United States v. Risha 445 F.3d 298

OPINION OF THE COURT

BECKER, Circuit Judge.

This is a government appeal from an order of the District Court granting a new-trial on Brady grounds to Defendant Jesse James Risha, who was convicted of attempted arson in violation of 18 U.S.C. §§ 844(i) and (2). See Brady v. Maryland, 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963). The District Court concluded that the government’s key witness, Frank Caito, expected consideration for testifying against Risha, and that his testimony in fact helped him to secure an extremely favorable plea agreement in unrelated state charges pending against him. The Court therefore held that a new trial must be granted because of the government’s failure to disclose these facts. Of course, a failure of the prosecution to disclose impeachment evidence, coupled with a duty to disclose, would result in a Brady violation. See Giglio v. United States, 405 U.S. 150, 92 S.Ct. 763, 31 L.Ed.2d 104 (1972).

No evidence has been advanced that the federal prosecutors in Risha’s case had actual knowledge of Caito’s expectations or of a pending plea agreement. The question ultimately presented here is whether cross-jurisdiction constructive knowledge can be imputed to the federal prosecution because of close involvement between the federal prosecution and state agents, or because impeachment information may have been “readily available.” Under certain fact findings, such knowledge might be imputed. In fact, we find it possible, indeed very likely, that a new trial should be granted. However, though the District Court made a conclusory determination that the jury should have been told that Caito expected a deal as a result of his federal cooperation, it did not address, in terms, the necessary question of constructive possession. That question involves factual and credibility determinations. Therefore, we will vacate the District Court’s order and remand for a determination whether there was constructive possession of Brady material.

I. Facts and Procedural History

The government alleges that on May 10, 1998, Risha offered to pay Caito to set fire to video machines owned by Risha so that Risha could recover insurance proceeds. Caito attempted to set fire to the Clairton, Pennsylvania property where the machines were located; however, the smell of gasoline was quickly reported to the authorities, and the fire did not ignite.

In September of 2002, more than four years later, Caito learned of an unrelated state warrant for his arrest for the illegal sale of firearms. He turned himself in to Pennsylvania authorities, later disclosing his involvement in the 1998 fire. As detailed below, proceedings on the state firearms charges were postponed until after Risha’s federal trial for the arson, presumably because of Caito’s cooperation. Ultimately, Caito received only probation on the state charges.

Risha was twice brought to trial for aiding and abetting the attempted arson of *300a building used in, and affecting, interstate commerce. His first trial ended in a hung jury, and his second in a conviction. Caito testified as a principal witness for the government in both trials. At Risha’s second trial, the one at issue here, Caito testified that he had been given immunity for his testimony. He also told the jury that testifying against Risha would not have any impact on the disposition of the state firearm charges against him.1 At one point during cross examination, Caito asserted that the state and federal cases had nothing to do with each other. He did, however, disclose the earlier continuances in his state case and the fact that an agent involved in the federal case against Risha had arrested him for the state firearms charge.

In its instructions to the jury, the District Court advised the jurors to take note of Caito’s federal immunity. However, the jury was at no time instructed to consider the pending state charges against Caito. And yet, the prosecutor in Risha’s trial emphasized in his closing that Caito had nothing to gain by testifying, stating that because he had immunity, he did not have “any particular reason not to tell the truth.” The controversy here is whether the government was obligated to disclose that Caito did in fact expect leniency and a forthcoming plea agreement in the state charges against him.

A. Disposition of the State Charges Against Caito

The proceedings on the state firearm charges were first scheduled for May, 2003. Trial was postponed for various reasons. On June 9, 2004, Caito moved for another continuance, this time in apparent reference to his testimony in Risha’s case:

Defendant’s participation in a federal court matter is not yet resolved (hung jury earlier this year; anticipated trial date late summer per AUSA Shawn [sic] Sweeney). Resolution of federal matter and unrelated state court matter will ultimately provide both parties with the factual basis to resolve this matter without a jury or non-jury trial.

Caito’s attorney, David Chontos, marked the “non-jury trial” box on the motion and wrote that he was requesting a plea. Bradley Hellein, Assistant Pennsylvania Attorney General and the prosecutor in the state case against Caito, consented to the motion. Risha argues, and the District Court agreed, that these communications indicated that “a plea agreement would be forthcoming,” following, and as a consequence of, Caito’s testimony against Risha.

