Defendant Carlos Martinez, a prison guard, was convicted of sexual abuse, aggravated sexual abuse, sexual abuse of a ward, and deprivation of civil rights under color of law. The prosecution failed to provide the defense with evidence consisting of a report summarizing an interview with a unit-mate of the victim and a witness in a related case, which strongly suggested that the victim's relationship with Martinez was consensual. Upon learning of the withheld evidence, Martinez, who has not yet been sentenced and who faces a life sentence under the Sentencing Guidelines, moved for a new trial.
BACKGROUND
According to her trial testimony, Maria, who testified under a pseudonym, worked as a cleaner during her incarceration at the Metropolitan Detention Center in Brooklyn, New York ("MDC"), starting in March 2015. Trial Tr. 58, 67-68. Beginning in May, she cleaned areas on the second floor of the MDC's East Building, such as the medical, legal, and lieutenants' areas. Id. at 68-69. Maria was asked to clean multiple times a week, often in the company of another inmate. Id. at 68-70. Sometimes Maria cleaned the area alone, but most of the time she and another inmate were both summoned, and they cleaned *227the second floor together. Id. Whenever she was summoned to clean, she was escorted from her unit to the second floor by a guard, who then left her under the supervision of whoever asked for a cleaner. Id. at 78-79.
Beginning in the fall of 2015, defendant Lieutenant Carlos Martinez started to regularly summon Maria to clean. Id. at 79, 81-82. Martinez usually called for Maria on Fridays, Saturdays, or Sundays, when the second floor was generally empty. Id. at 82. Maria testified that, when she first started cleaning for Martinez, he acted kindly, sharing personal information about himself, including facts about his family and living situation, and referring to Maria by her first name. Id. at 87-88. But Martinez's comments soon became uncomfortable, veering into improper sexual inquiries. See id. at 92-94.
Maria testified that Martinez's misconduct became criminal in December 2015, approximately two to three months after she first started cleaning for him. As usual, he asked Maria to come clean, although this time Maria did not recall if she was accompanied by another inmate. Id. at 100-01. While Maria bent over to retrieve cleaning supplies from a cabinet inside the Lieutenants' Office, Martinez pulled his penis out of his pants, grabbed her head and shoulder, and pushed Maria towards his penis, forcing it into her mouth. Id. at 102-04. Martinez explained to Maria that she would not get in trouble because he could view the feed from all the cameras monitoring the area via a monitor located on the desk in the Lieutenants' Office, so he could make sure no one would walk in on them. Id. at 104. Martinez then pushed Maria, chest-down, on to the desk, pulled her pants down, and penetrated her vagina from behind with his penis. Id. at 105-06. Maria testified that, while this was happening, she asked him to use protection but he did not. Id. at 106. She also testified that she "[d]id [not] want to have sex with him" and "he force[d] [her] to do those things." Id. at 107.
Afterwards, Maria noticed she was bleeding and demanded that Martinez buy her a morning-after pill. Id. at 108. Martinez explained that he had undergone a vasectomy, but eventually relented and agreed to purchase the pill for Maria. Id. at 109-10. He delivered it to her around midnight in her unit, at which point she took it. Id. at 110, 130-36. In the interim, Maria returned to her unit, and told her fellow inmates-Kiara Maldonado, Melva Vasquez, and Danilda Osoria-that Martinez had penetrated her and that she was bleeding. Id. at 115-17, 119-22. Initially, Maldonado responded as if this were a positive event, "[l]ike it was something ... happy ... something good." Id. at 119. Osoria had a similar reaction and joked about the incident, but Vasquez became upset and angry at her unit-mates' lack of sympathy. Id. at 122, 447-48, 560. After Maria began crying and divulged some of the details of what transpired, both Maldonado and Osoria realized something more serious had occurred. Id. at 122-23, 560-62. Maldonado testified that, in the days after the incident, Maria "didn't want to go to work" and "was always in bed," even though she usually socialized with the other inmates. Id. at 640. Osoria similarly testified that, in the weeks following the alleged rape, Maria appeared upset whenever she was called to clean by Martinez. Id. at 567.
