Reflections on the Brady Violations in Milke v. Ryan: Taking Account of Risk Factors for Wrongful Conviction

pdfCatherine Hancock∞

I. Introduction. 438

II. Milke in Context: Localized Brady Resistance in the Exoneration Era 440

III. Milke as a Case Embedded in a Culture of Brady Resistance. 444

A. The Personnel File and the Benefits of Non-Disclosure for the Milke Prosecutor 444

B. The “Game-Changer” Misconduct Evidence in the Court Records and Judicial Failure to Apprehend Its Significance 450

IV. Considering the Risk Factors for Wrongful Conviction in Milke. 454

V. The Next Milke Trial: Reflections on the Continuing Risks of Wrongful Conviction 460


“The jury [had] nothing more than [the detective’s] word that Milke confessed. Everything the [S]tate claims happened in the interrogation room depends on believing the detective’s testimony. Without [his] testimony, the prosecution had no case against Milke[.] [T]he Constitution requires a fair trial, and one essential element of fairness is the prosecution’s obligation to turn over exculpatory evidence. This never happened in Milke’s case and so the jury trusted [the detective] without hearing of his long history of lies and misconduct.”[1]



Brady violations contribute to wrongful prosecutions, wrongful convictions, and wrongful sentences, including death penalty verdicts.[2] The defendant in Milke v. Ryan, Debra Milke, spent 22 years on Arizona’s death row before her defense counsel succeeded in obtaining her release in 2013 based on the Ninth Circuit’s recognition of the prosecutor’s Brady violations in her case.[3] In a recent empirical study sponsored by the National Institute of Justice (NIJ Study), cases of wrongful convictions were compared with cases in which wrongful prosecutions were halted before they reached the wrongful conviction stage.[4] The purpose of the NIJ Study was to identify the factors that are uniquely associated with the failure of prosecutors to halt the prosecutions that become wrongful convictions. The NIJ Study determined that one of those risk factors is the existence of Brady violations[5]—the failure of the prosecutor to disclose to the defense “favorable evidence rising to a material level of importance.”[6] The interaction of other factors can magnify the risk of wrongful conviction, as when a Brady violation occurs in a case with weak prosecution evidence and a lying witness.[7] The Milke prosecution exhibited many of the risk factors identified in the quantitative results of the NIJ Study, as well as a variety of the factors that emerged in the qualitative observations.[8]

This article offers reflections upon Milke as a case study of how the NIJ risk factors can illuminate the significance of Brady violations and their impact upon a defendant’s ability “to present a complete defense” and avoid a wrongful conviction.[9] By violating Brady and Giglio v. United States,[10] the prosecutor in Milke was able to construct a trial narrative of guilt using the testimony of a detective with a history of misconduct, which included unconstitutional interrogations and false testimony about confessions.[11] Milke testified that she was innocent, that she never confessed to the detective, and that he violated her Miranda rights by ignoring her request for counsel.[12] The detective testified that he gave Milke the Miranda warnings, that she never invoked her rights to counsel or silence, that she knowingly waived her rights orally, and that she confessed to conspiracy to murder.[13] The prosecution’s confession narrative, as prepared and performed on the stand by an experienced police witness, was effectively unimpeachable without access to the undisclosed evidence of his misconduct.[14] Both the Milke prosecutor and the state trial judge exhibited their resistance to the enforcement of Brady by ignoring its well-settled constitutional mandate at every phase of the litigation.[15] For Milke, Brady’s Due Process doctrine provided no protection from the risk of a wrongful conviction.[16]

This article proposes that judges, prosecutors, and defense counsel need to become aware of how the risk factor of a Brady violation can interact with other risk factors and contribute to breakdowns in the adversarial process that increase the risk of wrongful conviction. Such awareness should lead courts to enforce Brady obligations to better insure against that risk. The Milke prosecution illustrates how resistance to the enforcement of Brady by prosecutors and judges reflects a lack of awareness of how the Brady duty of disclosure serves as a vital safeguard against very real risks. In the era of exonerations, courts can no longer rely on old assumptions about the reliability of convictions. Yet those assumptions continue to be reflected in the tolerance of Brady violations by some courts.

Part II of this article establishes the context for the Ninth Circuit’s Milke decision by contrasting the present awareness of the reality of wrongful convictions with the lack of such awareness during the era of Milke’s conviction. Part III explores the ways in which Milke’s case illustrates resistance to the enforcement of Brady by the prosecutor and the courts that reviewed the case before it reached the Ninth Circuit. Part III(A) analyzes the events of the ten-year saga of the Brady violation that occurred when the prosecutor and trial judge improperly rejected defense counsel’s request for disclosure of the detective’s personnel file as a source of Giglio impeachment evidence. Part III(B) analyzes the almost twenty-year saga of the multiple Brady violations based on the prosecutor’s failure to disclose the pattern of judicial orders and rulings regarding the detective’s history of lying under oath and conducting unconstitutional interrogations, as reflected in court records. These analyses reveal how the prosecutor obtained a variety of benefits from these Brady violations, and how reviewing courts failed to apprehend their significance, even in a capital case in which the defendant claimed to be innocent.

Part IV of the article explains how the Milke record exhibited five of the risk factors for wrongful convictions identified in the NIJ Study’s quantitative findings, as well as additional risk factors revealed in its qualitative findings. Part V considers the ongoing resistance to the Ninth Circuit decision displayed by the current Milke prosecutor, who is determined to obtain Milke’s conviction again by using the detective’s testimony, and who has obtained a pre-trial ruling to prevent the detective from invoking the Fifth Amendment. This section concludes with reflections regarding the continued existence of the risk factors for wrongful conviction in Milke’s case and the prosecutor’s resistance to the Ninth Circuit’s ruling.

Milke in Context: Localized Brady Resistance in the Exoneration Era

The Milke prosecution began in 1989 at a time when “virtually all observers assumed that the innocent were rarely convicted.”[17] By contrast, the Ninth Circuit’s validation of Milke’s Brady claim occurred in the present era of increased publicity regarding the growing numbers of exonerations.[18] There have been 1,430 exonerations since the first use of post-conviction DNA evidence for that purpose in 1989.[19] DNA evidence could not help Milke because her conviction was based on the detective’s testimony about her confession, not on physical evidence.[20] But more than 70% of exonerations have not depended on DNA evidence.[21] According to one estimate in 2005, the number of wrongful convictions during the prior 15 years “must be in the thousands, perhaps tens of thousands.”[22] However, in the earliest years of the exoneration era, when the Arizona courts rejected Milke’s Brady violation claim,[23] her innocence argument would not have raised a serious concern about her possibly wrongful conviction.

Yet Milke’s insistence on her innocence[24] was integral to her arguments that the detective fabricated her confession and lied about his adherence to Miranda, and that the prosecutor violated Brady in order to protect the detective’s credibility from impeachment. As more than two decades passed while Milke lived on death row, exoneration reforms expanded[25] and litigation exposed the reality of wrongful convictions, including those based on Brady violations[26] and on false confessions procured through coercive interrogations[27] or fabricated by police officers.[28] Thus, Milke’s arguments became more credible over time. The NIJ Study is one of the latest developments that supports the credibility of those arguments.[29]

One unchanging reality during the years following Milke’s conviction was the “highly localized character” of disclosure policies used by prosecutors’ offices,[30] ranging from “full open file discovery” systems that exceed Brady’s requirements to office practices that reflect a culture of resistance to Brady.[31] If state law requires open file discovery, then “most Brady evidence” will be disclosed routinely.[32] But when an open file policy is adopted by a prosecutor’s office, the scope of required disclosure may vary considerably.[33] For example, such policies may allow defense counsel to “view all information gathered in a case,” or may allow “substantial but not total access [to] files.”[34]

However, some offices choose to follow narrow policies that strictly limit their disclosures of Brady material,[35] and some prosecutors develop “shared understandings” that resistance to Brady is acceptable for some types of evidence.[36] Some offices are notorious for their culture of tolerance for Brady violations.[37] Like the selective use of the death penalty by prosecutors in some states,[38] the selective denial of Brady rights reflects the attitudes of local prosecutors and judges, not only toward Brady, but also toward the risks of wrongful prosecutions and convictions associated with Brady violations.[39] Not surprisingly, the lack of enforcement of Brady has provoked a constant stream of criticism[40] and proposals for reform.[41]

Given the entrenched nature of the autonomous and localized Brady cultures, the ability of Milke’s defense counsel to vindicate her claims was dependent on the culture of the Arizona courts.[42] One measure of that culture is reflected in the findings of a recent study described in the Arizona Republic: “In Arizona, prosecutorial misconduct is alleged in half of all capital cases that end in death sentences. Half the time, the Arizona Supreme Court agrees that misconduct occurred in those instances, but it rarely throws out a conviction or sentence because of it.”[43]


Milke as a Case Embedded in a Culture of Brady Resistance

One of the Brady and Giglio violations in Milke involved the prosecutor’s longstanding resistance to the disclosure of the detective’s personnel file sought by Milke’s defense counsel at trial and in subsequent proceedings.[44] The other violations concerned the resistance to the disclosure of court records, unknown to the defense, which contained judicial findings regarding the detective’s false testimony and unconstitutional interrogations in prior cases.[45] As it turned out, the discovery of the court records by post-conviction counsel accomplished more than support for a renewed request for access to the file. The “pattern” of misconduct revealed in the court records constituted “highly relevant” and “highly probative” evidence that “would certainly have cast doubt” on the detective’s credibility if used to impeach his testimony at trial.[46] Yet in the years before the Ninth Circuit’s decision, no reviewing court recognized either the impeachment value of the court records or the merit of Milke’s Brady violation claims regarding their non-disclosure.[47] Eighteen years after the discovery and presentation of the records in the post-conviction petition, the Ninth Circuit granted a new trial for Milke based on the prosecution’s concealment of the Giglio evidence in both the file and the records.

  1. The Personnel File and the Benefits of Non-Disclosure for the Milke Prosecutor

The Milke prosecution provides a vivid example of how Brady rights may be impossible to enforce, even when a prosecutor knowingly violates Brady and suppresses favorable and material evidence that is requested repeatedly by defense counsel.[48] The Milke detective’s personnel file contained “Brady and Giglio evidence of the most egregious kind,”[49] and the defense unsuccessfully sought to obtain disclosure of the file both during the trial[50] and in the post-conviction petition.[51] By the time the federal district court granted the renewed defense request for disclosure in the habeas petition, the prosecutor had already reaped a variety of benefits by resisting disclosure for ten years.[52]

The prosecutor gained three immediate benefits by not disclosing the file to Milke’s counsel before trial. First, the file contained an internal investigation report[53] that cast doubt upon the detective’s credibility because of his supervisor’s judgment that his “image of honesty, competency, and overall reliability must be questioned.”[54] The report described a five-day suspension, which had been imposed because the detective had traded sex in exchange for not arresting a female motorist on an outstanding warrant.[55] He had “steadfastly lied about the incident until he failed a polygraph test.”[56] The evidence demonstrated the detective’s willingness “to abuse his authority to get what he wants” and to lie “during the course of his official duties,” as well as his “misogynistic attitude toward female civilians.”[57] Moreover, these attitudes and behaviors were “highly consistent with Milke’s account of the interrogation.”[58] By withholding the report, the prosecutor was able to both “impugn Milke’s credibility” and “hide the evidence that undermined [the detective’s] credibility”[59] regarding his testimony about the interrogation.

