common law . . . is the most compelling [theory] on offer.”). —with a controversial constitutional common law evidentiary principle.

Epps does not really answer this important line of objection. He says that the “partial solution” that present doctrine reaches “is to be some­what disingenuous about when the real constitutional violation occurs.” 40 Id. at 2171. In other words, appellate courts today understandably act as if the Constitution requires the exclusion of all unconfronted evidence (as Van Arsdall says) even though it technically does not (per his theory). But what should courts do as we move ahead? Should they continue to be disingenuous? Presumably not. But then we are back to the problem—noted in the preceding paragraph—of a disconnect between what the law requires and how we want trial judges to act. And this is no small problem, given that there is little to no “acoustic separation” between the decision rules adopted by appellate courts and the law applied during criminal trials by trial judges. 41 See Meir Dan-Cohen, Decision Rules and Conduct Rules: On Acoustic Separation in Criminal Law, 97 Harv. L. Rev. 625, 630–34 (1984) (using the term “acoustic separation” to describe a situation in which regulated actors are aware of conduct rules but not the decision rules applied by government officials adjudicating disputes involving conduct rules). While Dan-Cohen’s concern was the relationship between judges and individuals subject to judicial decision, Epps’s concern is the relationship between appellate judges and trial judges who must be cognizant of appellate decisionmaking. Epps, supra note 1, at 2171.

Epps might point to the fact that the Supreme Court has built a harm calculation into the Brady and Strickland rights without unleashing trial-administration problems such as these. 42 See supra notes 17–19 and accompanying text. But such a response would not be persuasive. Leave aside the fact that Brady and Strickland case law has come under withering criticism for being insufficiently protective of the rights of criminal defendants. 43 See, e.g., Paul D. Butler, Poor People Lose: Gideon and the Critique of Rights, 122 Yale L.J. 2176, 2187 n.49 (2013) (criticizing Strickland ’s holding); Pamela R. Metzger, Fear of Adversariness: Using Gideon to Restrict Defendants’ Invocation of Adversary Procedures, 122 Yale L.J. 2550, 2558 (2013) (same); Justin Murray & John Greabe, Disentangling the Ethical and Constitutional Regulation of Criminal Discovery, Harv. L. Rev. Blog (June 15, 2018), [] (discussing the problems states have encountered in determining prosecutors’ ethical obligations to disclose exculpatory evi­dence caused by Brady case law’s inclusion of a harm calculus in the definition of the right). Even so, these rights materially differ from the Sixth Amendment confrontation right, which is set forth in the text of the Constitution in unqualified terms that clearly contemplate its application as a real-time conduct rule by trial judges. 44 See supra text accompanying note 3. Neither the Brady right nor the Strickland right is enumerated in the Constitution’s text, 45 The Brady right is an unenumerated due process right. See Brady v. Maryland, 373 U.S. 83, 86 (1963). The Strickland right derives from the Sixth Amendment’s promise of “the Assistance of Counsel for [an accused’s] defense,” U.S. Const. amend. VI, but the Constitution’s text makes no mention of “effective” assistance of counsel. See Strickland v. Washington, 466 U.S. 668, 685–86 (1984). Rather, the Supreme Court has located the right to effectiveness in the due process guarantee and not in the Sixth Amendment itself. See United States v. Gonzalez-Lopez, 548 U.S. 140, 144 –48 (2006). and neither right is typically capable of being asserted by its beneficiary and vindicated in real time—that is, before a violation occurs. It may therefore be defensible as a matter of court administration—if not opti­mal as a normative matter—for the Court to limit the contours of the Brady and Strickland rights to circumstances in which the challenged action or inaction had a tangible impact on the trial’s outcome or the defendant’s sentence. But why should the Court do that with the Sixth Amendment confrontation right, which (again) is set forth in unquali­fied terms and accurately describes the contours of the right—confrontation of witnesses—trial judges ought to enforce?