Maria testified that, after the December 2015 incident, Martinez would regularly summon her to the office to clean and rape her. Id. at 138-39, 161-62. One of these incidents occurred on a Saturday in either mid-to-late January or early February 2016. See Opp. 13 n. 9, ECF No. 84. Maria was escorted to the second floor to clean *228for Martinez as usual. While Maria was cleaning the bathrooms, she testified that Martinez asked the officer who escorted her to do something in another building in the MDC complex. See Trial Tr. 141-44. The errand would leave Martinez and Maria alone for about twenty minutes. See id. at 717. As soon as the other officer left, Maria testified that Martinez called for her, but she did not immediately respond. Id. at 143. As a result, Martinez came to the door of the bathroom where Maria was cleaning and ordered her to go to the Lieutenants' Office. Id. at 143-45. Once inside the office, Maria testified that Martinez groped her, pushed her down on the desk, pulled her pants down, and penetrated her. Id. at 146-48. The other officer did not return until after Martinez had finished and Maria was back in the bathroom. Id. at 159-60.
Maria also testified to another incident that occurred in January 2016. See id. at 165-80. During this incident, Maria and another inmate, Odis De La Cruz, went to clean the second floor. Id. at 165-66. De La Cruz told Maria that she would clean the area near the Legal Department and told Maria to clean around the Lieutenants' Office. Id. at 171-72. Maria testified that, when she went into the Lieutenant's Office to start cleaning, Martinez again grabbed her, pushed her against the desk, pulled her pants down, and penetrated her. Id. at 177-78. Maria noted that while he was penetrating her, he adjusted the video feed on the computer monitor so that he could see where De La Cruz was cleaning. Id. at 179.
In late January and early February, Maria contacted two friends outside of the prison. First, she called Noel Lopez. Id. at 181. She asked him to visit her and, when he arrived, to kiss her on the lips. Id. at 181, 186, 304-06. Maria testified that Martinez had commented that no one visited her at the MDC, and she hoped that Lopez's visit would demonstrate that she was not alone. Id. at 181. Even though he found the request odd, Lopez kissed Maria when he visited, but the interaction had the opposite effect Maria had hoped for: she testified that Martinez saw the interaction and subsequently forcibly kissed and groped her. Id. at 184-86, 306. But Maria also testified that, after Lopez visited and kissed her, Martinez told her that "he was different" and asked Maria to "get in touch with him" after she was released from prison, even providing her with his email address. Id. at 187-88.
From the MDC, Maria also called Larihelys Vargas, a friend and former coworker. Id. at 289-90. Maria asked Vargas to unfriend her on Facebook and look up Martinez. Id. at 291-92. Even though Maria did not tell Vargas why she wanted her to review Martinez's Facebook page, Vargas checked anyway. Id. at 293. She described photos from Martinez's page to Maria, including photos of Martinez with other women. Id. at 190-91. Vargas also friended Martinez and messaged with him briefly, asking if he was married. Id. at 623-24. While on the phone with Maria, Vargas asked about Martinez. Maria testified to the following exchange: "She [Vargas] would say 'does he kiss you?' I'd go, 'yeah, all the way.' So I went, 'yeah, look him up. Look him up.' ... She said 'all the way?' I said, 'yeah.' And she said 'chiqui, chiqui?' And I said, 'yes.' " Id. at 317 (quotation marks added).
Maria testified that she hoped Martinez would listen to her call with Vargas, become worried that other people knew about what he was doing, and stop abusing her. Id. at 190. Martinez ultimately did listen to the message when it was relayed to him by a fellow officer. Id. at 485. That officer testified that, upon listening to the call, Martinez stated he was "not going to *229bring these women down anymore" because he "[didn't] want problems." Id. at 485.
Nevertheless, after hearing the phone call, Martinez summoned Maria. Id. at 192-94. She testified that he chastised her for the phone call, explaining that there would be an investigation and that she might be held in prison longer because of it or sent to solitary confinement, also known as the Special Housing Unit or SHU. Id. at 196-98. Later, when Maria was interviewed by prison staff about the phone call, she "started to shake," became emotional and cried. Id. at 201, 202, 488. When asked if she liked Martinez, she said no. Id. at 488. When told that she could be sent to the SHU pending an investigation into the phone call, she stated that she did not want any trouble and did not want to be sent to the SHU. Id. at 202, 490. No further inquiry occurred.
The final sexual encounter, according to Maria's testimony, was on April 16, 2016, shortly before she was released from the MDC on April 29, id. at 209-20. Maria testified that, during this last incident, she initially thought she would be accompanied by fellow inmate Odis De La Cruz. Id. at 210-11. At the last moment, she learned that De La Cruz had a visitor and would not join her. Id. She reported to the Lieutenants' Office alone to clean the area and, after the escort officer and another officer left the area, Martinez raped her again. Id. at 214-20.
