join, dissenting from the order denying the petition for rehearing en banc:
There is an epidemic of Brady violations abroad in the land. Only judges can put a stop to it.
I
Kenneth Olsen was convicted by a federal jury of knowingly developing a biological agent for use as a weapon in violation of 18 U.S.C. § 175. United States v. Olsen, 704 F.3d 1172, 1177 (9th Cir.2013). Olsen admitted that he produced ricin, a highly toxic poison, but argued that he didn’t intend to use it as a weapon. Instead, he claimed that he was motivated by “an irresponsible sense of curiosity” about “strange and morbid things.” Id.
To show that Olsen acted with the requisite intent, the government produced a bottle of allergy pills found among Olsen’s possessions. An analysis by Washington State Police (WSP) forensic scientist Arnold Melnikoff determined that these pills might contain ricin. Because Melnikoffs lab wasn’t equipped to test for ricin, Melni-koff sent the pills to the FBI, which confirmed his suspicion. Id. According to the government, these spiked allergy pills were tangible proof that Olsen intended to use the poison as a weapon. Id. at 1185.
Olsen tried to cast doubt on this evidence by arguing that the pills were contaminated by Melnikoff before he sent them to the FBI. There was evidence that Melnikoff “handled and extensively manipulated” the pills. Melnikoff admitted to examining the pills “not by individually removing them from the bottle with forceps, but rather by dumping them onto his laboratory bench, albeit on- ‘a sheet of clean lab paper,’ after, he had examined other items on. the. same bench — which included scraping ricin-positive powder from some of these items.” Id. at 1178-79. This was especially important because the ricin test destroyed the pills, so we can’t tell whether the poison was inside them or merely on their surface. Id. at 1177. As a result, Melnikoffs competence and veracity were critical to the government’s case. Unfortunately for the prosecution, however, there were many reasons to doubt both.
Before joining the WSP, Melnikoff ran the Montana State Crime Laboratory. While there, he conducted a hair sample analysis that resulted in the conviction of Jimmy Ray Bromgard for raping an 8-year-old girl. When a DNA analysis exonerated Bromgard after he hád spent fifteen years in prison, officials in Washington and Montana took note. So did the New York Times1 Washington launched an investigation into Melnikoffs “misconduct involving courtroom testimony and/or case analysis,” which was expanded after state officials discovered that another innocent Montana inmate had been wrongfully convicted based on flaws in Melnikoffs work. Id. at 1179. A month before Olsen’s trial began, a third Montana inmate was exonerated on similar grounds. See Know the Cases: Paul D. Kordonowy, Innocence Project, http://goo.gl/rZJrME (last visited Nov. 27, 2013).
The findings of the WSP’s investigation were compiled into a highly critical report: The panel of experts who prepared it doubted “Melnikoffs diligence and care in the laboratory, his understanding of the scientific principles about which he testified in court, and his credibility on the witness stand.” Olsen, 704 F.3d at 1179-80. The report also contained an exten*627sive review of Melnikoffs work in Washington, recommending that the WSP redo lab work in 14 out of 100 randomly selected cases due to “the presence of unexplained contaminants in his laboratory, among other findings.” Id. A few quotes from the report illustrate its force:
• [T]here were many mistakes in case documentation, administrative documentation, evidence analysis, data interpretation, and written reports, for example. The casework analysis seemed to be built around speed and shortcuts.... This focus on speed, as expected, manifested itself in acceptance of weak or insufficient data to reach identifications;' in documentation mistakes and sloppy work; in reaching conclusions of identification without taking time to isolate the controlled substance ....
• [M]any of the results, collected as they were, were not necessarily sufficient to prove the identity of the questioned substance. Often the data was weak and there was an overreliance of basing identification from mixtures without taking the steps to separate the components. There were often contaminants or unexplained material in the blanks, and these contaminants were often not noted or identified, and no obvious attempt was made to remove them,
• There was a tendency for conclusions to become stronger as the ease developed, from notes to written report to testimony.
Melnikoffs employment with the WSP was eventually terminated, and the Washington Court of Appeals affirmed the termination based on a “finding that Melnikoff was incompetent and committed gross misconduct.” Melnikoff v. Washington State Patrol, 142 Wash.App. 1018, 2008 WL 40158, at *11 (2008).