At the District Court hearing, Hellein testified that he knew that Caito was cooperating in the federal investigation of Risha. He further testified that he told Caito, “every time [he] met him,” that any state or federal cooperation would be “taken into consideration” in resolving the state charges-.2 However, Hellein also testified that he never specifically stated that Caito would receive more “lenient treatment,” and that he did not have authority to make ultimate decisions regarding sen-*301fencing recommendations.3 David Chontes, Caito’s state-court attorney, also suggested that he expected Caito’s federal testimony against Risha to affect the disposition of the state charges. However, as indicated below, some of Chontes’ testimony on this matter is conflicting. On September 17, 2004, after Risha was convicted, Caito entered a plea to two counts of possession of a firearm without a license. Two more serious charges were dropped, as requested by the state. As felonies in the third degree, the crimes together carried a maximum penalty of 7 to 14 years incarceration and a fine of $30,000, yet Caito’s plea agreement was for a period of probation only, the length of which was to be determined by the court. At sentencing, Caito was placed on probation for only one year. During Caito’s sentencing, Hellein made clear his knowledge of Caito’s involvement in Risha’s case. In a statement that the District Court felt confirmed Caito’s expectations of a beneficial plea agreement, Hellein advised the Court of Caito’s cooperation:

We would also add that Mr. Caito has provided very valuable assistance to the Commonwealth and the United States of America with regard to a certain prosecution that occurred in the Western District of Pennsylvania resulting in a conviction approximately a month and a half ago.

The District Court found it undisputed that the cooperation referenced was Caito’s testimony against Risha.

B. The Overlap of State and Federal Agents

As described below, a finding of constructive knowledge as between federal and state forces may depend on the extent to which the forces overlapped.

It appears that the investigation of the arson began as a joint federal-state effort. It also appears that at least one state agent was simultaneously involved in the federal case against Risha and the state case against Caito. Though neither the District Court nor the government discuss Agent Paul Marraway at length, his role may be highly relevant. Marraway was an agent with the state Attorney General’s Office, who arrested Caito on the firearm charges. Marraway was also involved in the investigation of the attempted arson, and continued to be involved after Risha was indicted on federal charges. Notably, Marraway actually sat at the government counsel table during Risha’s first federal trial.

Caito’s state court attorney, David Chontes, suggested in his testimony that Agent Marraway spoke with Hellein and may have assisted Caito in receiving a deal on the state charges. At one point Chontes recollects a “powwow,” apparently before Risha’s second trial, during which Chontos, Caito, Hellein, and Marraway met to discuss Caito’s cooperation.4 Chontos affirmed that he told Caito “on numerous occasions” that in order to receive leniency, he needed to “satisfy” state Agent Marraway in the federal case. Marraway testified that he advised Hellein of Caito’s cooperation in federal court. He *302also represented that there was contact between himself and Shaun Sweeney, the federal prosecutor in Risha’s case. On the other hand, Marraway testified that he specifically told Caito that consideration from the federal government was a different matter from consideration in the state case, and that he was aware of no one from the state Attorney General’s Office who implied or directly stated that Caito’s federal cooperation would benefit him in the state system. Also, at one point, Caito claimed that he did not think there were manifestations by Sweeney or Hellein that federal cooperation would benefit Caito in the state charges, but that Marraway may have made such manifestations.

As explained below, if a team or joint investigation did exist here, or if any state agent was acting on behalf of the federal government, the federal prosecution may be charged with the knowledge of the state Attorney General’s Office — including the knowledge of Marraway and perhaps Hellein. Notably, the federal prosecution conceded, in a September 8, 2004 filing with the District Court, that it constructively possessed the state grand jury testimony of Richard Merlo. Merlo was, allegedly, a long-time associate of Risha. As here, the government maintained that it' had no actual knowledge of the 2000 grand jury testimony. But it stated:

Given the fact that government counsel was working with agents of the AG’s Office in the instant case, government counsel should have specifically inquired of the agents as to whether Merlo had testified before the grand jury.... Again, the government concedes that, as a matter of law, government counsel’s lack of specific knowledge of the transcript is not an excuse for failing to disclose it.