At trial, the defense advanced various alternative theories to the narrative presented above. Primarily, the defense argued that "Maria is an inherently unreliable opportunist who manipulate[d] her friends." Id. at 1013. The defense highlighted numerous instances when Maria purportedly lied to the prosecutors or other government agencies, or otherwise behaved inconsistently. See, e.g., id. at 1014-15. Additionally, even though various prison officials and guards received training on identifying victims of sexual abuse, none noted that Maria appeared distressed in any of their interactions. See, e.g., id. at 1017-18. The defense's ultimate explanation for these inconsistencies was that she was not raped; rather, "Maria was pursuing Lieutenant Martinez," hoping to initiate a consensual relationship. Id. at 1028-29. By this account, Maria contacted Noel Lopez and asked him to kiss her to make Martinez jealous, id. at 1031, and asked Larihelys Vargas to peruse Martinez's Facebook account because she wanted to learn more about the man she was interested in, id. at 1032-33.
The jury rejected this version of events and, on January 19, 2018, found Martinez guilty of all 20 counts of the indictment, including five counts each of deprivation of civil rights, aggravated sexual abuse, sexual abuse, and sexual abuse of a ward. Id. at 1163-66. The jury specifically found that Martinez caused Maria to engage in sexual acts by placing her in fear. Id. at 1165-66.
On August 22, 2018, defense counsel asked that the prosecutor identify whether she disclosed any interviews with an inmate identified as Jane Doe #4, a testifying victim in a separate trial against another MDC officer. Ricco Decl. 2, ECF No. 82; Opp. 31. The prosecutor responded that she had inadvertently failed to disclose a report summarizing an interview with Jane Doe wherein Jane Doe stated that, in April 2016, Maria mentioned having a relationship with Martinez. Ricco Decl. 2; Opp. 32. Specifically, Jane Doe's account indicated that Maria volunteered to clean the second floor alone because of her relationship with Martinez and declined to have a medical test done because she thought it might reveal the ongoing relationship. See Opp. 32. Based on this *230undisclosed evidence, Martinez moved for a new trial. See ECF No. 81.
LEGAL STANDARD
Defendants may move to "vacate any judgment and grant a new trial if the interest of justice so requires." Fed. R. Crim. P. 33. Such motions may be predicated on a violation of Brady v. Maryland , 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963). See, e.g., United States v. Douglas , 525 F.3d 225, 245 (2d Cir. 2008) ; see also 3 Wright & Welling, Federal Practice and Procedure: Criminal § 586 (4th ed. 2011). A Brady violation occurs when the prosecution suppresses "evidence favorable to an accused ... where the evidence is material either to guilt or to punishment." Brady , 373 U.S. at 87, 83 S.Ct. 1194. A Brady claim involves three elements: (1) "The evidence at issue must be favorable to the accused," (2) "that evidence must have been suppressed by the [prosecution]," and (3) "prejudice must have ensued." Strickler v. Greene , 527 U.S. 263, 281, 119 S.Ct. 1936, 144 L.Ed.2d 286 (1999). Evidence that may be used to impeach a prosecution witness, in addition to evidence that tends to exculpate a defendant, qualifies as "favorable to an accused." See United States v. Bagley , 473 U.S. 667, 676-77, 105 S.Ct. 3375, 87 L.Ed.2d 481 (1985) ; Giglio v. United States , 405 U.S. 150, 154-55, 92 S.Ct. 763, 31 L.Ed.2d 104 (1972). Even if evidence has both incriminatory and exculpatory aspects, it still qualifies as Brady material. See, e.g., United States v. Mahaffy , 693 F.3d 113, 130 (2d Cir. 2012).