Olsen’s lawyer knew that an investigation was' underway, but he didn’t know its scope. Nor did he know that .the WSP’s report had been completed and sent to the state decisionmaker two months before Olsen’s trial began. Rather than inform defense counsel and the court of these important developments, the Assistant U.S. Attorney prosecuting the case materially understated the scope, status and gravity of the investigation. He claimed that the investigation was “purely administrative” and revolved : around a decades-old complaint limited to DNA testing, which wasn’t, at, issue in Olsen’s case. Melni-koffs . lawyer, Rocco Treppiedi, made an appearance and represented that the WSP was “in the process of investigating” the matter and that, as of that time, there was “absolutely no evidence, no allegation that Mr. Melnikoff has ever done anything inappropriate with respect to anything other than his opinion testimony on the hair sampling.” The Assistant U.S. Attorney added that the WSP investigation was ongoing and represented that “[tjhere is nothing further that you should know about.”
Each of these statements was contradicted by the WSP report. But because the Assistant' U.S. Attorney failed to disclose the contents of the report to the district judge, the judge relied heavily on the Assistant U.S. Attorney’s inaccurate representations in evaluating Olsen’s request to cross-examine Melnikoff about the investigation. The judge stated: “[L]et me just say a few words about my understanding of this, and counsel should check me if I am wrong.”' He then said that “[t]he only issue here involved at all is whether or not there was some inaccuracy regarding his testimony in Montana about comparing hair samples on rape and homicide cases,” and that “there is nothing in here that I see that indicates that there *628was any problem at all during the state — - his tenure with the State of Washington.” He summarized: “There has been no investigation — or at least no conclusions. They are just bare allegations involving hair sample analyses, the subject of which is not in any way involved in this case.” The district judge surmised that Melni-koffs forced administrative leave during the investigation must have been taken “out of an abundance of precaution.” He then asked, “[Wjhere am I off the track?”
Nearly everything the district judge understood to be true was false. But the prosecutor did not correct the district judge, who then concluded that it would be “unfair to Mr. Melnikoff to allow counsel to delve into this issue” and “under an analysis of [Federal Rule of Evidence] 403, it just would be improper to go into that.” As a result, the government introduced the spiked allergy pills and the jury heard Melnikoff s testimony, all without ever being informed of these serious doubts about their reliability.
II
Olsen claims that the prosecutor’s failure to disclose the WSP investigative report violated Brady v. Maryland, 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963). Brady holds that a prosecutor violates due process when he (1) suppresses evidence (2) that is favorable to the defendant, when that evidence (3) is material to guilt or innocence. Id. at 87, 83 S.Ct. 1194. This extends to evidence that bears- upon the credibility, of a government witness. Gig-lio v. United States, 405 U.S. 150, 153-54, 92 S.Ct. 763, 31 L.Ed.2d 104 (1972). The panel expressly recognizes that the report was favorable to Olsen; nevertheless it dismisses Olsen’s complaint on the ground that the WSP report wasn’t material.2 Olsen, 704 F.3d at 1183-87."
Evidence is material under Brady if it creates “a ‘reasonable probability’ of a different result.” Kyles v. Whitley, 514 U.S. 419, 434, 115 S.Ct. 1555, 131 L.Ed.2d 490 (1995). “A reasonable probability does not mean that the defendant ‘would more likely than not have received a different verdict with the evidence,’ only that the likelihood of a different result is great enough to ‘undermine[ ] confidence in the outcome of the trial.’ ” Smith v. Cain, — U.S. -, 132 S.Ct. 627, 630, 181 L.Ed.2d 571 (2012) (quoting Kyles, 514 U.S. at 434, 115 S.Ct. 1555). To say that the undisclosed information wasn’t material, a court must conclude that the other evidence was so overwhelming that, even if the withheld evidence had been presented to the jury, there -would be no “reasonable probability” that it would have acquitted. This standard isn’t satisfied if “the State’s argument offers a reason that the jury could have disbelieved [the undisclosed evidence], but gives us no confidence that it would have done so.” Id.