B. Collateral Review

Epps acknowledges that his proposal would have implications for the doctrines governing the provision of relief on collateral review. That’s because, on collateral review, Chapman’s beyond a reasonable doubt standard does not supply the relevant harmless error decision rule. Rather, under Brecht v. Abrahamson, the error must have had a “substan­tial and injurious effect or influence in determining the jury’s verdict.” 46 507 U.S. 619, 631–32 (1993) (internal quotation marks omitted) (quoting Kotteakos v. United States, 328 U.S. 750, 776 (1946)) (holding that this more forgiving harmless error standard articulated in Kotteakos, which until Brecht had been applied only to non­constitutional errors in federal criminal cases, must be satisfied for a court to award a habeas petitioner relief for any constitutional error that occurred during a state criminal trial). In Brecht, the Supreme Court described this standard as “less onerous” (to the government) than the “stringent” Chapman rule. 47 Id. at 621, 632.

Epps begins his analysis of how his proposal might translate to col­lateral review by suggesting that, if it were adopted, the Supreme Court could simply say that Brecht was wrongly decided. 48 See Epps, supra note 1, at 2185 (noting that Brecht was “a hotly contested 5–4 decision”). Again, fair enough; by its very nature, doctrinal scholarship often involves calls for the Court to revisit precedent. But interestingly, in describing how the Court might proceed to overrule Brecht, Epps says that the Court could acknowledge that it “erred [in Brecht] by requiring federal habeas courts to rely on a different definition of constitutional rights (as [Epps’s] theory neces­sarily implies) than that which ordinary appellate courts must apply.” 49 Id. Thus, Epps implicitly concedes that, if his proposal were to be adopted but Brecht were to remain good law, habeas courts would be required to apply a different definition of the underlying constitutional right than the one that direct-review courts apply. 50 After suggesting that the Supreme Court might consider overruling Brecht, Epps goes on to say that the Brecht principle, if maintained, would simply require, as current law already does, that “federal habeas courts . . . resolve constitutional claims using a more deferential substantive test than would otherwise apply on direct appeal.” Id. While this is true, introducing a new, narrower definition of rights on collateral review would make an already muddled area of the law even more incoherent. See supra section II.A.

This would transform habeas into a fundamentally different regime. Under current law, as one moves from direct to collateral review, the right that is said to have been violated does not change. Indeed, applica­ble habeas exhaustion principles require a petitioner to have presented the claim of right on direct review to have the claim heard on the merits in habeas. 51 See, e.g., Duncan v. Henry, 513 U.S. 364, 365–66 (1995); see also 28 U.S.C. § 2254(b)–(c) (2012) (requiring state court exhaustion of habeas claims in nearly all circumstances). Rather, what changes is the tolerance for the error. On habeas, federalism and finality concerns lead courts to have a higher tol­erance for trial error in two ways. First, under the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), an alleged error cannot ground relief unless the state court’s decision “was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States.” 52 28 U.S.C. § 2254(d)(1). The Supreme Court has interpreted this provi­sion to apply to a state appellate court’s determina­tion that a constitutional violation was harmless under Chapman, and it precludes an award of relief unless the state court’s application of Chapman was unreasonable. 53 See Mitchell v. Esparza, 540 U.S. 12, 17–18 (2003) (per curiam). Second, even if the exacting AEDPA stand­ard is satisfied, the error still must be harmful under Brecht. 54 See Fry v. Pliler, 551 U.S. 112, 119–20 (2007). Confusingly, in Davis v. Ayala, the Supreme Court faulted the lower court for failing to conduct both an AEDPA–Chapman analysis and a Brecht analysis, see 135 S. Ct. 2187, 2198 (2015), even though the latter inquiry “obviously subsumes the former,” Fry, 551 U.S. at 120. For a more complete description of harmless error on collateral review, see Greabe, Harmless Error Revisited, supra note 5, at 76 –78.

Now, consider how all of this would work if the Supreme Court were to retain Brecht but also accept Epps’s proposal to treat the right asserted on direct review as being circumscribed by the Chapman standard. In such a regime, a direct appeal raising, say, a Sixth Amendment Confrontation Clause claim would assert denial of the right to confront a witness whose testimony cannot be deemed (beyond a reasonable doubt) immaterial to the later decision to convict. But should a habeas court entertaining such a claim after it was rejected on direct review follow this approach all the way down? Should it define the right on habeas as a right to confront witnesses whose testimony both cannot be deemed (beyond a reasonable doubt) immaterial to the later decision to convict and has been shown to have a substantial and injurious effect or influ­ence on the jury’s verdict?