(Emphasis added.)

We believe that this language, omitted by the government in its current appeal, is indicative that federal and state forces may have acted as a team. It appears that the federal prosecution itself viewed its relationship with the state Attorney General’s Office as collaborative. Of course, the District Court will take this information under consideration on remand.

C. The District Court’s Opinion

The District Court found a Brady violation and granted Risha’s motion for a new trial. It stated that the disposition of Caito’s state court charges “[gave] rise to the assumption” that he had a motive for lying. It cited the fact that Caito’s state proceedings were continued on a number of occasions and that Chontos testified that he requested postponements for the “express purpose” of allowing Caito to complete his testimony. Each time, it noted, Caito’s counsel and the deputy attorney general agreed to a postponement. The Court emphasized Chontos’ comment in conjunction with Caito’s June 3 postponement that completion of the federal testimony would “provide both parties with the factual basis to resolve th[e] matter without a jury or non-jury trial.”

The Court agreed with the government that “there was no representation, at any time, by either Assistant United States Attorney Shaun Sweeney, or the agent on the case, to Caito or Chontos that the United States would make any attempt to intervene in the state court proceedings.” (Emphasis added.) However, the Court did not opine as to whether there was actual or constructive knowledge of an expected benefit on the part of any “direct” member of the federal prosecution. At all events, the Court concluded that “the jury should have been informed that the government’s key witness expected to receive a benefit for testifying against the Defendant.” It therefore vacated the jury ver*303diet and ordered a new trial. The government filed a timely notice of appeal.

II. Jurisdiction and Standard of Revieiv

The District Court exercised jurisdiction over this case pursuant to 18 U.S.C. § 3281. This Court has jurisdiction pursuant to 18 U.S.C. § 3731. Brady claims present mixed questions of law and fact. This Court conducts a de novo review of the District Court’s conclusions of law, and a clearly erroneous review of findings of fact. See Virgin Islands v. Fahie, 419 F.3d 249, 252 (3d Cir.2005). Where the correct legal standard has been used, “weighing of the evidence merits deference from the Court of Appeals, especially given the difficulty inherent in measuring the effect of a non-disclosure on the course of a lengthy trial covering many witnesses and exhibits.” United States v. Thornton, 1 F.3d 149, 158 (3d Cir.1993) (quotation omitted).

III. Discussion

Under Brady v. Maryland, 373 U.S. at 87, 83 S.Ct. 1194, the prosecution’s suppression of evidence favorable to a criminal defendant violates due process when the evidence is material to guilt or punishment. To establish a Brady violation, it must be shown that (1) evidence was suppressed; (2) the evidence was favorable to the defense; and (3) the evidence was material to guilt or punishment. See, e.g., United States v. Pelullo, 399 F.3d 197, 209 (3d Cir.2005). This is an objective test, meaning that no bad-faith inquiry is required. United States v. Merlino, 349 F.3d 144, 154 (3d Cir.2003). This case turns on the answer to the first Brady inquiry — -whether the government suppressed relevant impeachment information.5

There is no question that the government’s duty to disclose under Brady reaches beyond evidence in the prosecutor’s actual possession. Since Giglio, 405 U.S. at 154, 92 S.Ct. 763, the Supreme Court has made clear that prosecutors have “a duty to learn of any favorable evidence known to the others acting on the government’s behalf in the case, including the police.” Kyles v. Whitley, 514 U.S. 419, 437, 115 S.Ct. 1555, 131 L.Ed.2d 490 (1995). Constructive possession has been defined by this Court as follows:

We construe the term “constructive possession” to mean that although a prosecutor has no actual knowledge, he should nevertheless have known that the material at issue was in existence. Accordingly, we consider whether the prosecutor knew or should have known of the materials even though they were developed in another case.

United States v. Joseph, 996 F.2d 36, 39 (3d Cir.1993) (emphasis added).