The suppression of evidence is prejudicial when that evidence is material, meaning that "there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different." Strickler , 527 U.S. at 280, 119 S.Ct. 1936 (quoting Bagley , 473 U.S. at 682, 105 S.Ct. 3375 ); see also Kyles v. Whitley , 514 U.S. 419, 433-34, 115 S.Ct. 1555, 131 L.Ed.2d 490 (1995). This assessment necessarily implicates the other evidence presented at trial. See United States v. Orena , 145 F.3d 551, 558 (2d Cir. 1998). Where the prosecution's case is "far from overwhelming," see, e.g., United States v. Triumph Capital Grp., Inc. , 544 F.3d 149, 163 (2d Cir. 2008), or where the suppressed evidence accords with the defense's theory and evidence, see United States v. Rivas , 377 F.3d 195, 200 (2d Cir. 2004), the suppressed evidence is more likely to be material. Where impeachment evidence has been suppressed by the prosecution, its materiality depends both on the credibility of the relevant witness and the degree to which the suppressed evidence is cumulative. See, e.g., Wearry v. Cain , --- U.S. ----, 136 S. Ct. 1002, 1006, 194 L.Ed.2d 78 (2016) ; Fuentes v. Griffin , 829 F.3d 233, 251-53 (2d Cir. 2016).
DISCUSSION
The prosecutors here failed to provide defense counsel with a report describing an interview with an individual identified as Jane Doe #4. In April 2017, Jane Doe, an inmate at the MDC at the same time as Maria, was interviewed by an agent of the Department of Justice's Office of the Inspector-General. She stated that she had spoken with Maria about her encounters with Martinez. A memorandum summarizing the interview was drafted and provided to the prosecutors before trial. The relevant portion reads as follows and centers on the April 2016 incident:
Maria told Jane Doe #4 that something was going on between her and Lieutenant Martinez. A few days before Maria left the MDC, she was called to clean the hospital area. Inmate Odis Delacruz [sic] asked Maria if she wanted her to go *231with her to clean. Maria said no. Jane Doe #4 asked Maria if she wanted her to go with her to clean and Maria said no and that she will go alone. Maria went to clean and returned approximately two hours later. A few days after this happened, Jane Doe #4 asked Maria why she did not want Delacruz or her to come along to clean. Maria told her that it was because she was leaving and was having relations with [Martinez]. This was the first time Maria told Jane Doe #4 about this. Maria told Jane Doe #4 that the ... [Bureau of Prisons] medical staff member[ ] asked Maria to have a Pap smear test. Maria did not agree to take the test because she was having relations with Martinez. Maria did not tell Jane Doe #4 how the relations between her and Martinez started.
Opp. 32 (brackets omitted).
This account differs markedly from Maria's. First, Maria testified that Odis did not come with her to clean because she had a last-minute visitor. See Trial Tr. 210-11. Conversely, Jane Doe reported that Maria chose to go to the Lieutenants' Office alone. Second, Maria drew a distinction at trial between having "relations" and being raped. See id. at 318-19. Nevertheless, Maria told Jane Doe that she was having "relations" with Martinez. Third, Maria testified that she was angry at another inmate for failing to report abuse. See id. at 319. But Jane Doe claimed that Maria purposely avoided interactions with prison staff that might reveal her relationship with Martinez.
The prosecution offers two replies: Jane Doe's statements are not favorable to Martinez because they are largely consistent with Maria's testimony and, in any event, the evidence against Martinez was so overwhelming that any inconsistent statements are not material.
Jane Doe's statements do corroborate the broad strokes of Maria's narrative, e.g. , Maria went, unaccompanied, to clean for Martinez in April 2016; the subsequent encounter was sexual; and they had previously had such interactions. But these elements only serve to inculpate Martinez as to the charges based on sexual abuse of a ward. They have the opposite effect as to the other sexual abuse counts, which require that Martinez engaged in sexual acts with Maria through force, by threatening her, or placing her in fear. The thrust of Jane Doe's statements is that Maria had a consensual relationship with Martinez. Jane Doe's account may not be explicit on that point, but Maria's purported attempts to see Martinez alone and to hide their "relations" from prison staff suggest a desire to preserve the relationship, in accordance with the defense's theory that Maria "pursued" Martinez. See Trial Tr. 1028-29; cf. Rivas , 377 F.3d at 200 (explaining that defendant "should have had the opportunity to bolster the defense theory" with suppressed evidence).
The prosecution attempts to recast Jane Doe's statement as consistent with the events described by Maria at trial because it purportedly demonstrates that Maria would have suffered "shame and embarrassment if someone saw her [being raped by Martinez], or worse, jeopardized her imminent release." Opp. 39-40. Putting aside the fact that Maria had already shared details of at least her first encounter with Martinez with her fellow unit-mates and that it is unclear how Martinez's misconduct would have affected Maria's release, Maria's alleged statements to Jane Doe are not necessarily indicative of shame or embarrassment. Indeed, Maria later told Jane Doe about her sexual relationship with Martinez, which she likely would not have done if she was ashamed or embarrassed by it. Moreover, she had already told her friend, Larihelys Vargas, *232that she and Martinez had "go[ne] ... all the way" with no indication whether the relationship was consensual or coercive. See Trial Tr. 317-18.