Epps suggests that the answer is yes. 55 See Epps, supra note 1, at 2183–85; see also supra notes 49–50 and accompanying text. But as mentioned above, this move would transform habeas from a context in which federal courts take a deferential second look at how state courts have handled a claim of federal right into a context in which federal courts entertain a differ­ent, narrower claim of federal right. True, to avoid this problem and maintain habeas as a second-look system, habeas petitioners could still be required to raise and exhaust this new claim of federal right within the state system. But in order to do so, state criminal defendants wishing to preserve their federal habeas rights during direct appeals and collateral proceedings in state courts would have to assert both that an alleged con­stitutional trial error was not harmless beyond a reasonable doubt and that it had a substantial and injurious effect or influence on the jury’s verdict. And, under AEDPA, habeas courts then would still have to decide whether the state court’s determination under this unwieldy new stand­ard was defensible under clearly established Supreme Court precedent. 56 See supra note 52 and accompanying text.

Doctrinally speaking, this would be madness. It would greatly com­plicate state criminal appeals and collateral proceedings, and it would introduce even more complexity into an area of federal law—the law of habeas corpus—that is already a conceptual disaster area. 57 For a critique of the mess that is harmless error on collateral review under pre­sent law, see Greabe, Harmless Error Revisited, supra note 5, at 109–14. It seems far better to continue to regard the Sixth Amendment right to confront wit­nesses as just that: a right to confront witnesses. Insofar as the Supreme Court remains committed to being stingier with relief for constitutional criminal procedure errors on habeas, it should continue to justify its differential treatment in remedial terms.

C. Section 5 of the Fourteenth Amendment

In taking note of the problems that can arise when the constitutional precedent is translated from one context to another, scholars have natu­rally focused on the different ways precedent is operationalized in differ­ent adjudicatory contexts—that is, as constitutional precedent is trans­ported from appellate courts to trial courts; from federal courts to state courts (and vice versa); and from direct review, to collateral review, to constitutional tort actions. But there is another constitutional lawmaking arena that could be impacted by judicial rulings about the scope of our constitutional rights handed down in other contexts: the authority con­ferred on Congress by section 5 of the Fourteenth Amendment to “enforce” against the states through “appropriate legislation” the due process and equal protection guarantees of section 1. 58 U.S. Const. amend. XIV, §§ 1, 5. Of course, there is not the slightest indication that the current Congress has interest in using its section 5 powers to ensure that state criminal justice systems show greater respect for federal constitutional rights. But things always could change. 59 In contemplating that Congress conduct oversight of state criminal justice sys­tems through use of its section 5 power, I am not the only dreamer. See, e.g., Cara H. Drinan, The National Right to Counsel Act: A Congressional Solution to the Nation’s Indigent Defense Crisis, 47 Harv. J. Legis. 487, 497–508 (2010) (proposing a new federal statute that would authorize a federal cause of action by indigent defendants seeking equi­table relief for systemic Sixth Amendment violations in state courts).

Imagine a wave election in which Congress comes under the control of politicians who are deeply concerned about whether federal rights are being sufficiently observed during state criminal trials. This new Congress holds hearings and develops an extensive body of evidence showing that in states with elected judiciaries in which judicial candidates are fre­quently former prosecutors who campaign on promises to be tough on crime, prosecutors regularly seek to admit—and trial judges regularly do in fact admit—evidence obtained in violation of constitutional criminal procedure rights spelled out in the Fourth, Fifth, and Sixth Amendments, at least as those rights are currently understood. Moreover, appellate courts in such states (also staffed with elected judges) regularly withhold remedies for such violations under Chapman.

Now, imagine that Congress relies on this legislative record to abro­gate prosecutorial and judicial immunity and permit, say, convicts who are later exonerated to file constitutional tort actions under 42 U.S.C. § 1983 against prosecutors and trial judges who participated in particu­larly egregious rights violations at their trials. 60 42 U.S.C. § 1983 (2012) states:
Every person who, under color of any statute, ordinance, regulation, custom, or usage, of any State or Territory or the District of Columbia, subjects, or causes to be subjected, any citizen of the United States or other person within the jurisdiction thereof to the deprivation of any rights, privileges, or immunities secured by the Constitution and laws, shall be liable to the party injured in an action at law, suit in equity, or other proper proceeding for redress, except that in any action brought against a judicial officer for an act or omission taken in such officer’s judi­cial capacity, injunctive relief shall not be granted unless a declara­tory decree was violated or declaratory relief was unavailable. For the purposes of this section, any Act of Congress applicable exclusively to the District of Columbia shall be considered to be a statute of the District of Columbia.
Under current law, when harm is not part of the definition of constitutional criminal procedure rights, the mere wrongful admission of evidence would suffice to estab­lish the necessary predicate of unconstitutional behavior by state actors to justify this intrusion on state sovereignty under section 5. 61 See City of Boerne v. Flores, 521 U.S. 507, 516 –20 (1997) (emphasizing that Congress’s power under section 5 is a power to “enforce” constitutional guarantees and must be exercised in a manner that is congruent and proportional to a pattern of uncon­stitutional state conduct identified by Congress to support the legislation). Such legis­lation would, in other words, likely constitute a permissible “enforcement” of the due process clause. 62 See id.