In United States v. Perdomo, we discussed the possibility of constructive possession at some length. 929 F.2d 967, 970 *304(3d Cir.1991). A Brady violation was found where defense counsel submitted requests for the criminal background of prosecution witnesses, and the prosecution failed to check local Virgin Islands records. A National Crime Information Center computer check that did not uncover local information was considered insufficient, and constructive possession of the local, unsearched, records was found. The panel reasoned that the criminal background information was indeed “readily available” to the prosecution. Id. at 970. Therefore, the failure to disclose was a suppression of exculpatory evidence. Id.

In Thornton, we also rejected a hands-off approach to information about government witnesses. Prosecutors were charged with constructive knowledge of DEA payments to government witnesses though they had no actual knowledge of the payments. We concluded that “prosecutors have an obligation to make a thorough inquiry of all enforcement agencies that ha[ve] a potential connection with the[ir] witnesses.” Thornton, 1 F.3d at 158 (finding that, though there was a duty to disclose, materiality was lacking because the witnesses were not critical to the trial).

Still, this Court has placed limitations on constructive knowledge in the Brady context. In Pelullo, we asked whether officials from the Pension and Welfare Benefits Administration (“PWBA”) who possessed relevant documents should be considered members of the “prosecution team.” 399 F.3d at 218. We concluded that there was no “constructive knowledge” because there was no reason to believe that the PWBA was acting on behalf of the prosecution. There was no indication that the PWBA and the prosecution were “engaged in a joint investigation” or that they “otherwise shared labor and resources.” Id. Instead, PWBA investigators played no role in the criminal case.

Id. We have also made clear that prosecutors are not required to undertake a “fishing expedition” in other jurisdictions to discover impeachment evidence. For example, prosecutors are not obligated to learn of all information “possessed by other government agencies that have no involvement in the investigation or prosecution at issue.” Merlino, 349 F.3d at 154 (quotation omitted).

It appears that in addressing the issue of cross-jurisdiction constructive knowledge, most courts of appeals have looked to the same questions that we have. Those questions include: (1) whether the party with knowledge of the information is acting on the government’s “behalf’ or is under its “control”; (2) the extent to which state and federal governments are part of a “team,” are participating in a “joint investigation” or are sharing resources; and (3) whether the entity charged with constructive possession has “ready access” to the evidence. We touch on each of these questions briefly, in turn.

First, there is the question whether a party with actual knowledge of the impeachment information was under the federal government’s control or acting on its behalf. See, e.g., Moreno-Morales v. United States, 334 F.3d 140, 146 (1st Cir.2003) (finding that “the Puerto Rico Senate was not acting on behalf of the federal government”); United States v. Leos-Hermosillo, No. 98-50546, 2000 WL 300967 at *3, 2000 U.S.App. LEXIS 5012 at *8 (9th Cir. Mar. 22, 2000) (holding that even though an officer “was not employed by ... the federal government, he was no less an agent of the federal government; he was acting on its behalf and subject to its control”). What is at issue here, at least generally, is the intermingling of state and federal forces. The record does not fully reflect the extent to which an officer of the state Attorney General’s Office knew that *305Caito’s testimony would benefit him or that a plea agreement would be forthcoming; nor do the facts fully indicate the extent to which state officers were acting “on behalf’ of the federal prosecution. We do know that at least Agent Marraway, a state agent, assisted the federal government and was at the counsel table during Risha’s first trial. However, the full scope of his role is, at this point, unclear.

Related to the question whether an agent is acting on behalf of the government is the question whether the forces are part of a team or are engaged in a joint effort. See, e.g., United States v. Beers, 189 F.3d 1297, 1303-04 (10th Cir.1999); Moon v. Head, 285 F.3d 1301, 1310 (11th Cir.2002) (refusing to impute to a Georgia prosecutor evidence possessed by a Tennessee Bureau of Investigation agent because the agencies shared no resources or labor); United States v. Brooks, 966 F.2d 1500, 1503 (D.C.Cir.1992) (holding that the prosecution must search files of other branches of government if they are “closely aligned with the prosecution” or have a “close working relationship”). Here, it appears that there may have been a “close working relationship” between state agents and the federal prosecution. The federal prosecution itself seemed to concede as much in its September 8, 2004 response to the District Court. But we cannot know without additional fact finding.