The second immediate benefit of withholding the file was the concealment of the detective’s annual reviews that included references to other investigations and prosecutions in which he had participated. For example, a list of six of the detective’s “high-profile cases” appeared in one of his evaluations.[60] The Ninth Circuit recognized that this information might have led the defense to locate judicial findings of the detective’s misconduct that were buried in court records,[61] none of which the prosecutor chose to disclose to the defense.[62]

A third benefit of withholding the file was that its disclosure would have revealed that it was suspiciously thin, as it contained only two annual reviews from the detective’s 21-year career.[63] Without access to the file, the defense counsel could not argue that the disappearance of the lost or destroyed annual reviews was evidence of an attempt to cover up the detective’s history of misconduct.[64] Ultimately, the non-disclosure of the file worked in several different ways to achieve the prosecutor’s goal of presenting the detective as the credible witness and Milke as the liar regarding her claim of innocence and her Miranda-violation claim. The rewards for violating Brady loomed large.[65]

Lacking access to the impeachment evidence in the file, Milke’s trial counsel sought to challenge the detective’s version of the interrogation by other means. Having failed to obtain a pre-trial ruling excluding the confession from evidence,[66] defense counsel called upon Milke to present her side of the swearing contest by rebutting the detective’s testimony regarding the interrogation.[67] Milke testified that after she received Miranda warnings, she told the detective that she did not understand them and that she needed a lawyer.[68] But the detective ignored her statements and continued to interrogate her. Such conduct would have violated his clear duty under Miranda to cut off interrogation after Milke’s invocation of the right to consult counsel. [69] Milke also testified that in his account of the interrogation, the detective “embellished and twisted [her] statements to make it sound like she had confessed.”[70] Thus, Milke’s testimony supported the defense argument that the detective was willing to lie about any aspect of Milke’s interrogation under oath, and to fabricate her confession in order to get her convicted.[71]

Defense counsel also challenged the detective’s credibility by calling the jury’s attention to his suspicious conduct regarding the interrogation. In effect, the defense counsel argued that by not recording the confession, the detective employed a calculated strategy or modus operandi for the purpose of immunizing his own credibility from later challenge at trial.[72] For example, he admitted that it was his practice not to record his interrogations.[73] He did not bring a recorder with him because he did not intend to record Milke’s interrogation.[74] He also admitted that he disobeyed an express directive to record that interrogation.[75] He even asked other police officers to leave him alone with Milke.[76] He destroyed his notes of the interrogation, which meant that no one could compare his contemporaneous observations with either his police report or his testimony.[77] Without witnesses and without a recording,[78] the detective knew that if Milke contradicted his account of the interrogation, it would be only her word against his.[79]

The trial judge did not allow defense counsel to introduce evidence in support of this argument about the significance of the detective’s failure to record Milke’s interrogation. For example, the judge rejected counsel’s request to present expert testimony about the detective’s interrogation practices,[80] and disallowed cross-examination of the detective about his interrogation methods in another case.[81] When the prosecutor asked the jurors to rely on the detective’s testimony, they lacked the experience to realize that a confession “proved by the say so of a single officer,” would be an unprecedented basis for a conviction.[82]

The Ninth Circuit determined that the state trial judge should have ordered the prosecutor to disclose the impeachment evidence in the detective’s personnel file, instead of granting the motion to quash the subpoena that Milke’s counsel had issued to obtain that evidence.[83] The state court “never complied with Brady,”[84] either at trial or in post-conviction proceedings, because the judge ignored the controlling authority of Brady and Giglio as well-established, “long-standing Supreme Court caselaw.”[85] As in Giglio, the Milke prosecutor violated Brady by failing “to turn over impeachment evidence about the key witness, whose testimony was essential to the case.”[86] The information in the personnel file clearly “fit within the broad sweep of Giglio.”[87] In rejecting Milke’s post-conviction Brady argument for the disclosure of the file,[88] the judge applied “the wrong legal authority” in faulting defense counsel for failing “to explain why the information was validly discoverable.”[89] The judge should have recognized that under Giglio, “material impeachment evidence isn’t just discoverable,” but “must be disclosed unilaterally as a matter of constitutional right.”[90]

The denial of Milke’s Brady right of access to the detective’s file was the result of the prosecutor’s knowing resistance to the fulfillment of its Brady duty, which was aided by the state judge’s misapprehension of Brady and Giglio.[91] The prosecutor was “charged with knowledge that there was impeachment material” in the file, namely the suspension report.[92] The Ninth Circuit observed that the prosecutor offered “no excuse for failing to turn over” the report before trial, “nor can we imagine any legitimate reason for this failure.”[93] As a consequence of the failure “to apply the correct controlling authority,” the state court’s Brady ruling that enabled the prosecutor to resist disclosure was “contrary to . . . clearly established Federal law,” and thus, a federal habeas court could “intervene” and consider the merits of the Brady claim.[94] Moreover, the prosecutor’s “misfeasance” created a second basis for such intervention.[95] By withholding the suspension report, the prosecutor “distorted the fact-finding process” of the state court and “rendered the fact-finding ‘process . . . defective.’”[96] Therefore, the court’s decision was “based on an unreasonable determination of the facts.”[97] On the merits, the Ninth Circuit briskly reversed the state trial judge’s decision regarding the detective’s file, finding that the prosecutor’s knowing suppression of favorable evidence had prejudiced Milke.[98]

Notwithstanding the significance of the prosecutor’s Brady violation regarding the suspension report, this evidence of the detective’s misconduct would turn out to be the proverbial tip of the impeachment iceberg. Most of the Ninth Circuit’s criticisms of the state post-conviction judge and the prosecutor were reserved for their conduct regarding the suppression of the detective’s long history of misconduct that was buried in court records and never disclosed to Milke’s counsel.[99]

  1. The “Game-Changer” Misconduct Evidence in the Court Records and Judicial Failure to Apprehend Its Significance

Milke’s post-conviction counsel anticipated the state trial judge’s continued unwillingness to grant access to the detective’s file, and therefore sought to find other records of the detective’s misconduct. The immediate goal of obtaining such evidence was to persuade the judge to grant disclosure of the file on the theory that it was likely to contain additional misconduct evidence.[100] With extraordinary assistance from others, counsel managed to direct an extensive excavation of the court records in a time-consuming search for evidence that would have been impossible to undertake before the trial. The successful collection of this evidence depended upon a “Herculean” effort[101] by a team of researchers, who spent 7,000 hours over 3.5 months looking for the detective’s name in the court records for criminal case files during the eight-year period preceding Milke’s sentencing.[102] The researchers found 100 cases in which he had played some role, and they made copies of documents taken from 18 cases that illustrated his misconduct.[103] The defense submitted this “trove of undisclosed impeachment evidence” as an appendix to the post-conviction petition that incorporated hundreds of pages.[104] However, the state judge summarily dismissed the petition without discovery or an evidentiary hearing.[105]

The Ninth Circuit recognized that the court records would have been “game-changer” impeachment evidence at Milke’s trial.[106] Ultimately, the court chose to focus on the records from seven cases[107] in which the detective’s testimony exhibited a “menagerie of lies and constitutional violations.”[108] In each case, judicial orders or rulings had been issued based on findings of the detective’s misconduct.[109] Four cases made out “a Giglio violation on their own”[110] by revealing the detective’s readiness to lie under oath “in order to secure a conviction or further a prosecution.”[111] In these cases, judges “threw out indictments or confessions because [the detective] had lied to a grand jury or judge.”[112] Four cases “presented additional Giglio evidence,” and the judges in these cases “threw out confessions or vacated conviction” because the detective had violated “Miranda and other constitutional rights during interrogations, often egregiously.”[113]

Thus, the pattern of misconduct in the court records could have been used at trial to support Milke’s testimony regarding both the detective’s Miranda violation and his false account of the interrogation. His history of unconstitutional interrogation practices “could have shown the jury that he habitually circumvented Miranda,” and “that [he] planned, from the outset, to conduct an illegal interrogation by confronting Milke alone, without a tape recorder.”[114] The trove of judicial findings would have been “highly probative” evidence for the jury’s determination about what happened when the detective interrogated Milke “behind closed doors, after which he emerged claiming to have extracted a confession.”[115]

When denying Milke’s post-conviction petition, the state trial judge’s resistance to the enforcement of Brady took the form of misapprehension regarding both the facts and the law related to the Giglio evidence in the court records. The judge’s first error was to declare that the court records “establishe[d] nothing” because they contained only “motions and testimony” that involved mere allegations of the detective’s misconduct.[116] This comment revealed that either the judge had not read the submitted records[117] or she “grossly misapprehended the nature and content of the documents,” which contained judicial orders and rulings.[118] This misapprehension “fatally undermine[d] the fact-finding process,” rendering the state court findings unreasonable.[119] The judge’s second error was her failure to consider that the evidence of the detective’s prior violations of Miranda and Due Process would have been admissible to impeach him at Milke’s trial.[120] As the Ninth Circuit noted, the evidence would have been admissible to impeach the detective’s credibility because it was both relevant and highly probative, and any decision to exclude such evidence would have violated Due Process.[121]

The Ninth Circuit’s assessment of the suppression of the court records left no doubt regarding its judgment concerning the prosecutor’s “cavalier attitude” toward Brady obligations during the time when Milke and her counsel were “fighting for her life.”[122] In surveying the favorable nature of the impeachment evidence in the court records, the Ninth Circuit described each example of the detective’s misconduct in detail, explaining the significance of each false statement under oath, and the particulars of each unconstitutional interrogation practice.[123] In assessing the prejudice caused by the prosecutor’s suppression of the records, the court concluded that since the detective’s credibility “was crucial to the state’s case against Milke,” it was “hard to imagine anything more relevant to the jury’s – or the judge’s – determination whether to believe [the detective] than evidence that [he] lied under oath and trampled the constitutional rights of suspects.”[124] Like the prosecutor’s knowing suppression of the detective’s file, the Ninth Circuit reasoned that the suppression of the court records must have been knowing, as the Milke prosecutor and his colleagues must have been “intimately familiar”[125] with the detective’s pattern of misconduct because “it had harmed criminal prosecutions.”[126] Unlike the suppression of the file, the suppression of so many instances of misconduct in the records demonstrated the prosecutor’s willingness to ignore repeatedly the detective’s known “propensity to commit misconduct”[127] in order to obtain confessions and convictions in high-profile murder cases.[128]

