Bell v. Bell

CLAY, Circuit Judge,

dissenting.

In its effort to avoid the conclusion that the state violated Brady v. Maryland, 373 U.S. 83, 87 , 83 S.Ct. 1194 , 10 L.Ed.2d 215 (1963), the majority makes three basic missteps. First, the majority disregards the reasoning of the Supreme Court and blames Petitioner for refusing to discover evidence the prosecution said did not exist. Second, the majority ignores the weight of the evidence and the absence of findings by the district court, and instead holds that no tacit agreement 'existed between William Davenport and the prosecution. Finally, the majority concludes that the prosecution’s withholding of evidence was not prejudicial, thereby implying that because Petitioner made an impeaching argument, it was irrelevant that he was de *238 prived of the evidence to support it. I disagree with the majority on all three points, and therefore respectfully dissent.

I.

The facts recounted by the majority and the Tennessee Court of Appeals are generally correct. Rather than repeat a lengthy recitation of the facts, I will limit this discussion to those aspects insufficiently developed by the majority.

Petitioner’s trial was held on or about the beginning of March, 1987. At trial, the prosecutor, Ronald Miller, called Davenport as a witness against Petitioner. Davenport claimed that Petitioner approached him in jail and asked what “they would do on a double murder.” J.A. at 256. Petitioner allegedly proceeded to discuss the charges in detail, recounting that he had gotten into an argument over his dog with Herman Wallace. Davenport stated that Petitioner admitted that he had been intoxicated, and had ended up shooting Mr. Wallace. Petitioner also allegedly confessed to shooting Jean Wallace, who was Mr. Wallace’s wife, because “she was there.” J.A. at 257. Davenport testified that Petitioner did not seem remorseful about killing the two people.

On cross-examination, Petitioner brought out Davenport’s five felony convictions for charges including forgery, fraud, and receiving and concealing stolen property. Davenport admitted that he was familiar with the criminal justice system, that he was currently incarcerated, and that he had a parole hearing on December 5, 1987. Davenport also admitted that he at one time adhered to the beliefs of the Ku Klux Klan.

During closing, Petitioner argued that Davenport lacked credibility because of his convictions for crimes involving fraud and false dealings, and because of his membership in the Klan. He also contended that Davenport should not be believed because Davenport had

a parole hearing coming up in a matter of months and if [he] can go to the Parole Board and [he] can say, “I have helped to convict Stephen Michael Bell,” that they might cut [him] some slack.... That is why [he] called the District Attorney’s office; that is why [he] spoke to the police and the District Attorney and that is why [he] came to testify.

J.A. at 275.

Miller argued in favor of Davenport’s credibility during closing. According to Miller, Davenport testified because Petitioner “came there and told him what had happened and he had no remorse and no sorrow whatsoever and that is why he [Davenport] was in here testifying and telling you what he heard.” J.A. at 277. Miller also claimed that though “[Petitioner] would have you believe that [Davenport] wants an early parole through our office or through me,” Miller did not “have any say-so with the Parole Board; they are going to let him go soon enough anyway. I have nothing to do with what they do in their own respective realms.” J.A. at 277. The jury convicted Petitioner for the first-degree murder of Mrs. Wallace, and the second-degree murder of Mr. Wallace.

On March 19, 1987, approximately two weeks after Petitioner’s trial, Miller sent a letter to the Board of Pardons and Paroles requesting that Davenport “be considered for parole at the earliest eligible date.” J.A. at 501. Miller stated that “the State did not have a strong case without the testimony of William Davenport” and that “[p]rior to talking to Mr. Davenport, we did not have any idea as to the motive for the killings.” J.A. at 501. According to Miller, early parole was justified on ac *239 count of Davenport’s cooperation and the possibility of retaliation from other prisoners. In June of 1987, the Board granted Davenport parole, apparently in advance of his release eligibility date. 1

After pursing a direct appeal and state post-conviction relief, Petitioner brought a petition for a writ of habeas corpus. The district court concluded that an evidentiary hearing was necessary to resolve certain issues raised by Petitioner, including the Brady issue. The court accordingly held an evidentiary hearing in January of 2004.

At the evidentiary hearing, Miller’s testimony revealed that Davenport had initiated contact with the prosecution by sending a letter to the district attorney that was eventually passed on to Miller. 2 Miller had met with Davenport on October 3, 1986, after receiving Davenport’s letter. Miller’s notes from that meeting (the “Notes”) indicate that Davenport “want[ed] to go to [the] Red Building,” which was a preferred minimum security facility, and that he was interested in work release. The Notes also suggest a discussion of Davenport’s prior convictions, as well as the date that Davenport would be eligible for parole.

Though Miller could not remember many of the specifics of his conversation with Davenport, he acknowledged that Davenport wanted something in exchange for his testimony. Miller denied, however, making any agreements with Davenport. Miller stated that he wrote the letter to the parole board on behalf of Davenport because “he testified at trial against someone I thought was dangerous, and I felt that he would now be labeled as a snitch, and it might be best that I did whatever I could do to get him out of prison.” J.A. at 475. Though Miller stated that the case against Petitioner would have proceeded to trial without Davenport’s testimony, Miller reaffirmed that the state did not have a good case without Davenport.