Although the investigative report provided objective, empirical support to corroborate Olsen’s claim that Melnikoff contaminated the pills, the panel concludes that there is no “reasonable probability” that the jury would have acquitted Olsen *629had the investigative . report been disclosed. According to the panel, the pills are just a sideshow. Quoting the government’s brief, the panel insists that the pills were “ ‘simply one more layer in an already overwhelming case against the Defendant’” — an “overwhelming” case built on “devastating evidence.” Olsen, 704 F.3d at 1185. “Even if Melnikoff s credibility as a witness had been totally destroyed we are confident beyond doubt that the jury would have found Olsen guilty, based on the overwhelming evidence presented by the government that he intended to use the ricin he possessed as a weapon.” Id. (emphasis added). •
So what is this “devastating,” “overwhelming” evidence? “Thought” evidence, of course: “ “What is unique about the evidence in this case is we have captured a thought process,’ ” the prosecutor said in closing. Id. at 1186. By this, the prosecutor meant the ambiguous evidence of a year’s worth of Olsen’s Internet browsing. Id. at 1185. In an investigation tfiat would have made Big Brother green with envy, the government produced 20,000 pages showing the websites Olsen visited and the searches he performed. Id. During that year, Olsen visited sites relating to ricin and other poisons. Id. He printed materials from these pages and ordered works with menacing titles such as: “ ‘How to Kill,’ ‘Silent Death,’ ‘Getting Even,’ and the ‘Poisoner’s Handbook.’ ” Id. at 1186. Olsen also searched for “ ‘silent killers,’ ‘death by poison,’ ‘tasteless poison,’ ‘hidden poison,’ ‘undetectable poisons,’ ‘untraceable poisons’ ... ‘deadly sleeping pills,’ and ‘common ingredients for death by sleep.’ ” Id.
The panel then tells us of Olsen’s handwritten notes in which he wrote out “the maximum doses, in milligrams,” of four antihistamine and sleeping drugs, including the one that had been spiked with ricin. Id. at 1186-87. ‘(Perhaps most incriminating of all,” the panel says, was the fact that “Olsen mathematically calculated the weight in kilograms of a 150-pound person.” Id. at 1187.' We’re told that Olsen’s wife, his mistress and his former boss “all weighed around 150 pounds,” id., as if this is proof that Olsen intended to kill one (or all?) of them.
Intriguing, in a Jerry Springer kind of way, but whom was Olsen planning to kill? We don’t know. And what was his motive? The panel doesn’t say. Given that the government so thoroughly “captured [Olsen’s] thought process,” id. at 1186, it’s surprising that these “thoughts” don’t shed light on the intended victim (or victims?). Surely somewhere in the 20,000 pages of Internet proxy logs . Olsen searched for “what to wear to your boss’s funeral” or “how to file a widower’s tax return,”, or maybe he watched “How to Murder Your Wife” on Netflix. But the. opinion makes no mention of it, which makes the materiality analysis that much weaker.
This is hardly the “overwhelming” evidence of intent that the panel promises. The evidence is consistent with Olsen’s intent to use the ricin as a weapon, of course, but it’s also consistent with the irresponsible curiosity that Olsen claims motivated him. The pills shed an entirely different light on the matter. They demonstrate that Olsen moved beyond curiosity and took concrete steps to use the poison. They are the “glue that held the prosecution’s case together,” providing “the only ‘direct’ evidence that connected [Olsen] to .the crime.” Horton v. Mayle, 408 F.3d 570, 579 (9th Cir.2005).
Had the jury seen the WSP report or been told of its contents, it may well have developed doubts about whether Olsen poisoned the allergy pills. But without the report, without any indication that Melni-koff s work had been condemned as incom*630petent by a panel of experts, without any indication that contaminating samples through careless handling was Melnikoff s modus operandi, the jury had no reason to doubt that it was Olsen who contaminated the allergy pills. And, once the jury believed Olsen went so far as to spike the pills, all the remaining, otherwise ambiguous evidence falls into place as further corroboration of his ill intent.
Ill
The panel's ruling is not just wrong, it is dangerously broad, carrying far-reaching implications for the administration of criminal justice. It effectively announces that the prosecution need not produce exculpatory or impeaching evidence so long as it’s possible the defendant would’ve been convicted anyway. This will send a clear signal to prosecutors that, when a case is close, it’s best to hide evidence helpful to the defense, as there will be a fair chance reviewing courts will look the other way, as happened here.