In its final statement after granting habeas relief to Milke, the Ninth Circuit declared that the clerk of court “shall send copies of this opinion” to the U.S. Attorney’s Office in Arizona and the Assistant U.S. Attorney General of the Civil Rights Division “for possible investigation into whether [the detective’s] conduct, and that of his supervisors and other state and local officials, amounts to a pattern of violating the federally protected rights of Arizona residents.”[129] Five months later, the U. S. Attorney’s Office in Arizona sent a letter to the Milke prosecutor’s office to say that the federal statute of limitations had expired for possible civil rights violations committed by the Milke detective, and therefore, no investigation would be opened.[130] By this time, the prosecutor had announced his intention to put Milke on trial again.[131]

Considering the Risk Factors for Wrongful Conviction in Milke

The NIJ’s quantitative study identified ten risk factors as being associated uniquely with wrongful prosecutions that were not halted before reaching the wrongful conviction stage.[132] The Milke prosecution exhibited five of those risk factors: a Brady violation, a lying witness, a weak defense, a local pro-death-penalty culture, and a weak prosecution case. Further qualitative findings identified additional risk factors that are evidenced in the Milke record as well.[133] The Study observed that wrongful convictions occur “because of a breakdown in the accuracy of human judgment at multiple levels: police investigation, prosecution, pre-trial motions, judicial rulings, [and] jury verdicts.”[134] Such breakdowns are produced by “a ‘perfect storm’ of system failure.”[135] Like the interaction of the risk factors in the NIJ Study’s wrongful conviction cases, the Brady violations in Milke’s case interacted with other risk factors to produce “a perfect storm of misfortune” for Milke.[136]

Two of the NIJ risk factors in Milke were documented in the Ninth Circuit’s opinion: the existence of Brady violations and the likely presence of a “lying witness” at trial.[137] Although the falsity of the detective’s testimony has remained a subject of dispute, his lack of credibility was evidenced not only by the Ninth’s Circuit’s analysis of his documented history of lying under oath,[138] but also by the expert’s report in the habeas petition regarding the implausibility of the detective’s testimony.[139]

A third risk factor in Milke was the weakness of the defense, as illustrated in several aspects of the record.[140] Usually this factor derives “from either poor defense counsel or a lack of exculpatory evidence—or both.”[141] In Milke’s case, it was both. Milke’s trial counsel “had been an attorney for under seven years,” and had tried only one capital case.[142] He failed to interview witnesses, obtain relevant records of Milke’s life, conduct timely investigations, and prepare for trial.[143] The Milke record also reflected a lack of exculpatory evidence, making it the type of case that the NIJ Study described as “particularly difficult to defend,” when there is “no physical evidence to test.”[144] Milke was not accused of being present at the scene of the crime, and there was neither any physical evidence linking her to the crime,[145] nor any physical evidence that could exonerate her.[146]

A fourth risk factor in Milke’s case was Arizona’s “death penalty culture.” The NIJ Study sought to measure the “punitiveness” of the legal culture of each state,[147] based on the hypothesis that a high level of state punitiveness “could contribute to more state actors assuming the defendant’s guilt,” and could lead state agents to “overlook or under-value evidence that contradicts assumptions of guilt.[148] The NIJ Study found that “defendants charged in jurisdictions with a strong attachment to the death penalty ha[ve] a greater chance of erroneous conviction than those” in other locations.[149] The NIJ Study used two types of execution data to assess a state’s death penalty culture,[150] and other data point to a comparatively strong death penalty culture in Arizona during the era of Milke’s trial and today. The state ranks eleventh in the total number of executions in the post-1976 era.[151] Currently, there are 118 people on Arizona’s death row, including 79 people from the county in which Milke was convicted.[152] This death row is now the eighth largest in the country.[153] Only nine of the 32 death penalty states conducted executions in 2013, and Arizona was one of them;[154] the state was one of only seven states that executed people in 2014.[155] At the time of Milke’s trial in 1990, no executions had taken place in the state since 1962.[156] Between 1980 and 1990, 108 death sentences were imposed in Arizona, and 157 were imposed between 1991 and 2013.[157] Arizona executed 22 people between 1992 and 2000, and 14 people between 2007 and 2013.[158]

A fifth risk factor for wrongful conviction in Milke was a weak prosecution case on the evidence.[159] The identification of this risk factor contradicted one of the hypotheses of the NIJ Study, namely that weaker evidence would be associated with the wrongful prosecutions that were halted, whereas prosecutions with stronger evidence would move forward to become wrongful convictions.[160] Instead, the data produced the opposite “counter-intuitive” finding,[161] and qualitative assessments of wrongful conviction cases helped to explain this result.[162]

The failure of prosecutors to stop prosecutions with weak evidence from moving forward has been attributed to the influence of the cognitive process known as “the escalation of commitment” or “tunnel vision.”[163] One consequence of these modes of thinking is that when a police officer or a prosecutor commits an error, thereby bringing an innocent person into the criminal justice system, this initial error may not be corrected and may even be compounded with additional errors.[164] No “rigorous testing of evidence” operates to correct the error because such testing has been dismantled, in effect, by the tunnel vision that begins to operate when a police officer or a prosecutor becomes convinced of a suspect’s guilt.[165]

The escalation of commitment occurs when resources, such as “money, time, and emotions,” are invested in “a narrative involving a suspect,” so that police and prosecutors become “less willing or able to process negative feedback that refutes their conclusions.”[166] They lose their motivation to uncover “weaknesses in the initial inculpatory evidence.”[167] Instead, they “want to devote additional resources in order to recoup their original investment.”[168] They become “so committed to proving the defendant’s guilt” that their tunnel vision leads them to ignore or devalue evidence “that points away from a suspect” and to overlook their own errors.[169] The NIJ Study recognized that even though tunnel vision is a risk factor that cannot be quantified, “many case studies of wrongful convictions show that errors attributable to tunnel vision are real and have grievous consequences.”[170]

The Milke record illustrates how the detective and the prosecutor may have become “so attached” to the prosecution of Milke that they failed to recognize the weakness of the evidence against her because of their tunnel vision as they went forward with the prosecution.[171] For example, according to the NIJ Study, one of the common cognitive errors made by police officers is their assumption that they can “become human lie detectors” and “accurately distinguish between truthful and false denials of guilt.”[172] The Milke detective testified that “he [knew] when someone is lying based on a ‘gut feeling.’”[173] But studies “have repeatedly shown” that “most people accurately make these types of judgments at rates no better than the flip of a coin,” and other studies “have suggested that police interrogators themselves” are no more accurate than other people.[174]

Similarly, even though the body language and behavior of suspects is “not a reliable indicator of guilt or deception,” when the detective interrogated Milke and she sobbed without tears, he saw this as evidence of her guilt.[175] The detective “had no training” in interrogation techniques[176] and he had a history of engaging in legally incompetent interrogations.[177] Under these circumstances, he could have been blind to his own cognitive errors and to the influence of tunnel vision regarding the pursuit of Milke.

The Milke record also supports the inference that the effects of tunnel vision may have influenced the prosecutor. The Ninth Circuit’s sarcastic rendition of the detective’s testimony concerning Milke’s statements implicitly echoes the opinions of Milke’s expert witness regarding some of the detective’s “wildly implausible” and “ludicrous” statements.[178] Yet the Milke prosecutor was willing to rely on the detective’s testimony, notwithstanding its dubious content and the impeachable character of the detective. This reliance suggested that the prosecutor accepted Milke’s guilt as a given, and after the Ninth Circuit decision, he recalled “that the facts were quite strong against Milke and there never was a question in his mind that she wasn’t guilty.”[179]

Yet the prosecutor’s Brady violations may suggest that there was a question in his mind whether Milke could be convicted without suppression of the impeachment evidence in the suspension report and the court records. As the NIJ Study observed, “weak facts may encourage prosecutors to engage in certain behaviors designed to bolster the case,” such as Brady violations.[180] If the prosecutor believed in Milke’s guilt, this belief would lead him to devalue the impeachment evidence because it casted doubt upon the detective’s story of the confession, and supported Milke’s claim of innocence. In this instance, the prosecutor’s escalated commitment to proving Milke’s guilt would have produced the tunnel vision that led him to devalue not only the significance of that impeachment evidence, but also the seriousness of his constitutional duty to disclose it to Milke’s counsel.

One of the qualitative risk factors identified as relevant in some NIJ cases was the behavior of a trial judge who failed to protect the defendant from a wrongful conviction.[181] Such a judge chooses not to take active steps to “level the field between prosecution and defense.”[182] Even though the Ninth Circuit found that Milke’s testimony “differ[ed] substantially” from the detective’s testimony,[183] the trial judge opined that their testimony was “substantially the same.” [184] Her mistaken interpretation of the court records led her to ignore the prosecutor’s Brady violations regarding this impeachment evidence. The judge’s attitude toward Milke’s argument is reflected in the conclusion of her order denying the post-conviction petition:

Notwithstanding the [defendant’s] inflammatory adjectives used to describe [the detective] and his conduct (over zealous, rampant overreaching, abusive interrogation techniques, coercive interrogation methods, manipulation of evidence, lying under oath, spurious practices, and oppressive and unprofessional interrogation tactics), the jury heard his version of the interview and they [sic] heard the defendant’s version of the interview. The jury decided the case.[185]

Although the NIJ Study found that judicial error appeared to be rare in its sample of wrongful conviction cases,[186] the Milke judge’s legal errors, both at trial and in the post-conviction proceedings,[187] were startling in the breadth of misapprehension of the law that they exhibited.[188] At Milke’s next trial, it may be hoped that the new judge will recognize the importance of serving “[the] gate-keeper function to prevent injustices.”[189] Otherwise, as in the first Milke prosecution, the “mistakes made . . . by police, prosecutors, . . . or defense attorneys” may be compounded by the judge’s failure to serve as a safeguard against the risks of wrongful conviction.[190]

