with whom PREGERSON, REINHARDT, W. FLETCHER, FISHER and PAEZ, Circuit Judges, join, dissenting from the denial of rehearing en banc:

I respectfully must note my strong disagreement with my colleagues who voted against rehearing this appeal en banc. The issue is one of exceptional importance to the administration of justice in criminal proceedings: the scope of the work product privilege in criminal discovery. 1 The panel majority ignores the plain meaning of the Federal Rules of Criminal Procedure, as Judge W. Fletcher ably details in his dissenting opinion. United States v. Fort, 472 F.3d 1106, 1122-31 (9th Cir.2007). Its interpretation of Rule 16 significantly alters the landscape of criminal discovery. In the course of its interpretative voyage, the panel majority works two sea-changes in existing law. First, it fashions from whole cloth a retroactive theory of agency *1101 between local and federal officials. Second, its holding as to the scope of the work product privilege directly conflicts with Supreme Court precedent, circuit court precedent, and the way prosecutors, defenders and district courts apply Rule 16 on a daily basis. The panel majority’s opinion threatens to reduce prosecutorial transparency in criminal prosecutions, provides tools for discovery gamesmanship, unwittingly hampers prosecutors by creating traps for reversible Brady error, and increases the costs and burdens on criminal defendants, much of which is born by the federal government. The panel majority’s opinion also has far-reaching effects, touching a vast number of criminal prosecutions in the Ninth Circuit. This is exactly the type of exceptional case that warrants rehearing en banc.

1. Background

This interlocutory appeal arises from a federal racketeering and drug conspiracy prosecution of some alleged members of the “Down Below Gang,” which operates in public housing projects in San Francisco. The eighty-six count second superseding indictment, which includes one hundred and three predicate acts as part of the racketeering and drug conspiracies, names eleven defendants. Eight defendants entered plea agreements following the issuance of the panel majority’s original unreasoned order that vacated the district court’s discovery rulings. The government has filed notices of intent to seek the death penalty against defendants Emile Fort and Edgar Diaz.

The predicate acts were investigated by local police authorities for years before there was ever any federal involvement. As a result, there are thousands of pages of local police reports critical to both the prosecution and defense of the case. The Down Below Gang, like many violent criminal organizations, has a documented history of using violence and intimidation to thwart investigators and prosecutors. Citing witness protection concerns, federal prosecutors have steadfastly refused to turn over unredacted versions of the local police documents in their possession. The redactions in those documents include witness names and both inculpatory and exculpatory evidence. 2

The district court found that the local police reports were “material to preparing the defense,” under Federal Rule of Criminal Procedure 16(a)(1)(E), and ordered the prosecutors to turn the reports over in unredacted form. However, to ensure witness safety, using its authority under Rule 16(d), the district court limited disclosure of the documents, fashioning a nuanced and thorough protective order. Nevertheless, prosecutors adamantly refused to produce the local police reports, asserting the work product protections of Rule 16(a)(2). 3 The district court unsurprisingly held that the Rule 16(a)(2) privilege applies only to documents made by agents of the federal government in connection with investigating or prosecuting the case. *1102 The district court further deemed local police officers, who at the time of the investigation of the case were working with federal investigators, federal “agents” for the purposes of the Rule 16(a)(2) privilege. However, such a joint-investigation showing, which would have justified the claim of privilege, was never made by the federal prosecutors. The district court therefore ordered evidentiary sanctions against the prosecutors for flouting its discovery holdings, and the government appealed.

II. Criminal Discovery

“There is no general constitutional right to discovery in a criminal case.” Weatherford v. Bursey, 429 U.S. 545, 559 , 97 S.Ct. 837 , 51 L.Ed.2d 30 (1977). Over time a patchwork of statutory and judge-made rules has evolved to govern criminal discovery. Those rules include: (1) the Jencks Act; 4 (2) the Federal Rules of Criminal Procedure; and (3) Brady v. Maryland, 373 U.S. 83 , 83 S.Ct. 1194 , 10 L.Ed.2d 215 (1963).

i. The Rules of Criminal Procedure

Rule 16 of the Federal Rules of Criminal Procedure governs criminal discovery. It has been significantly expanded since the Rules were first adopted, and it now imposes discovery obligations both on the government and the defense. See Fed. R.Crim.P. 16 (1966 Amendment); id. (1974 Amendment); id. (1975 Enactment). Unlike civil discovery, where some materials are automatically produced as a matter of right, in criminal discovery, the defendant must invoke the right to discovery. Compare Fed.R.Crim.P. 16(a)(1) with Fed.R.Civ.P. 26(a). Once a defendant makes a Rule 16 discovery request and the government complies, the government is entitled to seek reciprocal discovery from the defendant. See Fed.R.Crim.P. 16(b)(1). Work product of the United States and of the defense are not subject to discovery under Rule 16. See Fed. R.Crim.P. 16(a)(2), 16(b)(2).