In sum, the withheld information derived from Jane Doe's interview was exculpatory and its suppression violates Brady if her statements are material. See Mahaffy , 693 F.3d at 130 ("Where suppressed evidence is inculpatory as well as exculpatory, and 'its exculpatory character harmonize[s] with the theory of the defense case,' a Brady violation has occurred." (quoting Triumph Capital , 544 F.3d at 164 )). "Although ... the test of materiality is whether the withheld evidence reasonably undermines confidence in the verdict[,] ... the existence of substantial independent evidence of guilty is unavoidably relevant." See Orena , 145 F.3d at 558. Thus, the other evidence presented at trial must be evaluated.
As one of the Assistant U.S. Attorneys noted: "[T]he whole case really comes down to ... whether the jurors believe Maria.... [I]t's important for the Government to be able to corroborate her in every way that we can." Trial Tr. 688. Indeed, this was the express reason the prosecution fought to ensure that another woman-a fellow prison guard, not an inmate-with whom Martinez was having a consensual relationship would testify. The prosecution introduced this other guard's testimony to corroborate Maria's testimony that Martinez used the video screens in the Lieutenants' Office to avoid getting caught, see id. at 686-90, even though that fact was not disputed. The prosecutor's insistence that the testimony was necessary to corroborate Maria's testimony underscores a key point: This is not a case where "considerable forensic and other physical evidence link[ed] [the defendant] to the crime." See, e.g., Strickler , 527 U.S. at 293, 119 S.Ct. 1936. Instead, the prosecution relied primarily on the perceptions of those who interacted with Maria at the time of the first alleged rape to corroborate her account.
Contradictory evidence can destroy a jury's confidence in a witness's story, especially where the prosecution's case is built upon observations or perceptions of circumstances surrounding or related to the offense conduct, rather than the conduct itself. See Kyles , 514 U.S. at 441-43, 115 S.Ct. 1555 ; cf. Smith v. Cain , 565 U.S. 73, 76, 132 S.Ct. 627, 181 L.Ed.2d 571 (2012) (holding that evidence is material where it undermines "the only [testimony] linking [defendant] to the crime" (emphasis omitted)). This is particularly true where, as here, the most persuasive testimony offered by the prosecution comes from individuals with a personal connection to the victim, see, e.g. , Trial Tr. 161, 181, 188, 289, 306, 441, 447, 551, 629 (witnesses and victim indicating they were friends), and the suppressed evidence offers "independent corroboration of the defense's theory of the case by a neutral and disinterested witness." Boss v. Pierce , 263 F.3d 734, 745 (7th Cir. 2001).
Within that category of testimony from those personally connected to the victim is the testimony of three women who lived in the same unit as Maria. Moreover, the weight of their testimony is least affected by the undisclosed statements of Jane Doe. Although her unit-mates initially joked when Maria told them about the alleged rape upon returning to her unit on December 13, 2015, Maldonado, Osoria, and Vasquez all eventually recognized that she was distraught and upset. Indeed, they remarked that she continued to cry throughout the evening and that she appeared more somber and withdrawn in the days after the incident. Trial Tr. 115-23, 560-62, 640. If Maria had "pursued" Martinez, as the defense argued, or sought to maintain *233their "relations," as the Jane Doe interview suggests, Maria would not have been "upset" or felt "bad," as she testified she did, when her fellow inmates laughed or made jokes after she told them of the sexual contact. See id. at 119, 122, 447-48, 560.
Nevertheless, Jane Doe's statements would have bolstered defense counsel's cross-examination of Maria's fellow inmates. For example, defense counsel engaged in the following colloquy with Kiara Maldonado, one of the prosecution's witnesses and Maria's unit-mate:
Q: You met with the agents. You just told us, they came and spoke to you, right?
...
A: Yes.
Q: And what you told them was that Maria flirted with a male correction officer. Isn't that correct?
A: Yes.
Q: And you say - you also said that Maria told you that she gave him oral sex. Right?
A: Yes.
Q: And then the correction officer penetrated her on the desk and the chair in the lieutenant's office, right?
A: Yes.