The Next Milke Trial: Reflections on the Continuing Risks of Wrongful Conviction

It is unknown how many of the 1,389 people executed since 1976[191] were sentenced to death at trials in which the prosecution suppressed favorable Brady evidence.[192] Even with the benefits of the herculean effort of the post-conviction researchers and the Ninth Circuit ruling in her favor, Milke remains at risk for another death sentence at her next trial for two reasons. First, even assuming that the prosecutor does not engage in Brady violations, other risk factors for wrongful conviction will continue to exist. For example, the same prosecution case will still be weak, as evidenced by the Ninth Circuit’s repeated observations that, aside from the detective’s testimony, “[t]here were no other witnesses or direct evidence linking Milke to the crime.”[193] Additionally, even if competent counsel represents Milke, the risk factor of a weak defense case will exist as long as there is a lack of exculpatory evidence. Finally, the factor of strong support for the death penalty will continue to exist in Arizona,[194] even though public support for the death penalty is lower today in national opinion polls than it was in the 1990s.[195]

The second reason for Milke’s peril is that the Ninth Circuit’s decision is not self-enforcing in state courts. Even though the Ninth Circuit affirmed Milke’s constitutional entitlement to the disclosure of the Giglio evidence and to its use for impeachment purposes at trial,[196] the new trial judge may find fault with those federal holdings, just as the prior trial judge ignored “long-standing Supreme Court caselaw” interpreting Brady.[197] As the pre-trial proceedings have played out upon the public stage, the current Milke prosecutor has never expressed regret about his predecessor’s Brady misconduct, or concern about his predecessor’s reliance on the detective with the known history of performing unconstitutional interrogations and lying under oath.[198] Instead, he has challenged the Ninth Circuit’s ruling repeatedly in the media, describing its view of the Milke record as “irresponsible and incorrect,” and expressing confidence that he would “win a second conviction.”[199]

The prosecutor has gone even further in court, arguing to the trial judge that the confession should be admissible because “there are things in the Ninth Circuit ruling that are false,” and that “[y]ou don’t have to accept a Ninth Circuit ruling if they [sic] are wrong.”[200] On that occasion, the judge rebuffed this assertion, observing that the Ninth Circuit opinion will be “the law of the case.”[201] Similarly, the prosecutor has taken the position that none of the Giglio evidence is admissible, referring to that evidence as “repeated attempts to sully [the detective’s] reputation.”[202] Milke’s counsel has criticized that position as unsupportable, given the Ninth Circuit’s explicit holding that exclusion of the Giglio evidence would have violated Due Process.[203] The trial judge has chosen to describe the prosecutor’s position as an “appropriate legal strategy based on good faith arguments of existing law.”[204]

The fact that Milke must prepare for a new trial and face the risk of wrongful conviction again is a reminder of the limits of judicial intervention to challenge Brady resistance—and how that intervention can come too late. When government officials have custody of contested Brady evidence, courts and defendants are dependent upon those officials to preserve that evidence during the years that it may take to litigate a Brady violation claim. The Ninth Circuit repeatedly expressed concern about the missing records in the detective’s personnel file.[205] Specifically, the Milke opinion directed the district court to order the prosecutor to provide the detective’s “records covering all his years of service,” and to submit a statement “under oath from a relevant police official certifying that all of the records have been disclosed and none has been omitted, lost or destroyed.”[206]

This certification process revealed that the missing records actually had been destroyed, because the City “only retains personnel files for five years” after an employee retires.[207] The records presumably were purged sometime after July 1995, but the certifying officials could not identify the date this occurred.[208] An exception to the City’s policy is made when a litigation “hold” has been issued for a particular file,[209] but apparently no such hold was implemented for the file subpoenaed by Milke’s counsel. Thus, the prosecutor obtained yet another benefit from Brady resistance during the ten years before a disclosure order was obtained from the federal district court. By that time, even though the suspension report and two annual reviews remained in the file, the other records were long gone.

The Brady disclosure issue may have faded from view, but the Brady implementation issue remains at center stage. The prosecutor’s strategy of blocking the admission of the impeachment evidence[210] presupposes the availability of the detective to testify at the next trial. However, the detective now seeks to invoke the Fifth Amendment because of his fear of prosecution for civil rights violations.[211] The trial judge initially concluded that the detective was entitled to make a blanket assertion of the privilege,[212] and that he had shown “a reasonable apprehension of danger” of future prosecution for perjury as well as a continuing conspiracy to violate civil rights.[213] In the judge’s view, the declination letters of the U.S. Attorney and Department of Justice leave the door open to future prosecution based on the evidence in the court records of the detective’s misconduct.[214] The prosecutor obtained a reversal of that ruling from the Arizona Court of Appeals, based on the theory that the risk of prosecution is not “real and appreciable.”[215] An appeal from that decision is pending.[216]

There is no doubt that the herculean effort of Milke’s volunteers ultimately proved to be a “game-changer.”[217] They accomplished for Milke the disclosure of favorable evidence that was impossible for her to obtain from the prosecutor or the courts. Their discoveries of the court records laid the foundation for the Ninth Circuit’s repudiation of the prosecutor’s Brady resistance, and made it possible for Milke to have evidence to present in defense at the next trial. Without Milke’s possession of the court records, the detective would not be seeking Fifth Amendment shelter and the prosecutor would not be doing battle against his star witness. Even so, Milke’s possession of the Giglio evidence does not mean that the implementation of her Brady right will be less challenging in the next trial than it was in the first, especially considering the continued existence of the risk factors for wrongful conviction.

Despite initial skepticism toward innocence claims during the early years of the exoneration era, today “police and prosecutors appear to be taking increasingly active roles in reinvestigating possible false convictions, and to be more responsive to claims of innocence from convicted defendants.”[218] According to one study, 38% of known exonerations in 2013 “were obtained at the initiative or with the cooperation of law enforcement.”[219] Yet while some officials have taken on this new responsibility, others, like the Milke prosecutor, continue to oppose the disclosure of evidence to defendants[220] and defend convictions based on the testimony of witnesses like the Milke detective. One police witness can play a powerful role in contributing to many convictions, and when undisclosed Giglio evidence later comes to light, retrospective doubts about the reliability of that witness can infect many prosecutions.

Milke’s case is contemporary proof of the danger of wrongful conviction that can arise because of Brady violations, especially when such Brady violations occur in a prosecution that is imbued with other risk factors. As long as localized resistance to Brady remains an acceptable legal norm for prosecutors and judges alike, the enforcement of Brady will remain a matter of geographic justice, and some prosecutors will continue to operate in a Brady-free zone of their own making.*


∞ Geoffrey C. Bible & Murray H. Bring Professor of Constitutional Law at Tulane University School of Law. The author wishes to thank Josephine Vanderhorst and Xiaomei Li (Tulane J.D. 2015), Elizabeth Houts (Tulane J.D. 2016), and Annette Allard (Tulane LL.M. 2015), for their helpful research assistance. This article also reflects the benefits of inspiring conversations about Brady with Cynthia Alkon and members of the Tulane Criminal Law & Procedure Scholars group, including Anna VanCleave, Jancy Hoeffel, Pamela Metzger, Tania Tetlow, and Becki Kondkar. This article is dedicated to our colleague Claire Dickerson with gratitude for her constant encouragement regarding our scholarship.

[1].Milke v. Ryan, 711 F.3d 998, 1002–03 (9th Cir. 2013).

[2].See Jon B. Gould, Julia Carrano, Richard Leo & Joseph Young, Predicting Erroneous Convictions: A Social Sciences Approach to Miscarriages of Justice 89–93 (2012), (discussing correlation between violations of Brady v. Maryland, 373 U.S. 83 (1963), and wrongful convictions). See generally The Innocence List, Death Penalty Info. Ctr., (last visited Aug. 30, 2014) (listing 143 death row exonerations since 1973); DNA Exonerations Nationwide, Innocence Project, (last visited Aug. 8, 2014) (listing 317 post-conviction DNA exonerations since 1989, and noting that 18 of the 317 served time on death row and 16 were charged with capital crimes but did not receive death sentences).

[3]. Milke, 711 F.3d at 1001, 1003 (referring to the State’s violations of Brady and its progeny, Giglio v. United States, 405 U.S. 150 (1972)).

[4].See Gould, Carrano, Leo & Young, supra note 2, at 38–39 (describing the criteria for “erroneous convictions” as compared to “near misses”). The Study refers to a halted wrongful prosecution as a “near miss.” Id. at xiv.

[5].See id. at 31–37.

[6].Milke, 711 F.3d at 1012 (quoting Kyles v. Whitley, 514 U.S. 419, 438 (1995)).

[7].See Gould, Carrano, Leo & Young, supra note 2, at 67–68 (noting that lying by a non-eyewitness and withholding prosecution evidence are each positive risk factors); id. at 84–85 (most cases in the study involve more than one error, especially wrongful conviction cases, which usually involve a combination of errors that go unchecked).

[8].See infra Part IV; Gould, Carrano, Leo & Young, supra note 2, at 57–88 (summarizing relevant quantitative and qualitative factors).

[9].Milke, 711 F.3d at 1010 (citing Crane v. Kentucky, 476 U.S. 683, 690 (1986)).

[10]. 405 U.S. 150, 154–55 (1992) (applying Brady’s disclosure duty to impeachment evidence).

[11].See Milke, 711 F.3d at 1000–01, 1003.

[12].See id. at 1002.


[14].Id. (“The jury had no independent way of verifying these divergent accounts. [The detective] didn’t record his interrogation . . . bring a tape recorder to the interview . . . ask anyone to witness the interrogation . . . [or] hav[e] Milke sign a Miranda waiver.”); Opening Brief for Petitioner/Appellant at 16, Milke v. Schriro, No. 07-99001 (9th Cir. Dec. 12, 2007) (available from counsel) [Milke’s Brief] ([T]he detective “knew that he would receive the benefit of any doubt” because of “his long tenure as a police officer and confidence in his ability to sway a jury”).

[15].See, e.g., Answering Brief for Respondents-Appellees, Milke v. Schriro, No. 07-99001 (9th Cir. Mar. 3, 2008), 2008 WL 2196481, at 15–31 [State’s Brief] (responding to defendant’s Due Process arguments without mentioning Brady claim or doctrine).

[16].See Milke’s Brief, supra note 14, at 67–77 (summarizing report of interrogations expert in federal habeas record, who found confession narrative to be implausible and too untrustworthy to support a conviction, and concluded that a combination of risk factors “strongly raise the possibility” of wrongful conviction).

[17].See Gould, Carrano, Leo & Young, supra note 2, at 1 (citing Richard A. Leo, Re-thinking the Study of Miscarriages of Justice: Developing a Criminology of Wrongful Convictions, 21 J. Contemp. Crim. Just. 201 (2005)).

[18].See Laura Sullivan, Exonerations on the Rise, and Not Just Because of DNA, NPR: National Public Radio (Feb. 4, 2014, 3:47 AM), (“[In a] ‘sea change from just ten years ago,’ ‘[m]ore than 30 percent’ of the 87 exonerations in 2013 ‘occurred because law enforcement agencies reopened a long-closed case or handed over their records to someone else who wanted to take a look.’”).