Among the items that the government must produce upon request are:

Fed.R.Crim.P. 16(a)(1)(E). In the Ninth Circuit, federal prosecutors are deemed to have “possession and control” over material in the possession of other federal agencies as long as they have “knowledge” of and “access” to that material. United States v. Santiago, 46 F.3d 885, 893 (9th Cir.1995). Thus, for the purposes of Rule 16, possession can be imputed to federal prosecutors even if they do not physically possess the materials.

ii. Brady

Though not technically a rule of criminal discovery, Brady and its progeny play an important role in the process. Discovery decisions made by prosecutors must always be made with one eye on successful prosecution and the other on the potential for reversible error for failure to disclose information. See Brady, 373 U.S. at 87 , 83 *1103 S.Ct. 1194 (“[T]he suppression by the prosecution of evidence favorable to an accused upon request violates due process where the evidence is material either to guilt or to punishment, irrespective of the good faith or bad faith of the prosecution.”). The prosecutor’s duty to disclose information includes a duty to disclose information known to other agents of the government. See Giglio v. United States, 405 U.S. 150, 154 , 92 S.Ct. 763 , 31 L.Ed.2d 104 (1972) (citing Restatement (Second) of Agency, and noting that prosecutors speak for the government as a whole). This principle was reiterated in the context of a state court prosecution in Kyles v. Whitley, 514 U.S. 419, 433 , 115 S.Ct. 1555 , 131 L.Ed.2d 490 (1995) (imputing knowledge of state police investigators to the prosecutor). “[E]vidence is material ‘if there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different.’ ” Strickler v. Greene, 527 U.S. 263, 280 , 119 S.Ct. 1936 , 144 L.Ed.2d 286 (1999) (quoting United States v. Bagley, 473 U.S. 667, 682 , 105 S.Ct. 3375 , 87 L.Ed.2d 481 (1985)).

Rule 16 is in essence a procedural prophylactic for Brady error. “Generally, Brady violations first come before a court after the trial and the court may then consider the materiality of what was suppressed or omitted from disclosures made, in the context of the complete trial record. Indeed, it is not possible to apply the materiality standard in Kyles before the outcome of the trial is known.” United States v. McVeigh, 954 F.Supp. 1441, 1449-50 (D.Colo.1997). 5 The panel majority’s opinion and its impact must be evaluated through the Brady /Rule 16 lens.

III. Panel Majority’s Decision

There is no dispute that the local police reports at issue qualify under Rule 16(a)(1)(E) as documents “material to preparing the defense.” Thus, the only question before the panel was whether the local police reports qualified as privileged work product. This required the panel to interpret Rule 16(a)(2):

Fed.R.Crim.P. 16(a)(2). “The meaning of Rule 16(a)(2) is so plain that it should be unnecessary to do anything more than simply read the text in order to conclude that it does not protect documents prepared by the San Francisco Police Department without any involvement by the federal government.” Fort, 472 F.3d at 1124 (W.Fletcher, J., dissenting).

Nonetheless, the panel majority dissected the language of Rule 16 and imbued it with new meaning. The panel majority first concluded that the word “government” refers to the federal government. Id. at 1111 . It then concluded that the term “ ‘government agent’ includes non-federal personnel whose work contributes to a federal criminal ‘case.’ ” Id. at 1113 . Finally, it held that the phrase “in connection with investigating or prosecuting the case,” is so broad as to include any such *1104 work by any “government agent,” at any time, even before there is a federal case. Id. at 1119-20 . This combination of rulings transforms local police officers involved in local investigations years before any federal prosecution was even contemplated into federal “government agents.” Id. As a result, all “reports, memoranda, or other internal government documents” generated by the San Francisco Police Department while investigating the Down Below Gang magically became material not subject to discovery under Rule 16 at all, 6

