Reliability matters: reassociating Bagley materiality, Strickland prejudice, and cumulative harmless error.

• Author: Blume, John H.

INTRODUCTION

Certain errors ... are like apples, while others ... are like oranges. Both share the same shape (they are errors), but they have decidedly different textures, colors, and tastes, making it extremely difficult to judge their combined effect on the jury. (1)

Categories are made, not found.... [O]ur categories do not derive from the shape of the world but create it. (2)

Most commonly invoked after conviction and direct appeal, when a defendant may claim that his lawyer was ineffective or that the government failed to disclose exculpatory information, the Brady (3) doctrine, which governs the prosecutor's duty to disclose favorable evidence to the defense, and the Strickland (4) doctrine, which monitors defense counsel's duty to represent the client effectively, have developed into the principal safeguards of fair trials, fundamental to the protection of defendants' constitutional rights and arguably defendants' strongest insurance of a reliable verdict. (5) But the doctrines do not sufficiently protect these core values.

The doctrines, despite their common due process heritage and symbiotic development, are generally divided when assessing prejudice. Even in cases where the defendant alleges both that the prosecution withheld evidence and that his counsel was incompetent, courts assess the impact of each party's conduct on the verdict independently. Our objectives here are two-fold. Our more modest objective is to argue that courts should consider the impact of Brady violations and Strickland violations together when evaluating whether a guilty verdict or death sentence is reliable. Without considering their simultaneous impact on jurors, the reliability of a verdict cannot fairly be measured. A divide-and-conquer approach is inconsistent with reliability, the touchstone of both doctrines. This is important because the Brady and Strickland doctrines, besides being the most commonly raised types of post-conviction error, govern the core functions of the main players in the adversarial system. Few courts, and no commentators, however, have directly tackled this issue.

Our second objective is more ambitious. If Strickland and Brady errors should be considered jointly when assessing prejudice, then why shouldn't the impact of all errors that potentially affect the reliability of a verdict be taken into account? By this, we mean errors that affect the information the jury considers and errors that affect the manner in which the jury considers the information it receives: denials of expert assistance, (6) denials of the right to confront one's accuser, (7) and prosecutorial misconduct that skews the factual presentation, to name a few. Such a global reliability inquiry is the alleged goal of cumulative harmless error doctrine, a branch of due process analysis addressing claims for relief based on multiple errors that do not warrant relief in isolation. But while cumulative harmless error analysis could provide a vehicle for overarching inquiry, the doctrine is inconsistently and rarely applied. Fears that fundamental fairness determinations could swamp other rules of criminal procedure, or displace state interests in finality in favor of omnipresent federal review, motivate courts to narrow the cumulative-error calculus. For example, many courts exclude suppression of evidence or mistakes by defense counsel, categorizing them as non-errors if they fail to satisfy the Strickland or Brady prejudice requirements. A verdict's reliability cannot sensibly be measured by assessing deficiencies of counsel, prosecutorial misconduct, and any other errors affecting reliability in isolation from one another. Such a procrustean, divide-and-conquer approach sacrifices reliability. It preserves verdicts, but guarantees that many of them will be unreliable. In sum, reliability cannot be assessed piecemeal.

Part I begins by tracing the history of the Strickland and Brady doctrines. We focus on three defining moments. The first is the United States Supreme Court's decision in the well-known case of Powell v. Alabama, which established a criminal defendant's right to counsel as a fundamental right protected by the Due Process Clause of the Fourteenth Amendment, the same legal precept the Court relied on in prior cases where the prosecution withheld or presented false evidence. The second is the initial development of prejudice standards to govern failures of defense counsel and prosecutorial suppression, highlighted by Brady v. Maryland, which put forth the materiality requirement for suppression errors, and Gideon v. Wainwright, in which the Supreme Court (while delivering the Sixth Amendment right to counsel to the states) adhered to the due-process roots of the right it recognized in Powell. The third and culminating moment is the Supreme Court's creation during its 1984-85 term of the mirror-image prejudice standards for ineffective assistance of counsel and suppression claims in Strickland and in United States v. Bagley, respectively. This history shows that laws governing the right to counsel and suppression of evidence have long shared the same core value, reliability of outcomes, and have long applied a broadly inclusive measure of error, totality-of-the-circumstances review, to effectuate it. The historical discussion ends with a look at recent precedent, which emphasizes that totality-of-the-circumstances review as applied in Strickland and Bagley is a cumulative consideration. Together, the doctrines' history and character anchor our argument.