[19].The National Registry of Exonerations,
exoneration/Pages/browse.aspx (last visited Aug. 11, 2014).

[20].See Milke’s Brief, supra note 14, at 23, 96–98.

[21].See The National Registry of Exonerations, Exonerations in 2013: The National Registry of Exonerations 8 (2014),
Documents/Exonerations_in_2013_Report.pdf (“[For exonerations between January 1989 and December 2013,] 28% were cleared at least in part with the help of DNA evidence (363/1281) [and] 72% were cleared without DNA evidence (918/1281).”).

[22].See Gould, Carrano, Leo & Young, supra note 2, at 3 (citing Samuel R. Gross, Kristen Jacoby, Daniel J. Matheson, Nicholas Montgomery & Suijata Patil, Exonerations in the United States 1989 through 2003, 95 J. Crim. L. & Criminology 523, 551 (2005)).

[23].Milke’s brief, supra note 14, at 3 (post conviction relief denied in 1996). Between 1989 and 1996, when Milke’s state post-conviction petition was denied, there were 237 total exonerations, including 51 DNA exonerations. Exonerations by Year: DNA and Non-DNA, The National Registry of Exonerations,
Exoneration-by-Year.aspx (last visited Aug. 11, 2014).

[24].Milke v. Ryan, 711 F.3d 998, 1002 (9th Cir. 2013) (“Milke has always denied involvement in the murder[.]”).

[25].See Gould, Carrano, Leo & Young, supra note 2, at 2 (noting the establishment of “more than 70 non-profit innocence projects” at law schools and six state innocence commissions, as well as the enactment of 40 state statutes “to facilitate inmate access to biological evidence for post-conviction DNA testing”).

[26].See, e.g., Samuel R. Gross & Michael Shaffer, National Registry of Exonerations, Exonerations in the United States, 1989–2012 66 (2012), (“[M]ost common serious form of official misconduct [in exoneration cases] is concealing exculpatory evidence from the defendant and the court.”); Randall Sims, The Dawn of New Discovery Rules, Tex. Dist. & Cnty. Attorneys Ass’n (July–Aug. 2013), (last visited Jan. 25, 2014) ([U]nder the Michael Morton Act, mandatory discovery required for “offense reports, witness statements, and all material evidence.”); Olivia Messer, Perry Signs Michael Morton Act, Tex. Observer (May 16, 2013, 8:14 AM), (Morton’s DNA exoneration inspired legislation to curb prosecutor’s suppression of exculpatory evidence).

[27].See, e.g., FRONTLINE: The Confessions (PBS television broadcast Nov. 9, 2010), available at (interviews describing willingness of jurors to ignore contradictions between physical evidence and false confessions by “Norfolk Four” defendants); Gould, Carrano, Leo & Young, supra note 2, at 9–15 (summarizing research related to causes of false confession).

[28].See, e.g., Frances Robles & N.R. Kleinfield, Review of 50 Brooklyn Murder Cases Ordered, N.Y. Times, May 12, 2013, at A1, available at (describing investigation of all guilty verdict cases involving detective whose testimony provided confessions in some cases “from suspects who later said they told him nothing”); Gross & Shaffer, supra note 26, at 73, 80–84 (describing 13 scandals in which over 1,170 people were framed by police officers and exonerated in groups when police corruption was discovered).

[29].See Gould, Carrano, Leo & Young, supra note 2, at 7; infra Part IV.

[30]. Ellen Yaroshevsky & Bruce A. Green, Prosecutors’ Ethics in Context: Influences on Prosecutorial Disclosure, in Lawyers in Practice: Ethical Decision Making in Practice 269, 279, 275–89 (Leslie C. Levin & Lynn Mather eds., 2012) (identifying variety of factors that “influence prosecutors to take liberal or narrow views” of disclosure obligations).

[31].See, e.g., Ellen Yaroshevsky, New Orleans Prosecutorial Disclosure in Practice After Connick v. Thomson, 25 Geo. J. Legal Ethics 913, 916, 939–40 (2012) (contrasting the “history of noncompliance with Brady obligations” in the Orleans Parish prosecutor’s office with open-file disclosure statutes in Ohio and North Carolina).

[32].See Robert Mosteller, Exculpatory Evidence, Ethics, and the Road to the Disbarment of Mike Nifong: The Critical Importance of Full Open File Discovery, 15 Geo. Mason L. Rev. 257, 310 (2008).

[33].Yaroshevsky & Green, supra note 30, at 279.


[35].Id. at 274, 280.

[36].See id. at 281 (for example, resisting disclosure of “arguably false statements by police,” or of evidence relating to “police internal investigations,” or evidence that does “not in itself prove the defendant’s innocence”).

[37].See, e.g., Yaroshevsky, supra note 31, at 921–29 (describing the Orleans Parish prosecutor’s office).

[38].See States With and Without the Death Penalty, Death Penalty Info. Ctr., (last visited Jan. 25, 2014) (Eighteen states do not use the death penalty, and six states abolished it between 2007 and 2013: Connecticut, Maryland, New Jersey, New York, New Mexico, and Illinois).

[39].See Gould, Carrano, Leo & Young, supra note 2, at 121–24 (state-by-state listing of wrongful convictions and wrongful prosecutions in the NIJ Study).

[40].See, e.g., Stephanos Bibas, Brady v. Maryland: From Adversarial Gamesmanship Toward the Search for Innocence?, in Criminal Procedure Stories 129, 154 (Carol S. Steiker ed., 2006) (“Judges are too weak, prosecutors are too partisan, enforcement is too difficult, discovery is too limited, and plea bargains are too widespread for Brady to influence many cases.”); Steven D. Benjamin, Brady at 50, 37-MAY Champion 4, 4 (2013) (“Many fine prosecutors are scrupulous about the production of exculpatory information . . . [but Brady] has been a failure because [m]ost prosecutors [lack] the training and motivation to discover and produce the information that might help the defendant they are trying to convict.”); Bennett L. Gershman, Litigating Brady v. Maryland: Games Prosecutors Play, 57 Case W. Res. L. Rev. 531, 531 (2007) (“Prosecutors have violated [Brady’s] principles so often that it stands more as a landmark to prosecutorial indifference and abuse than a hallmark of justice.”); Norman L. Reimer, A Half-Century Struggle for Fairness, 37-MAY Champion 7 (2013) (“[T]he 50th anniversary of Brady [is] less a celebration than a lamentation.”); Scott Sundby, Fallen Superheroes and Constitutional Mirages: The Tale of Brady v. Maryland, 33 McGeorge L. Rev. 643, 644 (2002) (observing that the Court’s “development of Brady’s holding destined the doctrine to become less of a pre-trial discovery right and more of a post-trial remedy for prosecutorial and law enforcement misconduct”).

[41].See, e.g., Peter Goldberger, Codifying the Brady Rule, 37-MAY Champion 8, 8–10 (May 2013) (summarizing the reforms proposed in the Fairness in Disclosure of Evidence Act); Michael Serota, Stare Decisis and the Brady Doctrine, 5 Harv. L. & Pol’y Rev. 415, 419–25 (2011) (some courts support the proposed reform of ignoring the Brady “materiality” requirement in the pre-trial context).

[42].Milke v. Ryan, 711 F.3d 998, 1005 (9th Cir. 2013) (noting summary denial of Milke’s post-conviction petition by Arizona Supreme Court); id. at 1006–10 (explaining errors of state trial court in dismissing petition).

[43].Michael Kiefer, Prosecutors Under Scrutiny are Seldom Disciplined, Ariz. Republic, Oct. 28, 2013, at A1, available at
20131027wintory-prosecutor-conduct-day-2.html. This article was in a four-part series that included: When Prosecutors Get Too Close to the Line (Oct. 27, 2013), A Star Prosecutor’s Trial Conduct Challenged (Oct. 29, 2013), and Can the System Curb Prosecutorial Abuses? (Oct. 30, 2013), now published in a collection. See Michael Kiefer, The Gray Area of Prosecutor Conduct, Ni Modo Press (Apr. 21, 2014),

[44].See infra Part III.A.

[45].See infra Part III.B.

[46].Milke, 711 F.3d at 1008.

[47].These judges included the state trial judge who denied the post-conviction petition, the Arizona Supreme Court justices who denied review of the latter ruling, and the federal district court judge who denied the habeas petition. See Milke’s Brief, supra note 14, at 3–4.

[48].Milke, 711 F.3d at 1003–04 (Milke was entitled to disclosure of the impeachment evidence even without a request); id. at 1016 (even an inadvertent failure to disclose would violate Brady); id. at 1012 (favorable evidence includes “any evidence that would tend to call the government’s case into doubt”); id. at 1018 (Brady prejudice requires only a showing of “a reasonable probability of a different result” regarding conviction or sentence if the evidence had been disclosed to the defense).

[49].Id. at 1007.

[50].See id. at 1004 (Defense subpoena requested “all records of any Internal Affairs investigations . . . relating to his technique of methods of interrogation, violations of Miranda rights and/or improprieties during the course of interrogation, if any.”). The defense request occurred after the defense cross-examined the detective at trial. See Milke’s Brief, supra note 14, at 40.

[51].Milke, 711 F.3d at 1010 (Defense post-conviction petition requested evidentiary hearing and discovery of all documents (1) “concerning the evaluation of [the detective’s] performance of his duties,” (2) concerning “investigations or disciplinary actions taken or contemplated against [him],” and (3) assessing his “credibility, strengths and/ weaknesses as a witness and/or possible effects on a judge or jury.”).

[52].See Milke’s Brief, supra note 14, at 2–4 (trial in 1990, state post-conviction petition filed in 1995 and denied in 1996, habeas petition filed in 1998 and request for disclosure of the file granted in 2000).

[53].See Milke, 711 F.3d at 1010.

[54].Id. at 1012.

[55].Id. at 1012 (stating that the detective “leaned into her car, ‘took liberties’ with her, and acted in a manner ‘unbecoming an officer’”).



[58].Id. This last allusion to the detective’s misogyny most likely referred to Milke’s description of his conduct with her when he took her into a closed room, alone, “pulled his chair up to hers so that they were sitting face-to-face,” “leaned forward ‘right in front of’ her,” and “put his hands on her knees.” Milke’s Brief, supra note 14, at 41. This brief also alludes to sexualized comments about Milke that the detective made in two interviews, which appeared neither in his report nor in his testimony. Id. at 58.

[59].Milke, 711 F.3d at 1019.

[60].Id. at 1018.

[61].Id. at 1012–15.