If the Supreme Court were to include a harm metric in the defini­tion of these rights, it would be much more difficult to frame such legis­lation as enforcement under section 5. In such circumstances, opponents of the legislation would surely argue that Congress needed to do more than develop a record of widespread admissions of evidence in order to establish the necessary predicate of unconstitutional state action. Rather, these opponents would argue, Congress would need to have established a record showing that those admissions of evidence could not reasonably have been understood to amount to harmless error under Chapman. 63 Of course, the proposed, hypothetical legislation might be considered prophylac­tic legislation that is congruent and proportional to the pattern of unconstitutional behav­ior identified. Under section 5, Congress can regulate conduct that is not itself unconstitu­tional. See id. at 518. But the issue certainly would be contested. Obviously, for Congress to develop such a record of case-specific evi­dence of harmfulness under Chapman would not only be onerous; it would be nearly impossible.

This hypothetical may be fanciful given our current political climate. But things always change over time. Thus, it does not seem unduly rights essentialist to suggest that a wholesale, conceptual narrowing of constitu­tional criminal procedure rights could have down-the-line consequences for the scope of Congress’s power to enforce federal constitutional guar­antees under its section 5 power.

III. A Better Path for Reforming Harmless Error Doctrines

Nothing I have said should be taken as disagreement with Epps’s thesis that the Supreme Court should act to clear up the many mysteries of harmless error review. I only disagree that the way to do so is to recon­ceptualize any right whose violation may be harmless as including a harm metric in the definition of the right. Epps is absolutely correct to argue that the Court should provide much greater clarity about what harmless error review is, which rights violations should be subject to harmless error review, how harmless error review should be conducted, and the relationship between judicially created harmless error doctrines and 28 U.S.C. § 2111, the federal harmless error statute. 64 See supra note 13 and accompanying text. But greater clarity can be achieved without abandoning the understanding of harmless error as a set of doctrines governing the provision and withholding of remedies for constitutional violations.

In recent work, I have argued that the Supreme Court’s harmless error doctrines unduly privilege constitutional error vis-à-vis nonconsti­tutional error; 65 See Greabe, Harmless Error Revisited, supra note 5, at 96–101. prescribe an easily and often manipulated jurisprudence of labels to determine whether an error is amenable to harmless error review; 66 See id. at 101– 08. and unnecessarily complicate the application of harmless error analysis on collateral review while at the same time showing insufficient regard for rule of law values. 67 See id. at 109–14. My critique of the doctrine and Epps’s cri­tique are largely consistent.

To address these problems, I propose that we start with a very basic question: What, if anything, does the Constitution require of reviewing courts when confronted with a criminal judgment arguably tainted by constitutional error? Thus far, commentators have sought to answer this question by focusing on Chapman’s ontology and asking, what is the Chapman beyond a reasonable doubt harmlessness principle? 68 See supra notes 10–12 and accompanying text. I suggest the possibility of a conceptual breakthrough by supplementing this inquiry with an analysis of remedial function that focuses on how the reversal or vacatur of a criminal conviction serves as a remedy for a con­stitutional violation at trial. 69 See Greabe, Harmless Error Revisited, supra note 5, at 64, 79–86. Analyzing remedial function is useful because it helps to establish the dividing line between constitutionally necessary remedies and constitutionally discretionary remedies: As a general matter, the Constitution requires courts to provide specific rem­edies responsive to ongoing constitutional violations but permits courts to withhold substitutionary remedies responsive to wholly completed constitutional wrongs. 70 See id.