Ross Alderman, Petitioner’s trial attorney, also testified at the evidentiary hearing. His testimony revealed that, although documents reflecting the disposition of Davenport’s criminal charges were responsive to Petitioner’s Brady request, the prosecution did not provide Petitioner any documents concerning Davenport. 3 Instead, the prosecution claimed that it had provided Petitioner with “all discoverable information in [its] file.” J.A. at 499. Sometime after trial, Petitioner discovered that the same prosecuting office handling Petitioner’s case had nolle prosequied four pending criminal charges against Davenport, which consisted of two counts of grand larceny and two counts of concealing stolen property, after Davenport’s meeting ■with Miller but before his testimony at Petitioner’s trial. On the same date that the charges were nolle prosequied, Davenport pled guilty to writing worthless checks and concealing stolen property, and received prison time to be served concurrently with time he was already serving.

*240 In his post-hearing brief, Petitioner argued that Miller’s interactions with Davenport demonstrated an exchange of favorable treatment for testimony that should have been disclosed to Petitioner. This exchange, Petitioner contended, when combined with the evidence of the Notes , and Davenport’s nolle prosequied charges, entitled Petitioner to relief under Brady . The district court rejected Petitioner’s Brady claim. The court concluded that Miller’s letter to the parole board did not exist at the time of trial and could not be disclosed, and though the Notes should have been turned over to Petitioner, they were not material under Brady . The district court made no findings as to whether a tacit agreement existed between Davenport and Miller, and also failed to analyze Davenport’s nolle prosequied charges.

A panel of this Court reversed, holding that the evidence of a tacit agreement, the Notes, and Davenport’s nolle prosequied charges was improperly suppressed under Brady. Bell v. Bell, 460 F.3d 739, 750-57 (6th Cir.2006). The Court also concluded that this evidence was material, and therefore Petitioner was entitled to relief under Brady. Id. at 757-59 . Respondent moved for rehearing en bane, which this Court granted on December 15, 2006, thereby vacating the panel opinion.

H.

The sole issue is whether Petitioner established a Brady violation. I agree with the majority that we review this issue de novo, and that the Antiterrorism and Effective Death Penalty Act, Pub.L. No. 104-132, 110 Stat. 1214 (1996), does not constrain our review. Majority Op. at 231.

A.

In Brady , the Supreme Court held that “the suppression by the prosecution of evidence favorable to an accused upon request violates due process where the evidence is material either to guilt or to punishment, irrespective of the good faith or bad faith of the prosecution.” 373 U.S. at 87 , 83 S.Ct. 1194 . To prevail on a Brady claim,.a petitioner must establish three elements: (1) “the evidence at issue must be favorable to the accused, either because it is exculpatory, or because it is impeaching;” (2) “that evidence must have been suppressed by the State, either willfully or inadvertently;” and (3) “prejudice must have ensued.” Strickler v. Greene, 527 U.S. 263, 281-82 , 119 S.Ct. 1936 , 144 L.Ed.2d 286 (1999). Exculpatory and impeaching evidence are measured by the same constitutional standard. United States v. Bagley, 473 U.S. 667, 676-77 , 105 S.Ct. 3375 , 87 L.Ed.2d 481 (1985).

Brady represents a limited departure from our adversarial system, which is justified by the overarching concern that the defendant receives a fair trial. Id. at 675 & n. 6, 105 S.Ct. 3375 . The Brady rule recognizes that the state’s interest “in a criminal prosecution is not that it shall win a case, but that justice shall be done.” Kyles v. Whitley, 514 U.S. 419, 439 , 115 S.Ct. 1555 , 131 L.Ed.2d 490 (1995) (quoting Berger v. United States, 295 U.S. 78, 88 , 55 S.Ct. 629 , 79 L.Ed. 1314 (1935)); Strickler, 527 U.S. at 281 , 119 S.Ct. 1936 (noting the “special role played by the American prosecutor in the search for truth in criminal trials”). Requiring the prosecutor to disclose favorable evidence “tend[s] to preserve the criminal trial, as distinct from the prosecutor’s private deliberations, as the chosen forum for ascertaining the truth about criminal accusations.” Kyles, 514 U.S. at 439 , 115 S.Ct. 1555 .

B.

There is no dispute that both the Notes and the prosecutor’s decision to nolle pro- *241 sequi several of Davenport’s pending criminal charges constituted favorable impeaching evidence. Likewise, there is no dispute that the prosecution suppressed the Notes. The majority argues, however, that because evidence reflecting the disposition of Davenport’s criminal charges was available to Petitioner in the public records, the prosecution did not suppress such evidence. For this reason, the majority refuses to consider the disposition of Davenport’s pending criminal charges in its materiality analysis. The majority’s holding is erroneous as a matter of law, and it threatens to reintroduce the gamesmanship that Brady and its progeny were designed to prevent. Petitioner specifically requested impeaching evidence concerning the criminal arrests and convictions of the prosecution’s witnesses, which would have included the disposition of Davenport’s criminal charges. The prosecution responded to that request without alerting Petitioner to the existence of these documents, claiming instead that all discoverable documents had been disclosed, despite the fact that the documents were in the prosecution’s possession. Accordingly, the prosecution suppressed this evidence, notwithstanding its availability to Petitioner through the public records.