Beginning with Part II we address three problems that we see as dividing reliability determinations. The first, failure to cumulate defense counsel errors in Strickland prejudice analysis, is really a non-problem. The Court's recent decisions applying Strickland show that with respect to ineffective assistance of counsel claims defense counsel's conduct "taken as a whole" must be considered in assessing prejudice. (8) Still, we think it is significant to point out this trend that divides reliability determinations in some courts. In the process, we see flawed reasoning that will reemerge later when we discuss cumulative harmless error doctrine.

We next consider the relationship between the Brady and Strickland doctrines, in Part III. Our call is for integrating Strickland prejudice and Bagley materiality. To some what we call for may seem obvious: a totality of the circumstances test, which both Strickland and Bagley are, surely takes into account everything that the jury should have heard if the system had worked properly. Considering suppression missteps in the Strickland prejudice prong, and evaluating defense counsel's errors in assessing Bagley materiality is, therefore, already written in the law. But others may see a difficulty: how can a due process violation (Brady/Bagley) combine with a Sixth Amendment violation (Strickland) to invalidate a conviction? What would such a hybrid be called? Integrating the prejudice arising from suppression and ineffective-assistance is inherent in the Bagley and Strickland standards. Both are totality of the circumstances standards derived from due process; the Bagley and Strickland standards reflect the same core constitutional value, reliability. That the doctrines presently reside under different constitutional amendments with traditionally different focuses, we argue, is ultimately of little concern. Considering suppressed evidence within the contour of Strickland prejudice, or defense-counsel failures in the context of Bagley materiality, is only being true to the function of each, to guarantee reliable verdicts.

Part IV expands the argument for cumulation beyond ineffective assistance of counsel and prosecutorial suppression. Embarking upon our second objective, we consider the obstacles to unified reliability determination posed by cumulative harmless error analysis. We advocate including all errors at the trial that impact verdict reliability in the scope of cumulative error analysis. The same reason for integrating Strickland and Brady prejudice--that reliability cannot be measured piecemeal--supports unifying the prejudice associated with other errors that threaten reliability. We propose an approach that need not conflict with justified limits on collateral review.

Courts frequently assess reliability in pieces, considering the impact of a portion of errors, rather than assessing how the panoply of trial errors affect the reliability of a verdict overall. Our call here is for a rule of integration requiring courts to consider the unified impact on the verdict of all errors affecting reliability.

1. BACKGROUND: THE HISTORY AND CHARACTER OF BAGLEY AND STRICKLAND

   In general terms, this article concerns what errors can and ought to be considered together in determining reliability. A necessary starting point, then, is identifying what errors affect the reliability of a verdict. This is a question that, starting on a blank slate, would merit exploration with respect to every conceivable error. But we are not on a blank slate. We draw on hundreds of years of precedent, in which errors have already been defined. Among the current mechanisms in the law for gauging the reliability of verdicts, three are prominent: the Brady doctrine, which governs prosecutorial suppression; the Strickland doctrine, which governs performance of defense counsel; and cumulative harmless error analysis, which accumulates errors, alone harmless, to assess whether their collective effect undermines the fairness of the proceedings. Rather than set off on a free-form journey to define reliability-impacting errors, therefore, we examine the state of reliability assessment, focusing on the success or failure of these doctrines.

   It is our position that reliability has meaning only when assessed globally. If divided, considered error by error, one cannot measure it. If one focused on the reliability-impact of each individual error, one would...