The last question, asked in Perdomo, is whether impeachment information was readily available to the prosecution. See, e.g., Kasi v. Angelone, 300 F.3d 487, 506 (4th Cir.2002); United States v. Auten, 632 F.2d 478, 481 (5th Cir.1980). Risha argues that information of a pending deal for Caito could have been easily obtained. Caito was represented by the same attorney in the state court prosecution and the federal grant of immunity. Risha claims that the federal prosecutor was obligated simply to ask Caito, Chontos, or Hellein what arrangements or inducements had been made. He submits that “a simple inquiry [would] have yielded the response that Caito’s State court case was being continued so that he could continue to cooperate.” To this end, we note that the government concedes that the federal prosecutor knew that Caito faced unrelated state charges. Risha further alleges that the government knew that Caito’s state case had been continued. The record indicates that impeachment information may well have been readily available to the prosecution, but we have no further findings on this point.

We find apposite a Fifth Circuit case, cited in Perdomo, with facts redolent of those before us — United States v. Antone, 603 F.2d 566 (5th Cir.1979). Antone involved the cooperation of federal and state forces. A federal defendant appealed his conviction on Brady grounds, claiming that a false statement by a witness concealed the fact that the witness’ attorneys’ fees had been paid by the State of Florida. Id. at 567. An investigative task force of federal and state agents had been formed to solve a murder in which the witness was allegedly involved. Id. at 568. It was decided that the witness should be represented by an attorney. Id. A state agent agreed to “take care” of the matter and a lawyer was appointed using state funds. Id. The arrangement was not disclosed to federal agents or prosecutors, and the fee vouchers were not made available. Id.

Still, the state’s knowledge of the attorney’s appointment was imputed to the federal team because the forces had “pooled their investigative energies to a considerable extent” and the effort overall was “marked by [the] spirit of cooperation.” Id. at 569. The Fifth Circuit concluded that, in the context of Brady, “[i]mposing a *306rigid distinction between federal and state agencies which have cooperated intimately from the outset of an investigation would artificially contort the determination of what is mandated by due process.” Id. at 570. The Court adopted a “case-by-case analysis of the extent of interaction and cooperation between the two governments.” Id. We find Antone persuasive and agree that a case-by-case analysis is appropriate.

IV. Conclusion

In sum, a Brady violation may be found despite a prosecutor’s ignorance of impeachment evidence. “This may be especially true when the withheld evidence is under the control of a state instrumentality closely aligned with the prosecution .... ” United States ex rel. Smith v. Fairman, 769 F.2d 386, 391 (7th Cir.1985). It appears that here, at least one state agent was heavily involved in the federal charges against Risha. It also appears possible that federal and state forces engaged in a “joint investigation” to resolve the alleged arson. Last, it is very possible that the impeachment information was “readily available” to the prosecution. However, the District Court did not make such findings.

We think it very possible that Risha must be granted a new trial. However, given the relative competencies of our two courts to decide issues of fact, we conclude that it is most appropriate to remand this case to the District Court to make this fact-driven determination. By way of explanation, we note that much of the detailed factual information that we rely upon was not mentioned by the District Court in its opinion. In fact, a large part of it was not briefed by the parties, but rather found in the extensive record before us.

The District Court must determine, inter alia: (1) did Agent Marraway, Hellein, or any other state actor know, or have reason to know, of a deal or expectation involving the impact of Caito’s federal testimony on the state charges against him?; (2) was any state actor sufficiently involved with the federal case against Risha, or was impeachment information readily obtainable, such that a finding of constructive knowledge is appropriate? This latter question is informed by the factors laid out above.6 If these two inquiries are answered in the affirmative, there was a Brady violation, and Risha is entitled to a new trial.7

The dissent would find that impeachment information was indeed “readily available” to the prosecution and that the prosecution should have undertaken further inquiries. We do not dispute that this is a possible, or probable, conclusion. However, we find that this conclusion should be made by the District Court, which never appeared to contemplate constructive possession in the first instance. Therefore, the order of the District Court granting a new trial will be vacated, and *307the ease remanded for farther proceedings consistent with this opinion.

NYGAARD, J.,

Dissenting.