Two Supreme Court cases establish that a prosecutor’s false or misleading statement disclaiming the existence of Brady material obviates the need for a petitioner to conduct an independent investigation. Such material, if in the prosecution’s possession, and if not disclosed, is therefore suppressed under Brady , even if it is available through another source. In Strickler , the issue before the Court was whether the petitioner had “cause” for failing to raise his Brady claim before the state trial court, 527 U.S. at 283 , 119 S.Ct. 1936 , which, as the Supreme Court has explained, tracks Brady’s “suppression” element. Banks v. Dretke, 540 U.S. 668, 691 , 124 S.Ct. 1256 , 157 L.Ed.2d 1166 (2004). The warden argued that because facts suggesting the basis for the petitioner’s Brady claim were publicly available, through trial testimony and a newspaper article, the prosecution’s maintenance of an open-file policy that did not include the evidence in question was irrelevant. Strickler, 527 U.S. at 284 , 119 S.Ct. 1936 . The Supreme Court rejected this argument. Though the Court disagreed with the warden’s contention that the factual basis for the petitioner’s claim was publicly available, it did not rely on this fact in crafting the applicable legal standard. 4 Id. at 285 , 119 S.Ct. 1936 . Instead, the Court held that the petitioner established cause because (1) “the prosecution withheld exculpatory evidence,” (2) “petitioner reasonably relied on the prosecution’s open file policy as fulfilling the prosecution’s duty to disclose such evidence,” and (3) the state asserted that petitioner had received “everything known to the government.” Id. at 289 , 119 S.Ct. 1936 .

In Banks , the Supreme Court reaffirmed and extended Strickler , and reversed the Fifth Circuit’s holding that the petitioner could not demonstrate cause because he was not diligent in investigating his Brady claim. 540 U.S. at 695 , 124 S.Ct. 1256 . The Banks Court did not take a position on whether further investigation would have led to the suppressed material; instead, the Court found cause because the prosecution (1) “knew of, but kept back” *242 the Brady material; (2) “asserted ... that it would disclose all Brady material;” and (3) confirmed the petitioner’s reliance on that representation by denying contrary allegations in state habeas proceedings. Id. at 693, 124 S.Ct. 1256 . In rejecting the warden’s argument, Banks clearly indicated that a contrary rule “declaring ‘prosecutor may hide, defendant must seek’ is not tenable in a system constitutionally bound to accord defendants due process.” Id. at 696 , 124 S.Ct. 1256 (framing the state’s argument that the petitioner was not diligent as an argument that “the prosecution can lie and conceal and the prisoner still has the burden to discover the evidence” and rejecting this argument (alternation and internal citation removed)); see also id. at 695 , 124 S.Ct. 1256 (“Our decisions lend no support to the notion that defendants must scavenge for hints of undisclosed Brady material when the prosecution represents that all such material has been disclosed.”). The rule emerging from Strickler and Banks is clear: Where the prosecution makes an affirmative representation that no Brady material exists, but it in fact has Brady material in its possession, the petitioner will not be penalized for failing to discover that material.

The majority’s attempt to distinguish this case from Strickler and Banks on the basis of “absence of reasonable reliance” is utterly unpersuasive. Majority Op. at 236. Petitioner specifically requested that the prosecution provide impeaching evidence concerning its witnesses, which would have included Davenport’s sentencing documents. Accordingly, when the prosecution did not provide any of those sentencing documents and informed Petitioner that it had provided “all discoverable information in [its] file,” J.A. at 499, Petitioner was entitled to “presume that [these] public officials [had] properly discharged their official duties” and that no relevant documents existed. Banks, 540 U.S. at 696 , 124 S.Ct. 1256 (quoting Bracy v. Gramley, 520 U.S. 899, 909 , 117 S.Ct. 1793 , 138 L.Ed.2d 97 (1997)). Under Strickler and Banks , Petitioner’s ability to uncover the disposition of Davenport’s criminal charges by searching public records did not relieve the prosecution of its duty to respond honestly and completely to Petitioner’s discovery request.