[62].Id. at 1004; id. at 1018 (“[S]uppression of the personnel file and suppression of the court documents run together. Had Milke been given the full run of evaluations in [the] personnel file, she would have found cases [the detective] worked on.”).

[63].Id. at 1010.

[64].Id. at 1010–11 (“The state has never offered an explanation for its failure to produce the remaining reports.”); id. at 1001 (“[E]ven today, some evidence relevant to the detective’s credibility hasn’t been produced, perhaps because it’s been destroyed.”). See infra text accompanying notes 207–09 (describing prosecutor’s later disclosure after Ninth Circuit decision as to what happened to missing reports).

[65].By suppressing the misconduct evidence in the file, as well as the court records, the prosecution also obtained a fourth immediate benefit by depriving Milke of evidence that cast doubt upon the detective’s assertions of Milke’s guilt, which could have supplied a mitigating circumstance for a life sentence. Id. at 1015–16.

[66].See Milke’s Brief, supra note 14, at 2 (noting that trial counsel did not call Milke as a witness during the suppression hearing).

[67].See Milke, 711 F.3d at 1000 (“The trial was, essentially, a swearing contest between Milke and [the detective].”); id. at 1009 (“The issue of [the detective’s] Miranda compliance was strenuously disputed at trial.”).

[68].Id. at 1002. Compare id. at 1009 (The prosecutor challenged Milke’s testimony during cross-examination by implying that she was lying about asking for a lawyer: “You actually didn’t ask for an attorney in reality, did you?” When she repeated her testimony about her request, the prosecutor asked her, “I take it you said that out loud?” During his closing argument, the prosecutor emphasized that the jury should believe the detective’s testimony that Milke never asked for a lawyer because “if [Milke] had requested an attorney[,] [the detective] would have noted it.”).

[69].Id. at 1002. See also, e.g., Minnick v. Mississippi, 498 U.S. 146, 147 (1990) (citing Edwards v. Arizona, 451 U.S. 477, 484–85 (1981)) (“[O]nce the accused requests counsel, officials may not reinitiate questioning ‘until counsel has been made available’ to him.”); Milke, 711 F.3d at 1025 (Kozinski, C.J., concurring) (arguing that the conviction should be set aside “on the separate ground that it relied on an illegally-obtained confession” that should be inadmissible under Miranda v. Arizona, 384 U.S. 436 (1966), and Edwards v. Arizona, 451 U.S. 477 (1981)).

[70].Milke, 711 F.3d at 1002.

[71].See Milke’s Brief, supra note 14, at 16 (the detective fabricated some of Milke’s statements and twisted other statements out of context).

[72].See Milke, 711 F.3d at 1024 (Kozinski, C.J., concurring) (“In effect, [the detective] turned the interrogation room into a black box, leaving us no objectively verifiable proof as to what happened inside.”); Maxwell v. Roe, 628 F.3d 486, 502 (9th Cir. 2010) (using the term modus operandi to refer to the consistent techniques used by a police informant for inventing false confessions of cellmates).

[73].See Milke, 711 F.3d at 1023 (Kozinski, C.J., concurring); Milke’s Brief, supra note 14, at 34.

[74].See Milke, 711 F.3d at 1023 (Kozinski, C.J., concurring); Milke’s Brief, supra note 14, at 20.

[75].See Milke, 711 F.3d at 1023 (Kozinski, C.J., concurring); Milke’s Brief, supra note 14, at 14, 19–20, 33–34.

[76]. See Milke’s Brief, supra note 14, at 31.

[77].See Milke, 711 F.3d at 1024 (Kozinski, C.J., concurring); Milke’s Brief, supra note 14, at 15.

[78].Two Ninth Circuit amicus briefs were filed on behalf of Milke, and both briefs focused on interrogation practices and the importance of recording confessions. See Brief of Amicus Curiae by Arizona Civil Liberties Union in Support of Appellant, Milke v. Schriro, No. 07-99001 (9th Cir. Aug. 5, 2008); Brief of Amicus Curiae by Center on Wrongful Convictions in Support of Appellant, Milke v. Schriro, No. 07-99001 (9th Cir. Aug. 5, 2008).

[79].Milke’s Brief, supra note 14, at 28; see Milke, 711 F.3d at 1010 (Detective’s history of misconduct would support the inference that he “planned, from the outset, to conduct an illegal interrogation by confronting Milke alone, without a tape recorder.”).

[80].Milke’s Brief, supra note 14, at 20, 38.

[81].Id. at 38. At the suppression hearing, defense counsel was allowed to introduce expert testimony. See id. at 35–36 (describing the testimony of defense experts Dr. Fowler and Dr. Fritz). Defense counsel also cross-examined the detective about his interrogation in one prior case. The detective acknowledged that he had continued to interrogate that prior defendant who had invoked his rights to silence and counsel. The detective also opined that the law did not require an officer to stop interrogating after an invocation, but only required that statements after an invocation would be inadmissible. Id. at 34–35 (cross-examination involved interrogation in State v. Runningeagle, 176 Ariz. 59, 859 P.2d 169 (1993)).

[82].See Milke, 711 F.3d at 1024 (Kozinski, C.J., concurring) (noting that “in a quarter century on the Ninth Circuit, [Chief Judge Kozinski] can’t remember another case where the confession and Miranda waiver were proven by nothing but the say-so of a single officer”); Milke’s Brief, supra note 14, at 77 (noting experience of interrogations expert Richard Leo described in his report in the federal habeas record, “In the hundreds of cases I have studied, I have never seen a conviction rest on nothing more than a disputed, undocumented and unsigned confession.”).

[83].Milke, 711 F.3d at 1004, 1006.

[84].Id. at 1003.

[85].Id. at 1006.



[88].Id. at 1005 (noting Milke’s “egregious misconduct” argument to the post-conviction court and reliance on the “discussion of the Brady disclosure obligation” in Donnelly v. DeChristoforo, 416 U.S. 637, 647 (1974)).

[89].Id. at 1006.


[91].Id. (“What happened here is more akin to active concealment [by the prosecutor].”); id. at 1007 (“The report was clearly available to the state and it unquestionably constituted Brady and Giglio evidence of the most egregious kind, yet the state suppressed it for more than a decade.”).

[92].Id. at 1016.

[93].Id. at 1007; id. at 1011 (“The state has not explained why this highly relevant report was not produced before Milke’s trial.”).

[94].Id. at 1003 (referring to one statutory basis that exempts a federal court from observing the required deference to a state court decision under the habeas review standards of the Anti-Terrorism and Effective Death Penalty Act of 1996 (AEDPA)) (quoting 28 U.S.C. § 2254(d)(1) (2012)).

[95].Id. at 1007.

[96].Id. The Ninth Circuit noted that the report was “obviously” relevant to the detective’s credibility. The post-conviction petition required the state judge to make findings regarding his credibility, but they were based “on an unconstitutionally incomplete record,” which was “fatally undermined by the absence of evidence that the state was required by Brady and Giglio to produce.” Id.

[97]. Id. (referring to an alternative statutory basis for lack of deference to a state court decision under AEDPA, 28 U.S.C. § 2254(d)(2)).

[98].Id. at 1012, 1016, 1018–19.

[99].See id. at 1004.

[100].Id. at 1006.

[101].Milke’s Brief, supra note 14, at 28. The collection project occurred during the pre-digital era and required the examination of microfiche records at the office of the clerk of court. See Michael K. Jeanes, Clerk of the Superior Court in Maricopa County, Strategic Plan 2006-2008, at 8, 10, 14, (last visited Oct. 8, 2014) (describing initiative to transform court records into electronic format); Milke’s Brief, supra note 14, at 3 (post-conviction petition filed in November 1995).

[102].See Milke, 711 F.3d at 1018 (describing the time frame as 1982 to 1990). This period roughly coincided with the last 8 years of the detective’s 21-year career. He resigned three months before the Milke trial. Id. at 1018 n.7.

[103].Id. at 1018; Milke’s Brief, supra note 14, at 47.

[104].Id. at 1005, 1008. As the Ninth Circuit noted, the fact that documents reflecting misconduct were “available in the public record” did not “diminish the state’s obligation to produce them.” Id. at 1017. According to the Ninth Circuit’s version of the due diligence rule, the State is exempt from disclosure of exculpatory evidence only when the defense “has enough information to be able to ascertain the supposed Brady material on its own,” but not when the defense lacks sufficient information “to find the Brady material with reasonable diligence.” Id. at 1017–18. Given the fact that Milke’s post-conviction counsel “was able to discover the court documents” only after thousands of hours of research, the court concluded that “[a] reasonably diligent lawyer couldn’t possibly have found these records in time to use them at Milke’s trial.” Id. See Kate Weisburd, Prosecutors Hide, Defendants Seek: The Erosion of Brady through the Defendant Due Diligence Rule, 60 UCLA L. Rev. 138, 153 & n.81 (2012) (observing that the Supreme Court has not addressed the validity of the “due diligence” rule, and providing detailed analysis and critique of the rule in federal circuits). Cf. United States v. Tavera, 719 F.3d 705 (6th Cir. 2013) (abolishing rule).

[105].See Milke’s Brief, supra note 14, at 3.

[106].Milke, 711 F.3d at 1009. For other examples of capital cases with positive outcomes for Brady claims on habeas in federal circuit courts, see Browning v. Trammel, 717 F.3d 593 (10th Cir. 2013); Phillips v. Ornoski, 673 F.3d 1168 (9th Cir. 2012); Maxwell v. Roe, 628 F.3d 486 (9th Cir. 2010); Douglas v. Workman, 560 F.3d 1156 (10th Cir. 2009); D’Ambrosio v. Bagley, 527 F.3d 489 (6th Cir. 2008); Gantt v. Roe, 389 F.3d 908 (9th Cir. 2004); In re Lott, 366 F.3d 431 (6th Cir. 2004); Castleberry v. Brigano, 349 F.3d 286 (6th Cir. 2003); Jamison v. Collins, 291 F.3d 380 (6th Cir. 2002); Benn v. Lambert, 283 F.3d 1040 (9th Cir. 2002); Mitchell v. Gibson, 262 F.3d 1036 (10th Cir. 2001); Paradis v. Arave, 240 F.3d 1169 (9th Cir. 2001); Nickols v. Gibson, 233 F.3d 1261 (10th Cir. 2000).

[107].Milke, 711 F.3d at 1013–15, 1020–22. Note that although the court describes the seven cases as involving two groups of four cases illustrating different types of misconduct, one of the cases, State v. King, belongs to both groups. Id. at 1014.

[108].Id. at 1015.