Trial Tr. 642. While Maldonado's testimony is suggestive of consent, it is not as definitive as Jane Doe's statements. Accordingly, defense counsel could have strengthened the impact of this cross-examination by calling Jane Doe as a witness, using her statements to further cross-examine Maria, or providing a good-faith basis for asking additional questions of the other women. Cf. U.S. v. Payne , 63 F.3d 1200, 1210 (2d Cir. 1995) (noting that affidavit may be material where it "add[s] concrete evidence" to support impeachment). If the defense had a witness who could testify to the consensual nature of the relationship between Maria and Martinez, perhaps counsel would have pursued other lines of cross-examination, such as asking whether there was another reason Maria might have been crying. For instance, counsel might have asked whether Martinez's failure to use a condom or Maria's fear of being caught might have prompted Maria to be upset. Of course, counsel could have inquired about such things even without Jane Doe's statement, but he would not have been able to rebut a contradictory answer and may not have had a good-faith basis for making such an inquiry absent the evidence furnished by Jane Doe. Such additional testimony or cross-examination may have been feasible if defense counsel knew he could present an affirmative defense of consent.
More significantly, the Jane Doe interview colors much of the other testimony offered by the prosecution. In particular, Maria testified that she contacted Noel Lopez and Larihelys Vargas in her attempts to dissuade Martinez from continuing to abuse her. Trial Tr. 181, 186. Lopez testified that Maria asked him to visit her and kiss her during the visit. Id. at 304-06. Vargas testified that Maria asked her to examine Martinez's Facebook profile and report back on the information she found, including whether he was married or had photos with other women. Id. at 190-91, 292-93, 317-18, 623-24. Indeed, Maria admitted that she asked Vargas to "look [Martinez] up" immediately after confiding in Vargas that Martinez "kiss [her]" and that they "[went] ... all the way." Id. at 317. Significantly, at no point during her conversations with Vargas does Maria indicate that her relationship with Martinez was nonconsensual.
During summation, defense counsel argued that these were not attempts to deflect Martinez's interest; rather, they were *234calculated to inspire jealousy. Id. at 1031-33. Jane Doe's statement could have supported such a theory, as it indicates that Maria engaged in other acts to preserve her relationship with Martinez, such as asking Odis De La Cruz not to accompany her to clean or by declining to receive a medical test she believed might reveal her sexual relationship. Moreover, the narrative of an active interest in Martinez-and, accordingly, a consensual relationship with him-helps explain the otherwise unstated reasons for Maria asking Vargas to look at Martinez's Facebook page.
Put another way, the withheld evidence could have afforded the defense an opportunity to construct an alternative explanation by emphasizing how the Jane Doe statement intersected with, and illuminated the inferences that could be drawn from, the evidence presented by the prosecution. This is the same tactic the prosecution used-structuring its case as a series of interconnected and reinforcing statements aimed at corroborating Maria's version of events. For example:
• Maria testified that, during her first encounter with Martinez, he told her that he had a vasectomy and could not get her pregnant. Trial Tr. 109-10. A doctor confirmed that Martinez had a vasectomy. Id. at 650-51.
• Maria testified that Martinez sent a fellow officer who was present in the area of the Lieutenants' Office to another building in the MDC complex, and proceeded to rape her while the other officer was moving between the two buildings. Id. at 143-49, 159-60. That other officer testified that Martinez did in fact send him to the other building and, at times, it took him at least 20 minutes to return. Id. at 671-73, 717.
• Maria testified that, during the April incident, Martinez could not hold her down as forcefully as usual because his hand had been injured. Id. at 218-19. Documents obtained from the MDC indicated that he sustained a hand injury in early April. Id. at 865.
Each of these examples confirms elements of Maria's narrative, but none addresses the key factual issue raised by Jane Doe's statements-or indeed, the key factual issue at trial: whether the relationship between Maria and Martinez was consensual. Maria could have known each of these facts even if the relationship was consensual. The prosecution's hope was that proving that Maria told the truth about each of these related events would convince the jury to believe her about the core offense conduct. And the jury did. Yet, as noted above, the prosecutor thought that she needed more evidence to further corroborate Maria's story. See id. at 686-90. The mere fact that Maria accurately relayed collateral facts does not necessarily mean that a jury would have believed her as to the core elements of the crime in the face of contradictory testimony offered by Jane Doe, or more extensive cross-examination based on Jane Doe's statements. Cf. Kyles , 514 U.S. at 445, 115 S.Ct. 1555 (evidence material where "[d]amage to the prosecution's case would not have been confined to evidence of the eyewitnesses," but also affected "the probative value of crucial physical evidence and the circumstances" surrounding it).