The majority’s mischaracterization of Strickler and Banks is even more egregious in light of the courts of appeals’ consistent interpretation of these cases. Several of our sister circuits have recognized that prosecutors cannot knowingly misrepresent that Brady material does not exist without running afoul of Banks and Strickler , see Jennings v. McDonough, 490 F.3d 1230 , 1239 & n. 8 (11 th Cir.2007) (denying Brady claim where the petitioner had equal access to the evidence and “there is no allegation that the prosecution actively misled [the petitioner] about the existence of the [Brady evidence]”); Johnson v. Dretke, 394 F.3d 332, 337 (5th Cir.2004) (“[I]f the State failed under a duty to disclose the evidence, then its location in the public record, in another defendant’s file, is immaterial.”); Gantt v. Roe, 389 F.3d 908, 912-13 (9th Cir.2004) (“While the defense could have been more diligent ... this does not absolve the prosecution of its Brady responsibilities.... Though defense counsel could have conducted his own investigation, he was surely entitled to rely on the prosecution’s representation that it was sharing the fruits of the police investigation.”), as have several separate opinions of this Court. See United States v. Graham, 484 F.3d 413, 422 (6th Cir.2007) (Bat-chelder, J. dissenting) (“[T]he defense is entitled to rely on the prosecution’s representations regarding its compliance with its Brady obligations.”); Bell, 460 F.3d at 767 (panel opinion) (Gibbons, J. dissenting) (“Miller ... did not disclose ... Daven *243 port’s sentencing documents, ... which could have been used for impeachment. Thus, the failure to disclose them was contrary to the requirements of Brady and its progeny.”); 5 cf. Harbison v. Bell, 408 F.3d 823, 833 (6th Cir.2005) (distinguishing Banks because there “the prosecution had repeatedly asserted that all Brady material was disclosed, but nevertheless continued to conceal such material”); Puertas v. Overton, 168 Fed.Appx. 689 , 706 (6th Cir.2006) (unpublished) (“The Supreme Court has repeatedly held that a defendant can reasonably rely on the prosecution’s representation that it has disclosed all Brady evidence, and that once the prosecution confirms this representation, the defendant is under no further duty to investigate additional Brady materials.”).

As the majority notes, Matthews v. Ishee held that “[wjhere ... ‘the factual basis for a claim’ is ‘reasonably available to’ the petitioner or his counsel from another source, the government is under no duty to supply that information to the defense,” even though the government had misled the petitioner by assuring the jury that the material in question did not exist. 486 F.3d 883, 891 (6th Cir.2007) (quoting Strickler, 527 U.S. at 283 n. 24, 119 S.Ct. 1936 ). We are not, however, presently bound by Matthews. Because an examination of Matthews reveals that it misconstrues Banks and Strickler , we should decline to follow it.

First, Matthews misreads the portion of Strickler that purportedly supports its conclusion. Strickler asserted that the prosecution’s open-file policy and refusal to disclose Brady material were factors that established “cause” under Murray v. Carrier, 477 U.S. 478, 488 (1986), because they constituted “conduct attributable to the Commonwealth that impeded trial counsel’s access to the factual basis for making a Brady claim.” Strickler, 527 U.S. at 283 , 119 S.Ct. 1936 . In support of this proposition, Strickler quoted Murray : “[A] showing that the factual or legal basis for a claim was not reasonably available to counsel or that some interference by officials made compliance impracticable, would constitute cause under this standard.” Id. at 283 n. 24, 119 S.Ct. 1936 (emphasis added) (internal citations and quotation marks omitted) (quoting Murray, All U.S. at 488, 106 S.Ct. 2639 ). Clearly Strickler was quoting Murray to establish that “interference by officials” constituted cause — as discussed above, Strickler did not attach legal relevance to the petitioner’s potential inability to discover the Brady material by following up on the public documents. Matthews, however, quotes only the “not reasonably available” example of cause highlighted by Murray , while ignoring the “interference by officials” example actually at issue in Strickler. See 486 F.3d at 891 .

Furthermore, Matthews’ analysis is flawed because it draws a distinction between “cases when the government has sole possession of investigative notes or memoranda” and cases based on public information, id. at 890-91 , notwithstanding Strickler1 s and Banks’ implicit rejection of this distinction. In Banks , the Court disregarded the warden’s argument that the factual basis for the petitioner’s claim was available from other sources, see Banks, 540 U.S. at 694-96 , 124 S.Ct. 1256 ; likewise, in Strickler , the Court declined to *244 condition its test on the petitioner’s inability to locate the Brady material. Instead, the Court crafted a legal rule that focused on whether the prosecutor made false or misleading representations concerning information in its possession. Strickler, 527 U.S. at 289 , 119 S.Ct. 1936 ; Banks, 540 U.S. at 693 , 124 S.Ct. 1256 .

Finally, Matthews is inconsistent with the policies that Brady and its progeny are intended to promote. Forbidding prosecutors from making false representations concerning information within their possession furthers Brady’s goal of fair trials, and is consistent with the “ ‘special role played by the American prosecutor in the search for truth in criminal trials.’ ” Banks, 540 U.S. at 696 , 124 S.Ct. 1256 (quoting Strickler, 527 U.S. at 281 , 119 S.Ct. 1936 ). The contrary rule is antithetical to both of these concerns. The Matthews rule also encourages the type of unsavory gamesmanship that the Supreme Court has thought unfit for litigation that stakes the defendant’s liberty on the jury verdict. See id. (“Prosecutors’ dishonest conduct or unwarranted concealment should attract no judicial approbation.” (citing Kyles, 514 U.S. at 440 , 115 S.Ct. 1555 )). 6

In this case, Miller represented that there was no discoverable information in his file relating to Davenport. This representation was false or misleading, as his office was in possession of Davenport’s sentencing documents. Accordingly, the prosecution suppressed the evidence of its decision to nolle prosequi several of Davenport’s pending criminal charges.