A robust and rigorously enforced Brady rule is imperative because all the incentives prosecutors confront encourage them not to discover or disclose exculpatory evidence. Due to the nature of a Brady violation, it’s highly unlikely wrongdoing will ever come to light in the first place. This creates a serious moral hazard for those prosecutors who are more interested in winning a conviction than serving justice. In the rare event that the suppressed evidence does surface, the consequences usually leave the prosecution no worse than had it complied with Brady from the outset. Professional discipline is rare, and violations seldom give rise to liability for money damages. See, e.g., Connick v. Thompson, — U.S.-, 131 S.Ct. 1350, 1366, 179 L.Ed.2d 417 (2011). Criminal liability for causing an innocent man to lose decades of his life behind bars is practically unheard of. See, e.g., Nathan Koppel, Texas Ex-Prosecutor Gets Brief Jail Time for Misconduct, Wall St. J., Nov. 9-10, 2013, at A5 (reporting that Texas settled civil and criminal misconduct charges against former prosecutor Ken Anderson, whose suppression of evidence in a murder trial resulted in the defendant spending 25 years in prison, in exchange for Anderson “forfeiting] his law license, [ ] serving] up to 10 days in jail, paying] a $500 fine and performing] 500 hours of community service”). If the violation is found to be material (a standard that will almost never be met under the panel’s construction), the prosecution gets a do-over, making it no worse off than if it had disclosed the evidence in. the first place.
Olsen’s prosecution highlights the problem. The prosecutor just did not take his constitutional duty to disclose exculpatory evidence very seriously. This is not the usual case where the prosecutor was unaware of exculpatory evidence being held by the police without his knowledge. The Assistant U.S. Attorney knew Melnikoff was being investigated and promised the district court that he would get more information, but never followed through. The record shows that Elizabeth Brown, his state counterpart, repeatedly attempted to contact the Assistant U.S. Attorney concerning the Melnikoff investigation. On April 8, she left a message asking him to call her regarding Melnikoff. She tried again on April 15, this time providing the name and phone number of another state employee whom the Assistant U.S. Attorney could contact about the Melnikoff investigation. Ten weeks later, at a June 30 pretrial hearing, the Assistant U.S. Attorney promised the court that he would call the state, acknowledging that he “ha[d] not had a chance to call the state authorities” but “will continue to pursue it.” He stated: “I know that I have to follow up and *631contact the State of Washington again, which I have done in the past, and every time I have contacted them, they have told me exactly what Mr. Treppiedi has told me.” The Assistant U.S. Attorney didn’t say whom he spoke to, but it wasn’t Brown, who was in charge of this aspect of the investigation.
Somehow, the Assistant U.S. Attorney never managed to get ahold of Brown, even though she called his office yet again on July 3, the day Melnikoff testified. Brown spoke to the Assistant U.S. Attorney’s secretary and asked that he call her office, but he failed to inform the court or defense counsel of the call and, so far as the record reflects, did not return the call. Instead, he stood before the district judge and uttered falsehoods about the WSP investigation. See page' 627, supra. All told, the record shows only k single call from the prosecutor to Brown — three months before Olsen’s trial.3 He left a message, but failed to answer her return call later that afternoon.
Is there any doubt that if the Assistant U.S. Attorney had thought Brown had in-culpatory evidence to provide him, he would have managed to connect with her? Or if these messages had been from a doctor or broker or child’s schoolteacher? We can be sure that the Assistant U.S. Attorney would have found the time for an extra phone call. But protecting the constitutional rights of the accused was just not very high on this prosecutor’s list of priorities. The -fact that a constitutional mandate elicits less diligence from a government lawyer than one’s daily errands signifies a systemic problem: Some prosecutors don’t care about Brady because courts don’t make them care.