[109].See id. at 1012–13. In the eleven other cases in the compilation that the Milke opinion does not describe, some defendants entered a plea agreement before a suppression hearing occurred. Most of these other cases involved allegations of the violation of constitutional rights during interrogations, and two cases involved challenges to identification procedures. See Milke’s Brief, supra note 14, at 47–51. Cf. State’s Brief, supra note 15, at 27–28 (discussing the cases in the compilation).

[110].Milke, 711 F.3d at 1014.

[111].Id. at 1013.

[112].Id. at 1004. These cases included State v. Reynolds, CR88-09605 (Ariz. Super. Ct. Feb. 27, 1989) (order Granting Mot. For New Finding of Probable Cause); State v. Rodriguez, CR 161282 (Ariz. Super. Ct. Nov. 20, 1986) (order Granting Mot. For Redetermination of Probable Cause); State v. Rangel, No. CR89-08086 (Ariz. Super Ct. Oct. 16, 1989) (order Granting Mot. To Remand); and State v. King, CR90-00050 (Ariz. Super. Ct. Jun. 22, 1990) (ruling on inadmissibility of statements). See Milke, 711 F.3d at 1013–14.

[113].Id. at 1004. These cases included State v. Yanes, CR-130403 (Ariz. Super. Ct. July 26, 1984) (order Granting Mot. For New Trial) and State v. Yanes, CR-130403 (Ariz. Super. Ct. Nov. 26, 1984) (order Granting Mot. to Suppress); State v. Conde, Nos. CR 88-05881(B), CA 90-475 (Ariz. Super. Ct. Oct. 24, 1989); State v. Conde, 174 Ariz. 30, 846 P.2d 843, 845 (Ariz. Ct. App. 1992); State v. Jones, No. CR90-05217 (Ariz. Super. Ct. Nov. 29, 1990) (order Granting Mot. to Suppress); and State v. King, CR90-00050 (Ariz. Super. Ct. Jun. 22, 1990) (ruling on inadmissibility of statements). See Milke, 711 F.3d at 1014–15.

[114].Milke, 711 F.3d at 1010.

[115].Id. Cf. id. at 1024 (Kozinski, C.J., concurring) (“[G]iven [the detective’s] long history [of misconduct], one wonders how [he] came to interrogate a suspect in a high-profile murder case by himself, without a tape recorder or witness . . . . [Is] this par for [his] Police Department or was [he] called in on his day off because his supervisors knew he could be counted on to bend the rules, even lie convincingly, if that’s what it took to nail down a conviction in a high-profile case?”).

[116].Id. at 1008.

[117].Id. (“[T]he judge “claimed to have reviewed the exhibits[.]”).

[118].Id.; see also id. at 1009 (“[H]ad the state post-conviction judge realized that the documents contained judicial findings of [the detective’s] mendacity and disregard for constitutional rights, she may well have recognized their relevance as impeachment evidence that had not been disclosed as required by Giglio.”).

[119].Id. at 1008 (noting that deference to the state court’s findings could not be given under AEDPA because they were based on an unreasonable determination of facts in failing to consider all the evidence presented).

[120].Id. at 1010.

[121].Id. (citing Crane v. Kentucky, 476 U.S. 683, 690 (1986)).

[122].Id. at 1016–17.

[123].See id. at 1013–19, 1020–22.

[124].Id. at 1018–19.

[125].Id. at 1017.

[126].Id. at 1016.


[128].Id. at 1016–17. Cf. id. at 1024 (Kozinski, C.J., concurring) (“[The] Police Department and the [detective’s] supervisors there should be ashamed of having given free rein to a lawless cop to misbehave again and again, undermining the integrity of the system of justice they were sworn to uphold. As should [the] County Attorney’s Office, which continued to prosecute [his] cases without bothering to disclose his pattern of misconduct.”).

[129].Id. at 1019–20.

[130].See KTAR Newsroom, U.S. Attorney’s Office Won’t Investigate Debra Milke’s Accuser, KTAR News (Sept. 13, 2013, 1:15 PM), See infra text accompanying notes 211–16.

[131].See Debra Milke’s Retrial Moved to 2015, Dare to Think: Be Informed (Sept. 27, 2013),

[132].See Gould, Carrano, Leo & Young, supra note 2, at xv-xvii, 60 (Study examined 260 wrongful convictions and 200 wrongful prosecutions in cases from 1980 to 2012. Each case involved a violent felony crime against a person. 23 factors were tested and 10 factors were found to be correlated uniquely with wrongful convictions as opposed to wrongful prosecutions in the sample); id. at 38-39 (criteria for factual innocence required for cases). Note that Milke’s prosecution may illustrate a sixth risk factor because Milke was 25 years old when convicted, and younger defendants in the Study “were at an increased likelihood of conviction.” Id. at xx. The age of the defendants in the Study ranged from 14 to 76. Id. at 70. Four NIJ risk factors are not illustrated in the Milke record: the defendant’s prior conviction, an intentional misidentification of the defendant by an eyewitness, a forensic evidence error, and the testimony of a family member as a defense witness. Id. at xvii.

[133].The Study relied on an expert panel of criminal justice professionals to analyze a representative sample of 39 of 460 cases for “qualitative” findings. Id. at 72–73 (“Panelists included: two prosecutors, two retired judges, a defense attorney, a police sergeant, a forensic scientist, and [five] researchers on both police and prosecutor practices.”).

[134].Id. at 31. Compare id. at 21 (“Once convicted, innocent defendants often find it extremely arduous to establish their blamelessness . . . .”) with Exonerations in 2013, supra note 21, at 9 (for people exonerated between 1989 and 2013, more than 75% had been in prison for at least 3 years, and 50% for at least 8 years).

[135].Gould, Carrano, Leo & Young, supra note 2, at 84–85. See Exonerations in 2013, supra note 21, at 17–18 (“[F]alse conviction is not one pathology with a single set of contributing risk factors but a set of several different problems with different causal structures depending on the crime[.] Homicide cases . . . include a high rate of official misconduct [58%], and 75% of all false confessions in the database [of exonerations between 1989 and 2013].”).

[136].Milke’s Brief, supra note 14, at 24.

[137].See Gould, Carrano, Leo & Young, supra note 2, at 67–68 (describing “prosecution’s withholding of evidence and lying by a non-eyewitness” as risk factors); Milke v. Ryan, 711 F.3d 998, 1019 (9th Cir. 2013) (“The suppression of evidence of [the detective’s] lies and misconduct thus qualifies as prejudicial for purposes of Brady and Giglio”). Note that the detective now seeks to invoke the Fifth Amendment at Milke’s new trial. See infra text accompanying notes 211–13.

[138].Milke, 711 F.3d at 1012–14, 1020–21.

[139].Milke’s Brief, supra note 14, at 72–73 (describing Professor Leo’s report). Note that Leo is one of the authors of the NIJ Study.

[140].Gould, Carrano, Leo & Young, supra note 2, at 81 (“[T]he strength of the defense case helped to predict case outcome, with erroneous convictions having on average weaker defenses[.]”); id. at 51–56 (describing measure for strength of case using modified version of Police Foundation Rating Scale).

[141].See id. at 82 (qualitative assessment of expert panel).

[142].Milke’s Brief, supra note 14, at 100.

[143].Id. at 100–14 (summarizing counsel’s performance that violated prevailing professional norms and prejudiced Milke as ineffective assistance of counsel under the Sixth Amendment, which counsel conceded; this claim in Milke’s post-conviction and habeas petitions was not addressed by the Ninth Circuit).

[144].Gould, Carrano, Leo & Young, supra note 2, at 80.

[145].Milke v. Ryan, 711 F.3d 998, 1002 (9th Cir. 2013).

[146].See Milke’s Brief, supra note 14, at 23.

[147].Gould, Carrano, Leo & Young, supra note 2, at xviii–xix, 67 n.23.

[148].Id. at xix.

[149].Id. at 89. The NIJ Study noted that “exoneration rates for death sentences are much higher than for other murder convictions and for criminal convictions generally.” Id. at 35 n.6. See Exonerations in 2013, supra note 21, at 8 (“Eight percent of known exonerations occurred in cases in which the defendants were sentenced to death (105/1281). However, since death sentences are a tiny sliver of felony convictions—less than 1/100 of 1%—this reflects a uniquely high rate of exoneration.”). The Study observed that one possible reason is that capital cases “are sometimes afforded greater legal protections” and are often “highly scrutinized by a strong local activist community.” Gould, Carrano, Leo & Young, supra note 2, at 35–36 n.6. Such greater scrutiny “may result in more erroneous convictions being discovered,” and if so, “then the death penalty is not a source or predictor of the occurrence of erroneous convictions.” Id. In the alternative, it may be “that the death penalty is indicative of a local legal culture that is violent and punitive and often historically racist,” and that “[t]hese traits may make police, prosecutors, and the community” in a particular state “more likely to seek conviction despite evidence of innocence.”

[150].The Study measured the state “death penalty culture” variable for bivariate analysis using state executions post-1976 per state population; for the logit regression models, the study used state executions post-1976 per number of state murders. Gould, Carrano, Leo & Young, supra note 2, at 59 n.15.

[151].See Number of Executions by State and Region Since 1976, Death Penalty Info. Ctr., (last visited Aug. 8, 2014) (the top ten states in total executions include Alabama, Florida, Georgia, Missouri, North Carolina, Ohio, Oklahoma, South Carolina, Texas, and Virginia).

[152].See Death Sentences: County Breakdown, Ariz. Dept. of Corrections, (last visited Sept. 2, 2014).

[153].See Death Row Inmates by State and Size of Death Row by Year, Death Penalty Info. Ctr., (last visited Aug. 30, 2014).

[154].Courtney Subramanian, 39 Death Row Executions in 2013, a 10% Drop from Last Year, Time (Dec. 19, 2013) The other states in 2013 included Alabama, Florida, Georgia, Missouri, Ohio, Oklahoma, Texas, and Virginia.

[155].See Execution List 2014, Death Penalty Info. Ctr., (last visited Aug. 21, 2014).

[156].See Arizona Death Penalty History, Ariz. Dept. of Corrections, (last visited Sept. 2, 2014).

[157].See Death Sentences in the United States from 1977 By State and By Year, Death Penalty Info. Ctr., (last visited Sept. 2, 2014). Cf. Exonerations in 2013, supra note 21, at 8 (“[T]he total number of death sentences in the United States—which averaged over 280 a year from 1988 through 1999—has dropped rapidly since 2000 to 80 per year or fewer in the past three years.”).

[158].See Executions by State and Year, Death Penalty Info. Ctr., (last visited Aug. 31, 2014).

[159].See Gould, Carrano, Leo & Young, supra note 2, at 51–56 (describing measure for strength of case using modified version of Police Foundation Rating Scale).

[160].Id. at 37.