Thus, reviewing courts “must remedy an ongoing infringement of constitutional rights worked by conviction under a facially unconstitu­tional statute or a statute that cannot be constitutionally applied on the facts of the case.” 71 Id. at 64. But otherwise, “the Supreme Court and Congress are . . . free to craft harmless-error doctrines that reflect the lessons of experience.” 72 Id. This lawmaking discretion “flows from the fact that, once we set to the side the exception involving conviction pursuant to an unconstitutional or unconstitutionally applied statute, an order vacating or reversing a tainted judgment provides substitutionary relief for a wholly concluded wrong; it does not deliver a constitutionally compelled remedy.” 73 Id.

With the path to reform thus largely clear, how should the Supreme Court or Congress exercise lawmaking discretion to improve the various doctrines that combine to comprise harmless error review? My own view is that, absent action by Congress, the Supreme Court should do away with the Chapman beyond a reasonable doubt principle, the structural defect–trial error dichotomy that presently informs whether an error can be subject to harmless error review, 74 In Arizona v. Fulminante, 499 U.S. 279, 306 – 09 (1991), the Supreme Court held that courts conducting direct review of criminal convictions should always provide reme­dies for “structural defects” but should conduct harmless error review of all “trial errors.” But the Court subsequently put this framework into operation in unpredictable and unprincipled ways. See Greabe, Harmless Error Revisited, supra note 5, at 63, 101– 08. and the Brecht principle. 75 See Greabe, Harmless Error Revisited, supra note 5, at 65, 115–22. In their place, the Court should require, with respect to errors that are subject to harmless error review, a simplified, unitary, and transcontextual analysis—reconceived as an elaboration of 28 U.S.C. § 2111 76 Note that my approach, like Epps’s, solves any legitimacy problem arising from the fact that Chapman is best characterized as constitutional common law—a body of law whose legitimacy is open to challenge especially insofar as it is held binding on the states. See supra note 12. In my view, harmless error doctrine should be an elaboration of a fed­eral statute, 28 U.S.C. § 2111, and not seen as constitutional principle, as Epps argues. —that tracks the approach for which Justice Roger J. Traynor argued in his well-known book, The Riddle of Harmless Error. 77 Roger J. Traynor, The Riddle of Harmless Error 49–51 (1970). This approach would involve telling all reviewing courts (whether they are conducting direct or col­lateral review) to set aside a judgment tainted by any error (whether constitutional or not) unless they conclude that it is “highly probable” that the error did not affect the judgment. 78 See Greabe, Harmless Error Revisited, supra note 5, at 65, 115–22.

With respect to which errors ought to be subject to harmless error review, the Court should not lose sight of the fact that, even when factu­ally guilty of the crimes with which they were charged, appellants and petitioners asserting claims of constitutional error “serve as private attor­neys general and, in that role, function as essential instruments for ensur­ing proper regard for fundamental constitutional [values].” 79 Id. at 120. Thus, the Court should instruct reviewing courts to exercise their power to set aside judgments tainted by error in which, regardless of whether there is a high probability that the error affected the judgment, an exercise of remedial discretion is needed to vindicate such values. 80 Id. Certainly, such circumstances include errors that undermine the rights to an impartial judge, a jury instruction that correctly states the relevant standard of proof, the assistance of counsel for the accused, and a fair jury. 81 Id. at 120 & n.325 (taking note of an apparent consensus among Supreme Court Justices that the fair trial guarantee always requires the observance of these rights). In my view, they also should include, at the very least, judicial proceedings marred by unconstitutional discrimination on the basis of race, religion, ethnicity, national origin, or gender and intentional misconduct by gov­ernment officials such as judges, prosecutors, and policy or probation officers. 82 Id. at 121. For an excellent elaboration and analysis of why harmless error review must account for constitutional values beyond the accuracy of criminal convictions, see generally Justin Murray, A Contextual Approach to Harmless Error Review, 130 Harv. L. Rev. 1791 (2017).


Professor Epps has written an interesting article that makes a persua­sive argument that our harmless error doctrines should be reformed. But the path to reform should not involve a narrowing of the definitions of constitutional criminal procedure rights. Such a narrowing would engender serious trial-administration problems and precedent that would not travel well into other adjudicatory and lawmaking contexts. For all of their faults, the harmless error doctrines are properly concep­tualized as remedial.