C.

In Giglio v. United States, the Court held that Brady required the disclosure of an agreement between the prosecution and its witness. 405 U.S. 150, 154-55 , 92 S.Ct. 763 , 31 L.Ed.2d 104 (1972). This rule requires the disclosure of “promises of reward” and “inducements” offered to witnesses. Bagley, 473 U.S. at 683-84 , 105 S.Ct. 3375 . The Supreme Court has specifically held that making a promise of reward or inducement contingent on the prosecution’s satisfaction does not exempt it from disclosure. See id. But beyond this general parameter, the Court has provided little guidance on how broadly or narrowly “promises of reward” or “inducements” should be construed, and no specific instructions on how Giglio applies to tacit agreements.

The majority, like several of our sister circuits, does not take issue with the general principle that tacit agreements, like explicit agreements, must be disclosed under Giglio. See, e.g., Wisehart v. Davis, 408 F.3d 321, 323 (7th Cir.2005) (holding that the witness could be impeached by a “tacit understanding that if his testimony [were] helpful to the prosecution, the state would give him a break on some pending criminal charge”); United States v. Shaffer, 789 F.2d 682, 690 (9th Cir.1986) (“[F]acts which imply an agreement ... bear on [the witness’s] credibility and would have to be disclosed.”); cf. Giglio, 405 U.S. at 153 n. 4, 92 S.Ct. 763 (suggesting that disclosure is required where the United States Attorney’s affidavit “contains at least an implication that the Government would reward the cooperation of the witness,” which “tend[ed] to confirm rather than refute the existence of some understanding for leniency”). Reading Giglio and Bagley to require prosecutors to disclose tacit agreements is undoubtedly *245 the correct result, as the same policies justifying the disclosure of explicit agreements also compel the disclosure of tacit agreements. Like explicit agreements, tacit agreements are likely to be relevant to credibility, and therefore should be disclosed to the jury. See Giglio, 405 U.S. at 155 , 92 S.Ct. 763 . Indeed, tacit agreements may be more likely to skew the witness’s testimony. In the case of an explicit agreement, the testifying witness will know what he can expect to receive in exchange for his testimony, and will know the conditions he must fulfill. When a witness is instead led to believe that favorable testimony will be rewarded in some unspecified way, the witness may justifiably expect that the more valuable his testimony, the greater the reward. See R. Michael Cassidy, “Soft Words of Hope:” Giglio, Accomplice Witnesses, and the Problem of Implied Inducements, 98 Nw. U.L.Rev. 1129, 1154 (2004) (“The more uncertain the inducement, the greater the witness’s incentive to tailor his testimony to please the government, precisely because the witness does not know exactly what he will get for his cooperation, and hopes for the very best.”); cf Bagley, 473 U.S. at 683 , 105 S.Ct. 3375 (“The fact that the [witnesses’] stake was not guaranteed through a promise or binding contract, but was expressly contingent on the Government’s satisfaction with the end result, served only to strengthen any incentive to testify falsely in order to secure a conviction.”). 7

The threat of incorrect jury verdicts is further increased by tacit agreements because, when testifying, a witness whose agreement is tacit, rather than explicit, can state that he has not received any promises or benefits in exchange for his testimony. By definition, tacit agreements are not concrete or explicit. See Blacks Law Dictionary 325, 1465 (7th ed.1999) (defining “tacit” as “[i]mplied but not actually expressed; implied by silence or silent acquiescence,” and defining “tacit contract” as “[a] contract in which conduct takes the place of written or spoken words in the offer or acceptance (or both)”). And given the absence of any formal arrangements, a witness’s statement that he has no expectation of favorable treatment concerns only his subjective understanding, and cannot typically be demonstrably falsified. Likewise, the prosecutor can argue to the jury that the witness is testifying disinterestedly, 8 which artificially increases the witness’s credibility — artificially, that is, because the premise of the argument is false. See Cassidy, supra, at 1132 (noting that the prosecutor’s ability to argue that no benefits have been conferred is advantageous). If the tacit agreement is not disclosed, the defendant is left with only argument, not evidence, to attempt to counter the credibility that improperly accrues to the wit *246 ness on account of his supposedly pure motive.

In order to guard against these dangers and facilitate Brady’s ultimate purpose of ensuring fair trials, this Court should hold that two types of evidence are subject to disclosure. First, any evidence that reasonably suggests that the prosecutor conveyed an expectation of favorable treatment to the testifying witness should be disclosed. See Shaffer, 789 F.2d at 690 ; Reutter v. Solem, 888 F.2d 578, 582 (8th Cir.1989) (holding that the state must disclose evidence of the witness’s impending commutation hearing notwithstanding the district court’s finding that no agreement existed between the state and the witness); cf. Giglio, 405 U.S. at 153 , 92 S.Ct. 763 . This expectation creates the incentive for false testimony, and the jury should be allowed to consider this evidence and decide for themselves whether the prosecutor’s conduct affected the witness’s testimony, regardless of whether the witness acknowledges a subjective understanding that he is testifying pursuant to a quid pro quo exchange.