IV. Consequences

In bypassing the plain meaning of Rule 16(a)(2), the panel majority ignored well-established canons of statutory construction, relied on inapposite cases and commentaries, and failed to consider the far-reaching consequences of its ruling, especially its redefinition of agency. Judge W. Fletcher’s dissent expertly provides a proper textual analysis of Rule 16, id. at 1123-26 , and explains the fallacies in the panel majority’s reasoning, id. at 1126-31 . That said, we do not rehear en banc every appeal where the panel gets it wrong. See Fed. R.App. P. 35(a) (limiting en banc rehearing to “question[s] of exceptional importance” and conflicts with Supreme Court or existing Ninth Circuit precedent). There are numerous reasons that this appeal meets these criteria and should have been reheard en banc.

i. Work Product Rule

One of the major justifications underpinning the panel majority’s expansion of the scope of Rule 16(a)(2) is that the rule was intended “to be a rule of discovery, related to the work product doctrine but not synonymous or coextensive with it.” Fort, 472 F.3d at 1116 . The panel majority’s novel assertion that Rule 16(a)(2) is not a work product exception is directly contrary to the way that Rule 16 has consistently been interpreted by federal courts. In United States v. Armstrong, 517 U.S. 456, 463 , 116 S.Ct. 1480 , 134 L.Ed.2d 687 (1996), the Supreme Court noted that “under Rule 16(a)(2), [a defendant] may not examine Government work product in connection with his case.” See United States v. Nobles, 422 U.S. 225 , 239 & n. 13, 95 S.Ct. 2160 , 45 L.Ed.2d 141 (1975) (describing Rule 16(a)(2) as embodying the work product exception). That Rule 16(a)(2) is a work product exception has been reiterated by our and other courts of appeal. See, e.g., Virgin Islands v. Fahie, 419 F.3d 249, 257 (3d Cir.2005) (“The exception in Rule 16(a)(2) applies to work product. The ATF Report was not government work product of a type exempted from discovery. It did not contain mental impressions, conclusions, opinions or legal theories concerning litigation of an attorney or other representative of a party.”) (citations and quotations omitted); United States v. Fernandez, 231 F.3d 1240, 1247 (9th Cir.2000) (“Rule 16 of the Federal Rules of Criminal Procedure recognizes the work product privilege.”).

ii. Vast Expansion of Government Agency

The panel majority relies on its mistaken notion that physical possession of materials by the federal government is the touchstone of discoverability analysis under Rule 16. Fort , 472 F.3d at 1118— 19(discussing United States v. Gatto, 763 F.2d 1040 (9th Cir.1985)). Our law is to the contrary. We have held that “[t]he prosecutor will be deemed to have knowl *1105 edge of and access to anything in the possession, custody or control of any federal agency participating in the same investigation of the defendant.” United States v. Bryan, 868 F.2d 1032, 1036 (9th Cir.1989) (discussing Rule 16(a)(1)(C), which has since been moved to Rule 16(a)(1)(E)); see also Santiago, 46 F.3d at 895 . 7 Rule 16 discoverability analysis thus hinges on federal government agency, not on federal government possession. If local authorities are “government agents” for the purposes of 16(a)(2) privilege, then they must also be government agents for the purposes of 16(a)(1) production. The panel majority’s broad expansion of “government agents” portends a number of deleterious downstream effects.

1.Expanded Prosecution Claims of Privilege

Here, the documents at issue are local police reports. But, the panel majority does not draw any lines as to the type of state or local agency that may be deemed an after-the-fact agent of the federal government under Rule 16. So, depending on the prosecution, its holding readily embraces reports made by state environmental or business regulation bodies long before a prosecution is even contemplated by the federal government. 8 Those reports need not contain any mental impressions particularly geared towards federal prosecution of a particular case under the panel majority’s rule, but rather must simply be made “in connection with investigating or prosecuting the case.” This means that many documents that were previously discoverable by the defense can now be flagged as privileged work product.

2. Defense Claims of Privilege

What’s sauce for the goose is sauce for the gander. Put another way, the panel majority’s opinion means that a defendant may invoke the retroactive agency holding to .make any individual who has material relevant to the defense case an after-the-fact agent, and similarly lock up “reports, memoranda, or other documents made by ... the defendant’s ... agent, during the case’s investigation or defense” as privileged work product under Rule 16(b)(2). The symmetrical structure of Rules 16(a) and 16(b) demands that the panel majorb ty’s novel interpretation of agency applies for defendants as well as for the government. This will lead to broad invocations of privilege designed by defendants to thwart criminal prosecution by locking up as much potentially relevant material as possible.