Although I agree with much of the majority’s discussion, I reach the opposite conclusion. In remanding, the majority poses certain questions to be answered by the District Court. The answers may be helpful; but on this record they are unnecessary. I conclude that the District Judge, who presided over the trial and who saw and heard the critical testimony, was correct in finding that a Brady violation occurred, and that the federal prosecutor should have made reasonable inquiries into the existence of exculpatory information in the concurrent state court proceeding. Hence, I dissent.

I.

It is well settled that a Brady claim must set out three distinct elements: (1) the prosecution must suppress or withhold evidence, (2) that evidence must be favorable, and (3) material to the defense. See United States v. Perdomo, 929 F.2d 967, 970 (3d Cir.1991). Under the first element, a prosecutor’s lack of knowledge concerning exculpatory material does not automatically defeat a Brady claim. Instead, the prosecutor may still be considered to have suppressed evidence if he has not sought out information readily available to him. Thus, in essence the prosecution has an obligation to produce evidence that is constructively in its possession. Constructive possession means that although a prosecutor has no actual knowledge, he should have known that the material at issue was in existence. See United States v. Joseph, 996 F.2d 36, 39 (3d Cir.1993). This requirement discourages the prosecution from behaving disingenuously by turning its head from information that may be exculpatory or undertaking a minimal or sham investigation. Further, it recognizes that often times the government is in a position of superior knowledge with respect to obtaining any exculpatory material. See United States v. Pelullo, 399 F.3d 197, 211 (3d Cir.2005).

II.

In Perdomo, we held that the prosecution’s failure to conduct a basic search of their star witnesses’s criminal background was sufficient to charge them with constructive possession of the exculpatory information they would have found had they undertaken the investigation. In so concluding, we emphasized that, “[i]t is well accepted that a prosecutor’s lack of knowledge does not render information unknown for Brady purposes.” Perdomo, 929 F.2d at 970. Additionally, we “declined to excuse non-disclosure in instances where the prosecution has not sought out information readily available to it.” Id. (citing United States v. Auten, 632 F.2d 478, 481 (5th Cir.1980)). For purposes of constructive possession, then, we established that the prosecution cannot avoid its Brady responsibilities “by failing to take the minimal steps necessary to acquire the requested information,” even where the prosecution was unaware that the material existed. See Joseph, 996 F.2d at 40 (identifying the “linchpin” of Perdomo’s holding that the prosecution should be charged with constructive possession).

The rule we formulated in Perdomo, that a prosecutor is obliged to produce information if such information is readily available to him, however, has undergone some shifts which are important to understand the contemporary framework of the Brady constructive knowledge requirement. Initially, in Perdomo, we clarified that “the availability of [exculpatory or impeachment] information is not measured *308in terras of whether the information is easy or difficult to obtain but by whether the information is in the possession of some arm of the state.” Id at 971. Our statement revealed an awareness that investigation arms of the prosecution team are as responsible as the actual prosecutor in locating and disclosing exculpatory information. See id at 970.

However, in Joseph, we shifted our analysis away from looking only at the location of the information as in Perdomo. There, we refused to charge the prosecution with constructive possession of exculpatory Brady material because the material was located in an entirely unrelated case that bore no relation to the case under prosecution. See Joseph, 996 F.2d at 40. We noted that “this case is very different than Perdomo” because, “unlike the defendant in Perdomo, the appellants did not direct the prosecutor’s attention toward the type of information they were seeking,” which was located in an unrelated case file. Id.

Distinguishing the present situation, we held that the prosecution should not be charged with constructive possession if the material was located in a “file unrelated to the case under prosecution” unless a defendant “make[s] a specific request for that information — specific in the sense that it explicitly identifies the desired material and is objectively limited in scope.” Id at 41. This holding reflected our recognition that “it would be unreasonable to expect the prosecutor to search all unrelated files in his office to look for exculpatory material.” Id at 40. However, we counseled that Perdomo prevented prosecutors from “ignoring] the very records likely to reveal germane information.” Id Thus, while clearly “in possession of some arm of the state,” we reasoned that the amount and scope of material for which a prosecutor would have to be responsible would place an unreasonable burden on prosecutors. See id. Under this new formulation, the scope and amount of information a prosecutor would be required to search, in addition to a request from the defense, are factors in a determination of whether constructive knowledge exists.