Second, the prosecution should be required to disclose any evidence in its possession that suggests that the witness actually harbors an expectation of favorable treatment, regardless of whether the prosecution created such an expectation. See United States v. Risha, 445 F.3d 298 , 303 n. 5 (3d Cir.2006) (“There can be no dispute that the information in question is favorable to the defense because [the witness’s] expectation of leniency in the state proceedings could have been used to impeach him.”); Todd v. Schomig, 283 F.3d 842, 849 (7th Cir.2002) (“Todd cannot prove an agreement existed. He argues that at the very least [the witness] had an ‘expectation’ of benefit. But what one party might expect from another does not amount to an agreement between them. And Todd does not argue that the state knew of [the witness’s] expectation.” (emphasis added)). Brady’s concern is not with prosecutorial bad faith, but with accurate outcomes. See 373 U.S. at 87 , 83 S.Ct. 1194 ; see also Bagley, 473 U.S. at 675 , 105 S.Ct. 3375 . Even if the prosecution was wholly innocent in creating the witness’s unfounded impression that favorable testimony would be rewarded, that impression nevertheless has the ability to color the witness’s testimony in favor of the prosecution. The prosecution should not be able to knowingly suppress evidence of a witness’s expectation of favorable treatment merely because the prosecution did not willfully cultivate it.

Construing “promises of reward” or “inducements” to include these two types of evidence would promote the disclosure of evidence actually likely to bias prosecution witnesses. In contrast to the rule proposed by the majority, which would require something akin to a formal agreement before any evidence was subject to disclosure, see Majority Op. at 235; see also Shabazz v. Artuz, 336 F.3d 154, 165 (2d Cir.2003) (“The government is free to reward witnesses for their cooperation with favorable treatment in pending criminal cases without disclosing to the defendant its intention to do so, provided that it does not promise anything to the witnesses prior to their testimony.”), this rule would foreclose a crafty prosecutor’s strategy of eschewing a formal agreement, only to achieve the same result through innuendo or implication. Additionally, it would resolve the nebulous issue of determining whether subjective expectations had given rise to a mutual understanding between the prosecution and the witness by making that issue one for the jury. If the prosecution made statements implicitly offering leniency in exchange for testimony, or if the witness made statements implying that he possessed such an expectation, the jury *247 could consider whether an agreement existed, and weigh the witness’s testimony accordingly.

The majority mischaracterizes the rule proposed by the dissent as creating “a new definition of Brady material that includes possible post-trial witness favorable treatment.” Majority Op. at 234. The rule proposed by the dissent does not require pre-trial discovery of post-trial witness treatment. Rather, the dissent simply proposes that we adopt the conclusion compelled by Giglio and Bagley — namely, that the prosecution must disclose evidence of any possible tacit agreements between it and prosecution witnesses prior to trial.

In the case before us, though Miller’s testimony at the evidentiary hearing confirms that no explicit agreements existed, his testimony does not preclude the possibility that he conveyed an expectation of lenient treatment that he intended to fulfill. See J.A. at 475 (“I didn’t promise Davenport anything, and I didn’t make any agreements with him....”). Similarly, the district court made no factual findings on this matter. If such a mutual understanding occurred, then a tacit agreement existed between Davenport and the prosecution, which was subject to disclosure under Giglio, 405 U.S. at 154-55 , 92 S.Ct. 763 .

The evidence in this case strongly suggests that a tacit agreement existed between the prosecution and Davenport. This inquiry involves the choice between two competing hypotheses: Either the prosecution provided favorable treatment to Davenport because of an implied understanding that valuable testimony would be rewarded, or the prosecution never reached an implied understanding with Davenport, and Davenport independently provided valuable testimony, at which point the prosecutor decided to reward him. If the former occurred, the tacit agreement should have been disclosed; if the latter occurred, no Brady violation occurred because no Brady material existed. Compare, e.g., Shaffer, 789 F.2d at 689 , with Todd, 283 F.3d at 849 .

The first hypothesis is significantly more plausible, for two reasons. First, the course of dealings between the parties suggests that a mutual understanding existed. Davenport approached Miller seeking to provide testimony in exchange for favorable treatment. Miller made a notation of the preferential treatment that Davenport sought. Shortly thereafter, and before Davenport testified, the prosecuting office nolle prosequied two counts of grand larceny and two counts of concealing stolen property against Davenport. And approximately two weeks after Davenport provided incriminating testimony and Petitioner was convicted, Miller wrote a letter to the parole board — which accords with the Notes’ indication that Davenport’s parole date was discussed. This course of conduct is unlikely to be the result of coincidence, thus lending support to the contention that an unstated understanding existed between Miller and Davenport.