Moreover, Jane Doe's statements highlight other inconsistencies in the prosecution's theory. The additional information related to Maria's purported reluctance to report the mistreatment she suffered is at odds with her criticism of another inmate for refusing to report her own sexual abuse. See Trial Tr. 318-20. Similarly, the prosecution argued that prison officials would not have been able "to detect the signs of abuse and rape" exhibited by Maria *235due to poor training or limited interaction with her. Id. at 1049-51. But the Jane Doe interview suggests that Maria may not have shown any such signs because she was not a victim of abuse. In sum, the defense could have used the withheld evidence to demonstrate that the trial evidence was equally indicative of consensual sex as nonconsensual sex.
Finally, the prosecution contends that, because defense counsel thoroughly and comprehensively impeached Maria, any additional evidence for that purpose would merely be cumulative. See, e.g., Shabazz v. Artuz , 336 F.3d 154, 166 (2d Cir. 2003) (quoting United States v. Avellino , 136 F.3d 249, 257 (2d Cir. 1998) ); see also United States v. Zagari , 111 F.3d 307, 320-21 (2d Cir. 1997). Nevertheless, few, if any, of the attacks on Maria's credibility went to the core of her testimony-the nonconsensual nature of her interactions with Martinez. See, e.g. , Trial Tr. 243-46 (prior false statements to law enforcement), 257 (same), 261-62 (same), 271-72 (withheld information from asylum petition), 327 (same), 373-74 (inconsistent testimony), 404-05 (inconsistent statements on asylum application and during plea), 433-34 (financial interest in the outcome from filing a $20 million suit against the MDC). Moreover, even though her general credibility was heavily attacked, the defense could not present a compelling alternative explanation for the sexual encounters-specifically, that the relationship was consensual. Jane Doe's statement provides the basis for this alternative narrative, elevating the value of the withheld testimony beyond merely cumulative impeachment evidence.
Taking the value of Jane Doe's statement as a whole and weighing it against the other evidence presented at trial, "there is a 'reasonable probability' that disclosure of the evidence would have led to a different outcome-i.e. , an acquittal or hung jury rather than a conviction" as to the counts where the use of force, threats, or fear is an element. Turner v. United States , --- U.S. ----, 137 S. Ct. 1885, 1897, 198 L.Ed.2d 443 (2017) (Kagan, J., dissenting) (noting that the majority and dissent "agree on the legal standard by which to assess the materiality of undisclosed evidence"). "[T]he testimony that [Jane Doe] would have given ... is of a different nature than that given by" the other witnesses. Boss , 263 F.3d at 745. She references specific acts Maria undertook to hide her relationship with Martinez, offering an alternative narrative to the one presented by the prosecution. That Maria may have taken steps to minimize the likelihood that others would find out about her relations with Martinez lends a great deal of support to the defense's theory of the case, especially in light of Maria's own testimony distinguishing between "relations" and rape and her criticism of other inmates for failing to report sexual misconduct.
Moreover, this new testimony cuts to the very core of the prosecution's approach: they sought to corroborate Maria's version of events "in every way that [they could]." Trial Tr. 688. If a jury ultimately concluded that even some aspect of Maria's relationship with Martinez was consensual-and, more significantly, that Maria was lying about it-the bulk of the case would have unraveled. In other words, Maria's credibility as to any one incident is inextricably linked with the jury's verdict on every other count involving force, threats, or fear. See Wearry , 136 S. Ct. at 1006 (evidence is material where "trial evidence resembles a house of cards, built on the jury crediting" government witness's account). In sum, the withheld evidence substantially corroborates the defense's primary theory, indicating that it is material. See Rivas , 377 F.3d at 200 ; see also *236United States v. Gil , 297 F.3d 93, 104 (2d Cir. 2002) (evidence material where "arguments and inferences [it] could support or reinforce ... bear importantly on the central issue at trial").
CONCLUSION
Jane Doe's testimony may not inexorably lead to an acquittal in this case. But even if the suppressed evidence does not mandate acquittal, it raises a reasonable probability of a different outcome. Accordingly, a new trial must be ordered. Defendant's motion for a new trial is granted, except as to the counts based on sexual abuse of a ward, as consent is not relevant to those counts and the undisclosed evidence primarily goes to the issue of consent.
SO ORDERED.