Then, as the majority notes, in United States v. Thornton we held that “prosecutors have an obligation to make a thorough inquiry of all enforcement agencies that ha[ve] a potential connection with the[ir] witnesses.” United States v. Thornton, 1 F.3d 149, 158 (3d Cir.1993) (emphasis added). There, the Drug Enforcement Agency (DEA) paid two witnesses in exchange for their testimony. This exculpatory material was not disclosed to the defense and, in an effort to excuse the non-disclosure, the prosecution argued that it had made a disclosure request to the DEA but that the DEA agents made no reply. Id. at 158. We responded that the prosecution’s inquiry was deficient and that, under the constructive possession doctrine, the prosecution was obliged to do more than make a mere request. In sum, while we ultimately ruled that the withheld evidence was not material, we concluded that the prosecution should be required to make thorough inquiries into the existence of any exculpatory information concerning its witnesses. Id

Finally, we limited the scope of the constructive possession requirement in Pelullo when we held that “the prosecution is under no obligation to ferret out evidence from another pending proceeding with a tenuous connection to the prosecution.” Pelullo, 399 F.3d at 217 (quoting United States v. Pelullo, 185 F.3d 863 (3d Cir.1999)). There, we determined that based on the tenuousness of the relationship between the Department of Labor investigation and the prosecution, to expect the prosecution to undertake a massive hunt for any possibly related documents within *309the Department of Labor would be to place an unreasonably onerous burden on the prosecution. See id.

III.

These cases all struggle with the question whether and under what circumstances it would be reasonable to hold the prosecution responsible for not knowing or finding out certain exculpatory information. The contours of this “reasonableness” inquiry have traditionally been quite murky and none of our cases has explicitly developed anything near a clear test. At first, we thought we could draw a bright line around any information contained within any state arm. See Perdomo, 929 F.2d at 971. Later, we recognized that the scope of material such a requirement could possibly encompass — including information from prior unrelated cases going back potentially indefinitely — would place an unreasonable burden on prosecutors, so instead we shifted our inquiry to whether the defendants made a request for the information and how much information the prosecutor would have been responsible for if he was to try and learn about the exculpatory information.

In light of these cases, it would be fair to say that a constructive knowledge test hinges on a number of different factors, including: (1) the location of the information; (2) the size and scope of the investigation it would take to uncover the information; (3) actions taken by the defense in asking for specific materials; and (4) the connection or relationship between the instant case and the proceeding in which the material is located.

The key to harmonizing these factors, then, would be to articulate a test that could accurately encompass the factors without losing sight of the desire to hold prosecutors responsible for disclosing information they should be able to discover. This test might best be described as a reasonableness test, utilizing the different factors to help determine the reasonableness of the prosecution’s behavior and might look something like this:

In order to establish a Brady violation based on constructive knowledge, the defendant must establish that: (1) the prosecution was put on notice either through specific defense requests for information or, under the circumstances, that exculpatory information may possibly exist; (2) once the prosecution is put on notice, it must take objectively reasonable steps to discover the potentially exculpatory information.8 Factors that should be considered include: (a) the location of the information; (b) the size and scope of the investigation it would take to uncover the information; and (c) the connection or relationship between the instant case and the proceeding in which the material is located; and (3) if the prosecution fails to take these objectively reasonable steps, constructive knowledge of any exculpatory information will be attributed to the prosecution.

IV.

Applying this reasonableness test and the factors previously identified to the facts in the instant case, the District Court did not err by finding that the prosecution had violated Brady. First, the prosecution was put on notice that possibly exculpatory or impeachment evidence was in existence when, during cross-examination, the de*310fense vigorously questioned Caito as to his agreements with and postponement of the state prosecution of his separate weapons offense. The repeated questioning by the defense on the connection between Caito’s testimony in the federal prosecution and the possible effect it might have on his state criminal prosecution should have alerted the prosecution to the possibility that some connection did, in fact, exist.