[161].Id. at 78; id. at xix (“Discussions with [expert panel] suggested that weak facts may encourage prosecutors to engage in certain behavior designed to bolster [a] case, [such as Brady violations] . . . . Eventually, despite the weak evidence, the players involved become so committed to proving the defendant’s guilt that evidence illustrating the contrary is ignored or discounted.”).

[162].Id. at 38, 67–68.

[163].See id. at xxi, 15–16.

[164].See id. at 85, 86.

[165].Id. at xxi, 15.

[166].Id. at 86.


[168].Id. at 86.

[169].Id. at 86–87. Accord id. at xix.

[170].Id. at 16.

[171].Id. at 80. See infra text accompanying notes 178–80.

[172].Id. at 11.

[173].Milke’s Brief, supra note 14, at 31.

[174].Gould, Carrano, Leo & Young, supra note 2, at 11.

[175].Milke’s Brief, supra note 14, at 73 (describing expert report concerning detective’s reaction). See Milke v. Ryan, 711 F.3d 998, 1001–1002 (2013) (describing detective’s testimony about his reaction).

[176].See Milke’s Brief, supra note 14, at 31 n.24.

[177].See Milke, 711 F.3d at 1013-15, 1020-21.

[178].Compare Milke, 711 F.3d at 1001–02 (2013) with Milke’s Brief, supra note 14, at 72–73.

[179].Crimesider Staff, Debra Milke, Arizona Death Row Inmate, Has Conviction Overturned, CBS News (March 15, 2013, 12:15 PM),

[180].Gould, Carrano, Leo & Young, supra note 2, at xix.

[181].Id. at 83.


[183].Milke, 711 F.3d at 1002.

[184].Milke’s Brief, supra note 14, at 53.

[185].Milke’s Brief, supra note 14, at 63.

[186].Gould, Carrano, Leo & Young, supra note 2, at 84.

[187].See, e.g., Milke, 711 F.3d at 1005–10 (analysis of errors of state trial judge regarding denial of post-conviction petition); Milke’s Brief, supra note 14, at 51–66 (arguments regarding judge’s errors).

[188].One year after the judge issued the order denying the post-conviction petition in Milke, she was transferred from hearing criminal cases to hearing domestic cases as a result of her “questionable decisions.” See Milke’s Brief, supra note 14, at 66 (citing Victoria Harker, Controversial Judge Transferred from Criminal to Domestic Cases, Ariz. Republic, Nov. 8, 1997, available at (last visited November 1, 2013)).

[189].Gould, Carrano, Leo & Young, supra note 2, at 84.

[190]. Id. Notably, the NIJ Study includes several proposals for nationwide reforms to reduce and deter wrongful convictions, Brady violations and false confessions. For example, states or individual prosecutor’s offices could initiate open-file discovery rules, or make it a “standard practice” to videotape police interrogations and interviews, and to “allow defense counsel to request the police records of officers.” Id. at 99–100. For state statutory reforms related to interrogations, see, for example, State v. Dabas, 71 A.3d 814, 828–30 (2013) (when officer destroyed contemporaneous notes from interrogation, court upheld statutory sanction of adverse-inference instruction at trial, directing jury to presume content of notes would have been unfavorable to the State).

[191].Number of Executions by State and Region Since 1976, Death Penalty Info. Ctr., (last visited Sept. 20, 2014).

[192].See, e.g., Gershman, Litigating Brady, supra note 40, at 533 (“[V]iolations of Brady are the most recurring and pervasive of all constitutional procedural violations, with disastrous consequences: innocent people are wrongfully convicted, imprisoned, and even executed.”); Bennett L. Gershman, Reflections on Brady v. Maryland, 47 S. Tex. L. Rev. 685, 688 & n.18 (2006) (citing examples of cases in which Brady violations contributed to the conviction of innocent individuals); Executed But Possibly Innocent, Death Penalty Info. Ctr., (last visited Aug. 16, 2014) (describing investigations that cast doubt on the convictions of ten executed men and providing examples of three posthumous pardons of wrongly executed men and women).

[193].Milke v. Ryan, 711 F.3d 998, 1000 (9th Cir. 2013). Accord at 1002, 1018.

[194].See supra text accompanying notes 147–158.

[195].See Jeffrey M. Jones, U.S. Death Penalty Support Lowest in More than 40 Years, Gallup (Oct. 29, 2013), (60% of the public supports the death penalty for murder, which is the lowest percentage since 1972; the highest support was 80% in 1994).

[196].See Milke, 711 F.3d at 1016.

[197].Id. at 1006.

[198].See generally Mary Ellen Resendez, Maricopa County Attorney Vows New Trial for Debra Milke, ABC, (September 13, 2013, 9:15 PM), (describing prosecutor’s agenda) (last visited September 23, 2014).

[199].See id. Similar confidence in the erroneous character of the Ninth Circuit’s ruling was expressed by the Arizona Attorney General when announcing his decision to seek Supreme Court review and his commitment to argue the case personally. See Press Release, Arizona Attorney General Tom Horne, Press Releases, Statement by Attorney General Tom Horne: “Milke Case Will Be Appealed,” (March 15, 2013), available at (last visited Sept. 3, 2014) (“In my last case, the Supreme Court accepted my argument and overruled the Ninth Circuit’s decision unanimously.”).

[200].Michael Kiefer, Prosecutor in Debra Milke Case Won’t Drop Questionable Confession, The Ariz. Republic, (August 23, 2013, 10:35 PM)
20130823arizona-debra-milke-case-prosecutors-arguments.html (last visited Aug. 15, 2014) (“[T]he prosecutor made it clear that he intends to debate the higher-court ruling and keep the tainted confession in evidence.”).


[202].Defendant’s Motion to Disqualify Maricopa County Attorney’s Office as Prosecuting Agency in This Case at 5 n.2, State v. Milke, No. 1989-01-012631 (July 31, 2013) (quoting the State’s Response to Defendant’s Motion to Set Bond at 2).

[203].Id. at 5; see Milke v. Ryan, 711 F.3d 998, 1010 (2013).

[204].Order Denying Defendant’s Motion to Disqualify Maricopa County Attorney’s Office as Prosecuting Agency in This Case at 4, State v. Milke, No. 1989-01-012631 (August 26, 2013) (Mroz, J.) (“[The State] has taken the position that it will either move to preclude the impeachment evidence . . . at trial or show that the impeachment evidence [w]as [sic] false, misleading, wrong or incomplete . . .”).

[205].Milke, 711 F.3d at 1001, 1010–11, 1018.

[206].Id. at 1011, 1019. The Ninth Circuit also affirmed that, “The state also had an obligation to produce the documents showing [the detective’s] false and misleading statements in court and before grand juries, as well as the documents showing the Fifth Amendment and Fourth Amendment violations he committed during interrogations.” Id. at 1016.

[207].Supplemental Notice of Filing Detective[’s] Phoenix Police Personnel Records Pursuant to This Court’s May 29, 2013 Order, Milke v. Ryan, CV 98-00060-RCB, June 5, 2013, Exhibit E at 1–2.

[208].Id. (The certification of a City Human Resources Officer declared that the records “were likely destroyed in the 1990’s in accordance with the City’s five year records retention schedule,” and it was unknown “how or why there were records turned over to the district court in 2001 from [the] file when all records should have been destroyed at that point.”).

[209].Supplemental Notice of Filing Detective[’s] Phoenix Police Personnel Records Pursuant to This Court’s May 29, 2013 Order, Milke v. Ryan, CV 98-00060-RCB, June 5, 2013, Exhibit F at 2.

[210].See supra text accompanying notes 202–204.

[211].Under Advisement Ruling at 2–3, State v. Milke, No. 189-01-12631 (December 18, 2013) (Mroz, J.) (Mroz Ruling); See Brian Skoloff, Judge Won’t Dismiss Charges in Debra Milke Case, AZ Family, (January 22, 2014, 10:21 AM), (last visited Sept. 2, 2014).

[212].Mroz Ruling, supra note 211, at 6–7.

[213].Mroz Ruling, supra note 211, at 3–6.

[214].Id. at 4–5 (noting that the statute of limitations has not started to run for civil rights violations of some defendants in the misconduct cases analyzed by Ninth Circuit who are serving sentences or appealing convictions). Among the 18 defendants whose court records appeared in Milke’s post-conviction petition, five served prison terms for sentences between 10 and 18 years; two died in prison while serving life sentences; one was executed; two defendants are still serving life sentences; and two remain on death row. Supplement to Defendant’s Response to State’s (Second) Memorandum Regarding Witness Invocation of Fifth Amendment Privilege at 2, State v. Milke, No. 1989-01-012631 (December 13, 2013) Exhibit A, at 1–8.

[215].Petition for Special Action from the Superior Court of in Maricopa County at 3–4, State ex rel. Montgomery v. Mroz, No. CR1989-012631 A (April 17, 2014). See id. at 4 (finding no evidence of a conspiracy and describing federal precedent establishing that the Fifth Amendment may not be invoked based on fear of a perjury prosecution).

[216].See Michael Kiefer, Court Says Ex-Cop Must Testify in Milke Retrial, The Ariz. Republic, (April 17, 2014, 9:34 PM)

18/court-says-ex-cop-must-testify-milke-retrial/7857129/. Similarly, the trial judge’s ruling that the new trial does not violate Milke’s Double Jeopardy right, based on the judge’s finding that the prosecutor did not “knowingly withh[o]ld” evidence in the first prosecution, is also on appeal. See Ryan Owens, Former Death Row Inmate Hopes Double Jeopardy Defense Will Set Her Free, ABC News, (June 25, 2014, 9:28 AM),

[217].See infra Part III.B.

[218].Exonerations in 2013, supra note 21, at 3; see id. at 10 (“Overall, 29% of known exonerations in the United States since 1989 included cooperation by police or prosecutors or both . . . . [T]he number and proportion of such cases appear to be increasing over time.”).

[219].Exonerations in 2013, supra note 21, at 3 n.11.

[220].Notably, eight months after the Ninth Circuit’s Milke decision, the Arizona Supreme Court adopted an amendment to Supreme Court Rules of Court 3.8 that requires prosecutors to disclose evidence regarding wrongful convictions to defense counsel and the presiding judge at trial. AZ: Supreme Court Adopts Wrongful Conviction Provision Despite Montgomery’s Opposition, The Open File, (November 18, 2013), (last visited Sept. 3, 2014). The Milke prosecutor opposed the amendment, urging the Court that it would be “confusing and burdensome,” and that there is “no convincing evidence that Arizona has a ‘problem’ of wrongful convictions.” Id.

* Editor’s Note: After Milke won the appeal described in note 216, supra, all charges were dismissed. See Milke v. Mroz, 236 Ariz. 276, 339 P.3d 659 (Ariz. Ct. App. 2014).