Second, neither prosecutors nor jailhouse informants are generally disposed to altruistic behavior, and the fact that a testifying jailhouse witness subjects himself to personal danger makes such behavior even less likely to stem from disinterested motivations. See J.A. at 475 (testimony of Miller) (“I felt that [Davenport] would now [after testifying] be labeled as a snitch, and it might be best that I did whatever I could do to get him out of prison.”). Prosecutors are aware of this expectation, and have every incentive to encourage it. See J.A. at 476 (“Everybody wants something, and I’m sure Davenport wanted something.”). The parties’ *248 incentives therefore encourage the creation of tacit agreements.

In sum, the record in this case strongly, though not conclusively, suggests the existence of a tacit agreement. Generally speaking, whether the prosecution and the witness reached an unspoken understanding is a factual matter for the state court or federal district court. Here, Petitioner argued before the district court that the record demonstrated a tacit agreement, but the district court made no factual findings on the matter. In light of this record, the majority errs by unjustifiably resolving this factual matter against Petitioner, thereby expounding an unduly restrictive and incorrect view of tacit agreements as a matter of law. 9

D.

A petitioner can establish prejudice by showing that “there is a reasonably probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different.” Bagley, 473 U.S. at 682 , 105 S.Ct. 3375 . This standard does not require that it is more likely than not that the verdict would have been different without the evidence, and is substantially less demanding than a sufficiency of the evidence test. Kyles, 514 U.S. at 434 , 115 S.Ct. 1555 . The prejudicial effect of any favorable evidence suppressed by the state must be viewed cumulatively, as opposed to item-by-item, and if the suppressed evidence collectively establishes prejudice, no harmless error inquiry is necessary. Id. at 435-36 , 115 S.Ct. 1555 .

Davenport’s testimony was critical to the prosecution. Petitioner was convicted of first-degree murder with respect to Mrs. Wallace, and Davenport provided the only direct evidence of premeditation, which was a necessary element of the offense. See State v. Johnson, 661 S.W.2d 854 , 860 n. 1 (Tenn.1983) (quoting TenmCode. Ann. § 39-2-202 (defining first degree murder), repealed, 1989 Pub. Acts ch. 591, § 1). Davenport’s testimony was also crucial to the prosecution’s second-degree murder case because it provided the only direct evidence that Davenport was the shooter, thereby greatly strengthening what was otherwise a weak and entirely circumstantial case. 10

In light of the state’s insubstantial case, the importance of Davenport’s testimony cannot reasonably be disputed. Miller, in fact, acknowledged that “the state did not have a strong case without the testimony of William Davenport.” J.A. at 501. Miller reaffirmed this statement during habe-as proceedings. If the jury disbelieved Davenport’s testimony, they could easily have acquitted Petitioner on both counts, as the remaining evidence required the jury to make considerable inferential leaps in order to connect the gaps in the state’s case.

Had the Notes and the evidence of the nolle prosequied charges been disclosed by the state, there is a reasonable probability that the jury would have returned a verdict of not guilty on both counts. These pieces of evidence, considered cumulative *249 ly, would have supplied proof of Davenport’s self-interested motive, and a juror considering Davenport’s testimony in light of this evidence might well have concluded that Davenport had seized an opportunity to fabricate a story to serve his own purposes. To recount, the Notes demonstrated that Davenport had approached Miller seeking specific benefits. The Notes also indicated that Miller knew what benefits Davenport sought, and that Miller was aware of Davenport’s parole eligibility. This evidence would have allowed the jury to infer that Davenport was testifying due to an expectation of lenient treatment, which in turn would have provided the jury with a substantial reason to doubt the veracity of Davenport’s testimony. See Shaffer, 789 F.2d at 688-89 (finding materiality where the evidence impeached a witness whose credibility was vital to the outcome of the trial); Reutter, 888 F.2d at 581-82 (finding a reasonable probability of a different outcome where the jury was not informed of the existence of an impending commutation hearing of a key witness).

The nolle prosequied charges were also valuable impeaching evidence. In evaluating this basis for impeaching Davenport’s credibility, the relevant question is not whether the disposition of Davenport’s charges and his offer to testify were actually linked. Instead, the evidence would have proved valuable if the jury could have found that Davenport believed that these events were connected. The jury here could easily have found that Davenport thought that his offer to testify caused the favorable result he received on his pending criminal charges. Davenport would have known that he and Miller discussed his convictions, and that the same prosecuting office had extended Davenport a highly favorable plea arrangement approximately one month later. This evidence would have aided Petitioner’s impeachment efforts in two distinct but mutually reinforcing ways. First, Petitioner could have argued that Davenport’s receipt of favorable treatment from the prosecution motived Davenport to skew his testimony out of gratitude or a sense of obligation. See Wisehart, 408 F.3d at 324 (noting that testimony can be affected by benefits previously bestowed). Second, the receipt of favorable treatment in the past may have amplified Davenport’s expectation that a good performance on the witness stand would lead to lenient treatment, providing Davenport with an increased incentive to shape his testimony to benefit the prosecution. The logic is simple: If the prosecution rewarded Davenport for making efforts to cooperate, it stands to reason that further cooperation will lead to additional rewards. This incentive would be especially strong considering that the benefits that Davenport sought had not yet been bestowed.