This is certainly not a situation that would require the prosecution to unreasonably “infer” that exculpatory information might exist like in Joseph where we held that, due to the defendant’s failure to alert the prosecution to the possibility that exculpatory information existed, there was no way the prosecution could have known of its existence. See Joseph, 996 F.2d at 40. There, the information was located in an old case file that was entirely unrelated to the prosecution’s current case. Id. at 39. Here, the defense’s continued questioning of the prosecution’s star witness over the effect his testimony would have on his state court trial clearly notified the prosecution that this information was germane to the case. Additionally, the material was located in a concurrent case and so was, in some sense, concurrently being produced. In sum, the prosecution here was put on notice as to the location and nature of the possible exculpatory or impeachment evidence.

Second, the “sporting theory” of justice has no place in criminal law. As the District Judge obviously recognized, it would have been objectively reasonable for the prosecution to undertake an inquiry into the possibility that Caito received some benefit from his testimony in Risha’s federal prosecution. Amazingly, the prosecution failed to undertake any inquiry into the existence of exculpatory information related to their star witness, Caito. Despite their knowledge that Caito faced concurrent state charges, they did not investigate into the disposition of those charges as it might have related to his help in the federal case against Risha. To permit this to go unchecked would be to encourage a “don’t ask, don’t tell” deliberate ignorance. The location of the information was easily identifiable and accessible. Had the prosecution merely asked Caito, his lawyer Chontos, or the Assistant Attorney General Hellein what arrangements or inducements had been made in Caito’s state case, the prosecution would have discovered the exculpatory material. Moreover, the connection between the two cases was not tenuous, but rather strongly related, as they both involved the government’s star witness, and the scope of the investigation would have been minimal — the prosecution merely could have asked Caito or Chontos, Caito’s lawyer, what arrangements or inducement had been made.

It is true, as the majority notes, that we have refused to charge the prosecution with constructive possession where they would be required to undertake a “fishing expedition” in other jurisdictions for exculpatory information. Maj. Op. at 304-05. Thus, prosecutors are not required to “learn of all information possessed by other government agencies that have no involvement in the investigation or prosecution at issue.” Id. (citations omitted). However, this case does not present a situation even remotely resembling the “fishing expedition” the majority apprehends. Instead, the exculpatory information could have been found had the prosecution made only the most cursory inquiry into their star witnesses’s concurrent state case. The information did not lie in some dark cavernous warehouse, nor was it possessed by an arcane wing of an unrelated agency. As Perdomo, Joseph, and Thornton teach, the prosecution must make reasonable inquiries into the existence of exculpatory material surrounding one of *311their witnesses. A failure to do so will not excuse the prosecution from being charged with a Brady violation if exculpatory material is later found. In all, to charge the prosecution here with constructive possession merely recognizes the well-accepted rule that the prosecution must investigate that which is likely to reveal germane information. See Joseph, 996 F.2d at 40.

V.

The majority excuses the prosecution’s failure to undertake any investigation or inquiry whatsoever into its star witness’s state case, despite having knowledge about it. This is precisely the sort of reverse incentive that motivated us to craft a constructive possession doctrine in the first place. See Perdomo, 929 F.2d at 970 (“To do otherwise would be inviting and placing a premium on conduct unworthy of representatives of the United States Government.”) (quoting Auten, 632 F.2d at 481).

It is certainly true that the other factors the majority discusses may reveal an even deeper fault attributable to the prosecution. For instance, it is likely that the federal and state agencies were engaged in a “joint investigation” to resolve the alleged arson and, additionally, that at least one member of the state investigation, Agent Marraway, was acting on behalf of the prosecution. But for me, these will only serve as a further indictment of the prosecution’s already inexcusable behavior and are unnecessary, under our current constructive possession requirement, to charge the prosecution with constructive possession where the prosecution failed utterly to make the most basic inquiry into an area that might reveal germane and exculpatory information.

I conclude the District Court was correct in determining that a Brady violation had occurred because the prosecution should have reasonably undertaken an inquiry into the impact that Caito’s testimony would have had on his state court proceeding. Fundamental fairness demands no less than this. Because the prosecution failed to do this, they should be charged with constructive knowledge of the Brady material, thereby satisfying the first and only contested prong in the Brady analysis. The order of the District Court granting Risha’s motion for a new trial should be affirmed.