I wish I could say that the prosecutor’s unprofessionalism here is the exception, that his propensity for shortcuts and indifference to his ethical and legal responsibilities is a rare blemish and source of embarrassment to an otherwise diligent and scrupulous corps of attorneys staffing prosecutors’ offices across the country. But it wouldn’t be true. Brady violations have reached epidemic proportions in recent years, and the federal and state reporters bear testament to this unsettling trend. See, e.g., Smith v. Cain, — U.S. -, 132 S.Ct. 627, 181 L.Ed.2d 571 (2012); Urtited States v. Sedaghaty, 728 F.3d 885 (9th Cir.2013); Aguilar v. Woodford, 725 F.3d 970 (9th Cir.2013); United States v. Kohring, 637 F.3d 895 (9th Cir. 2010); Simmons v. Beard, 590 F.3d 223 (3d Cir.2009); Douglas v. Workman, 560 F.3d 1156 (10th Cir.2009); Harris v. Lafler, 553 F.3d 1028 (6th Cir.2009); United States v. Zomber, 299 Fed.Appx. 130 (3d Cir.2008); United States v. Triumph Capital Grp., Inc., 544 F.3d 149 (2d Cir.2008); United States v. Aviles-Colon, 536 F.3d 1 (1st Cir.2008); Horton v. Mayle, 408 F.3d 570 (9th Cir.2004); United States v. Sipe, 388 F.3d 471 (5th Cir.2004); Monroe v. Angelone, 323 F.3d 286 (4th Cir.2003); United States v. Lyons, 352 F.Supp.2d 1231 (M.D.Fla.2004); Watkins v. Miller, 92 F.Supp.2d 824 (S.D.Ind.2000); United States v. Dollar, 25 F.Supp.2d 1320 (N.D.Ala.1998); People v. Uribe, 162 Cal. App.4th 1457, 76 Cal.Rptr.3d 829 (2008); Miller v. United States, 14 A.3d 1094 (D.C. 2011); Deren v. State, 15 So.3d 723 (Fla. Dist.Ct.App.2009); Walker v. Johnson, 282 Ga. 168, 646 S.E.2d 44 (2007); Aguilera v. State, 807 N.W.2d 249 (Iowa 2011); DeSimone. v. State, 803 N.W.2d 97 (Iowa 2011); *632Commonwealth v. Bussell, 226 S.W.3d 96 (Ky.2007); State ex rel. Engel v. Dormire, 304 S.W.3d 120 (Mo.2010); Duley v. State, 304 S.W.3d 158 (Mo.Ct.App.2009); People v. Garrett, 106 A.D.3d 929, 964 N.Y.S.2d 652 (N.Y.App.Div.2013); Pena v. State, 353 S.W.3d 797 (Tex.Crim.App.2011); In re Stenson, 174 Wash.2d 474, 276 P.3d 286 (2012); State v. Youngblood, 221 W.Va. 20, 650 S.E.2d 119 (2007).
When a public official behaves with such casual disregard for his constitutional obligations and the rights of the accused, it erodes the public’s trust in our justice system, and chips away at the foundational premises of the rule of law. When such transgressions are acknowledged yet forgiven by the courts, we endorse and invite their repetition.
Olsen’s case points to another important problem — that of rogue investigators and forensic experts. Melinkoff s long history of misconduct, resulting in the wrongful conviction of numerous innocent people, is hardly unique. Just last month, Annie Dookhan, a Massachusetts crime-lab technician, was sentenced to-3-5 years imprisonment after spending several years filing positive results for samples she had not properly tested. Her misconduct tainted over 40,000 drug samples, implicating several thousand defendants (hundreds of whom have already been released). See Katharine Q. Séelye & Jess Bidgood, Prison for a State Chemist Who Faked Drug Evidence, N.Y. Times, Nov. 23, 2013, at A9; Jennifer Levitz, Crime Lab’s Ex-Chemist Is Sentenced, Wall St. J., Nov. 23-24, 2013, at A2. Such incidents have become distressingly common. See, e.g., Joseph Goldstein, New York Sees Errors on DNA In Rape Cases, N.Y. Times, Jan. 11, 2013, at A1 (noting a city investigation that confirmed mishandling of evidence in at least 26 rape cases, prompting officials to explore over 800 more cases); Agency: Houston crime lab worker had history of poor work, ABC (Apr. 5, 2013), http://goo. gl/9o8pfA (recounting the substandard work of a Texas lab technician affecting nearly 5,000 cases in 36 counties). Even the vaunted FBI Laboratory at Quantico, Virginia hasn’t been immune from charges of falsification and pro-prosecution bias. See Spencer S. Hsu, Justice Dept., FBI to review use of forensic evidence in thousands of cases, Wash. Post (July 10, 2012), http://goo.gl/9GQUFw (announcing that the FBI was undertaking the largest-ever review of potential lab misconduct — over ten thousand cases going back to at least 1985 — due to numerous problems with forensic analyses).