The majority’s first error in considering prejudice stems directly from its failure to consider Davenport’s nolle prosequied charges suppressed. Because this evidence is not included in the majority’s calculus, the majority ignores the fact that Petitioner was deprived of the opportunity to argue that Davenport had already received favorable treatment from the prosecution. This error also affects the overall magnitude of the impeaching evidence, which of course must be considered cumulatively. See Kyles, 514 U.S. at 436 , 115 S.Ct. 1555 .

The majority errs again in discrediting the suppression of the Notes because “[t]he jury was apprised of Davenport’s status and the possible other reasons for his decision to testify, namely, that he wished to secure early parole as a result of his participation in the Bell case.” Majority Op. at 237. Here, the flaw in the majority’s reasoning is that it allows argu *250 ment to substitute for evidence. Though the jury was aware that Davenport had an upcoming parole hearing, they were provided with no evidence that Davenport’s parole hearing was connected to his decision to testify. As it was, Petitioner’s claim that Davenport was testifying in order to increase his chances of a favorable result at the parole hearing must have struck the jury as unsubstantiated speculation. Had Petitioner instead proved his claim with reliable evidence provided by the prosecutor himself, there is a reasonable probability that Petitioner’s argument would have convinced the jury, which in turn would have discredited Davenport.

Additionally, the majority ignores two other valuable arguments that Petitioner could have advanced had he been in possession of the Notes and the nolle prose-quied charges. First, as noted above, Miller suggested to the jury that Davenport was motivated to testify by his abhorrence of Petitioner’s lack of sorrow and remorse. Armed with the suppressed evidence, Petitioner could have met this suggestion with powerful evidence that Davenport’s testimony was opportunistic, not altruistic. Second, Miller misleadingly told the jury that he had no “say-so” with the parole board. By demonstrating that Miller had seen fit to record the date of Davenport’s parole eligibility, Petitioner could have mustered some argument that Miller, though lacking formal authority, might be able to exercise some measure of influence. At the very least, the fact that Davenport knew that his parole hearings were discussed with Miller would have provided a basis for arguing that Davenport thought Miller capable of bestowing such a favor on his behalf.

In sum, the suppressed evidence handicapped Petitioner’s efforts to discredit Davenport. Had the evidence been disclosed, there is a reasonable probability that the jury would have doubted Davenport and acquitted Petitioner.

III.

For the foregoing reasons, I respectfully dissent.

. This letter is not in the record, and counsel for Petitioner stated at oral argument that the letter disappeared while in possession of the state, and has never been disclosed to Petitioner.

.Specifically, Petitioner asked for "[t]he nature and substance of any preferential treatment given at any time by any state agent, whether or not in connection to this case to any potential witness” and "[t]he FBI and state arrest and conviction records of each witness the state intends to call in this matter.” J.A. at 441.

. The Supreme Court declined to reach the question of whether a different result would follow if “the defendant [were] aware of the existence of the documents in question and knew, or could reasonably discover, how to obtain them.” Strickler, 527 U.S. at 288 n. 33, 119 S.Ct. 1936 . This question is not raised by the instant case, because Petitioner was not aware of the existence of Davenport’s nolle prosequied charges.

. The authoring judge of the majority opinion originally believed that Davenport’s sentencing documents were suppressed by the prosecution, as indicated by her now-vacated panel dissent. She now argues in the majority opinion herein that the sentencing documents were not suppressed by the prosecution. The fact that the judge authoring the majority opinion herein has taken contradictory positions on this issue certainly detracts from the persuasiveness of the majority opinion.

. The majority also relies on Spirko v. Mitchell, 368 F.3d 603, 611 (6th Cir.2004), but that case is inapposite because it did not involve a false or misleading statement made by the prosecution.

. Empirical evidence also supports the conclusion that implicit promises are as likely to skew a witness’s testimony as explicit promises. See Saul M. Kassin & Karlyn McNall, Police Interrogations and Confessions: Communicating Promises and Threats by Pragmatic Implication, 15 Law & Hum. Behav. 233 , 248 (1991) ("In short, ... our data indicate that because people often process information 'between the lines’, [explicit and implicit promises and threats] are functionally equivalent in their impact.”).

. This danger is not hypothetical. Here, the prosecutor implied that Davenport was motivated by altruistic motives. See J.A. at 277 (Miller’s closing argument) ("Bill Davenport told you that the man [Petitioner] came there [to Davenport in jail] and told him what happened and he had no remorse and no sorrow whatsoever and that is why he [Davenport] was in here testifying and telling you what he heard.” (emphasis added)).

. Even under the majority’s erroneous view of prejudice, the case should be remanded so that the district court can resolve this factual issue. However, because the Notes and Davenport’s nolle prosequied charges establish prejudice under Brady , remand is unnecessary.

. The physical evidence in the case consisted of laboratory results indicating that Petitioner could have fired a gun, and bullets of the same type used in the murder found on Petitioner at the time of his arrest. The state lacked the murder weapon, eyewitnesses to the crime, or physical evidence, such as blood, that connected Petitioner to the victims.