Because modern criminal trials frequently turn on forensic reports, these incidents of misconduct raise the frightening prospect that many of the over 1.5 million people now populating state and federal prisons might, in fact, be innocent. See E. Ann Carson & Daniela Golinelli, Prisoners in 2012 — Advance Counts, Bureau of Justice Statistics (July 2013). How do rogue forensic scientists and other bad cops thrive in our criminal justice system? The simple answer is that some prosecutors turn a blind eye to such misconduct because they’re more interested in gaining a conviction than achieving a just result. See, e.g., Goldstein v. City of Long Beach, 715 F.3d 750, 751-52 (9th Cir.2013) (recounting the Los Angeles District Attorney’s Office’s repeated use of a jailhouse informant despite serious (and undisclosed) concerns about his reliability, including his heroin addiction and a history of providing unsubstantiated testimony in exchange for sentence reductions), rev’d 555 U.S. 335, 129 S.Ct. 855, 172 L.Ed.2d 706 (2009). Montana’s investigation of Melnikoffs work is emblematic of the problem. The state’s Attorney General conducted such a perfunctory investigation that a group of concerned citizens, includ*633ing three retired justices of the Montana Supreme Court, filed a petition accusing him of “abandoning justice.” Maurice Possley, Lab etrors cited in 'petition on 200 Montana cases, Chi. Trib. (Aug. 26, 2004), http://goo.gl/Wxf6a8. The Attorney General, for his part, “said he believes Melnikoff was competent and there are no problems in the crime lab.” Id.
We must send prosecutors a clear message: Betray Brady, give short shrift to Giglio, and you will lose your ill-gotten conviction. Unfortunately, the panel’s decision sends the opposite message. The panel shrugs off an egregious Brady violation as immaterial. Had Melnikoff been fully impeached, the only evidence from which the prosecutor could’ve proven Olsen’s intent to use ricin as a weapon would have been a few Google searches and bookstore receipts. This is surely enough to show a reasonable probability of a different result. By raising the materiality bar impossibly high, the panel invites prosecutors to avert their gaze from exculpatory evidence, secure in the belief that, if it turns up after the defendant has been convicted, judges will dismiss the Brady violation as immaterial.
On these facts, I would easily find a Brady violation. My only question is whether the Assistant U.S. Attorney’s failure to discover the WSP report was willful or recMess, and if so, whether an Order to Show Cause should be entered inquiring why he and his supervisor should not be sanctioned. There is room for reasonable disagreement on this point, but I fail to see any plausible argument that the prosecutor did not violate Olsen’s constitutional rights when he failed to disclose evidence casting serious doubt on the reliability of the only dispositive piece of evidence in the ease. By turning a blind eye to this grave transgression, the panel has shirked its own duty and compounded the violence done to the Constitution by the Assistant U.S. Attorney.
. Adam Liptak, 2 States to Review Lab Work of Expert Who Erred on ID, N.Y. Times, Dec. 19, 2002, http://www.nytimes.com/2002/12/19/us/ 2-states-to-review-lab-work-of-expert-who-erred-on-id.html.
. Because the panel found the report immaterial, it didn't address the "complex question” of whether knowledge of the state report could be imputed to the federal prosecutor. Olsen, 704 F.3d at 1183. This court has not addressed this issue, but several of our sister circuits have found such imputation proper in at least some circumstances. See, e.g., United States v. Risha, 445 F.3d 298, 304 (3d Cir. 2006); United States v. Antone, 603 F.2d. 566, 570 (5th Cir.1979). Because the panel declined to address this issue, however, we must attribute to the prosecutor the knowledge that Melnikoff was incompetent and routinely contaminated samples entrusted to him for testing. But the Assistant U.S. Attorney failed to disclose any of this to the defense or the district court.
. Olsen sought an evidentiary hearing on his Brady claim in the district court, at which he planned to depose the Assistant U.S. Attorney. Although the court initially scheduled one, it was subsequently canceled. Thus, the Assistant U.S. Attorney has never directly spoken to the attempts he made'to contact the WSP.