3. Brady Implications

The panel majority asserts that “nothing in this opinion should be interpreted to diminish or dilute the government’s Brady obligations.” Fort, 472 F.3d at 1110 . Indeed, the panel majority’s opinion has the exact opposite effect: it enhances the scope of the federal government’s Brady obligations. The extension of Brady to knowledge not personally held by the prosecutor has been driven by theories of government agency. See Giglio, 405 U.S. at 154 , 92 S.Ct. 763 (applying agency principles to prosecutors as spokesmen for the federal government). As noted previously, Rule 16 and Brady are in many ways two sides of the same coin. If a local agency is a “government agent” for Rule 16 purposes, it should also be deemed an agent *1106 for Brady purposes. This extends the federal government’s Brady duties to include information in the control of local agencies that participated in the “case.”

And, in turn, the scope of the prosecutor’s policing responsibilities with respect to them non-federal information sources is now greatly extended. While the DOJ likely has internal “procedures and regulations ... to carry that[5ro<% ] burden and to insure communication of all relevant information [held by DOJ appendages] on each case to every lawyer who deals with it,” Giglio, 405 U.S. at 154 , 92 S.Ct. 763 , it seems unlikely that any prosecutor would want the additional duty of ferreting out any and all relevant Brady material from every state or local (or foreign) agency that provides investigative material to the federal government. The panel majority’s opinion expands the government’s Brady obligations into a space where even the best-intentioned prosecutor will never be able to comply.

4. Conflicting Doctrines of Agency

The conduct of federal agents is constrained by laws and rules that do not apply to local officials or private parties. To ensure that federal agents cannot avoid those rules by using straw men, we have applied agency concepts to extend their applicability. For example, Federal Rule of Criminal Procedure 41 governs search and seizure by federal agents, and imposes various restrictions related to warrants. We have extended the strictures of Rule 41 to local officials only when they perform a search that is “federal in character,” that is, “from the beginning it was assumed a federal prosecution would result.” United States v. Palmer, 3 F.3d 300, 303 (9th Cir.1993). This interpretation of who is a federal agent sensibly takes into consideration the temporal aspect of the relation between the investigating official and the prosecution. Thus, only if there is some reason at the time of the search to deem it a federal search does Rule 41 apply to searches performed by local officials.

The panel majority’s definition of a “government agent” ignores examples like Rule 41 and runs roughshod over the traditional tenet of agency that a principal-agent relationship must exist at the time of the agent’s actions in order to ascribe culpability to the principal. See, e.g., United States v. Veatch, 674 F.2d 1217, 1221-22 (9th Cir.1981) (“A private person cannot act unilaterally as an agent or instrument of the state; there must be some degree of governmental knowledge and acquiescence. In the absence of such official involvement, a search is not governmental.”) (quoting United States v. Sherwin, 539 F.2d 1, 6 (9th Cir.1976) (en banc)); United States v. Birnstihl, 441 F.2d 368, 370 (9th Cir.1971) (per curiam) (holding that no Miranda warnings were required because “[t]he evidence was completely inadequate from which to conclude that the [private security] guard in this ease was an actual or ostensible agent of the police”); see also United States v. Jones, 231 F.3d 508, 517 (9th Cir.2000) (“A person is a government agent when the government authorizes, directs and supervises that person’s activities and is aware of those activities.”) (quotation and citations omitted).

iii. Practical Concerns

1. Encouraging Gamesmanship

Rule 16 prevents criminal trials from becoming “a sporting contest in which the game of cat and mouse is acceptable.” United States v. Howell, 231 F.3d 615, 626 (9th Cir.2000); see also Berger v. United States, 295 U.S. 78, 88 , 55 S.Ct. 629 , 79 L.Ed. 1314 (1935) (“The United States Attorney's] ... interest, therefore, in a criminal prosecution is not that it shall win a case, but that justice shall be done.”). When defendants invoke Rule 16, they open themselves up to discovery to gain access to information held by the govern *1107 ment; thus, they invariably give the government some inkling of the defense theory of the case. Through this process, prosecutors become more informed about their own case and the evidence in their possession. This interplay during criminal discovery also often causes prosecutors to reevaluate the possible Brady relevance of material. In a criminal prosecution like this, where there are thousands of pages of local police documents, it is even more likely that prosecutors will be unaware of the potential importance of materials that they, or the local police, may possess. The panel majority drives a significant wedge into the discovery process. First, the defense is unlikely to even ask for discovery if the government is entitled to the broad privilege created by the majority. If the government is aggressive in labeling material “privileged,” little material of any use to the defense is likely to be turned over in any event. Second, even if the defense does participate, it will now be able to assert a similar privilege for materials garnered from “agents.”

As a practical matter, in a complex case, it is impossible for prosecutors to mentally catalog everything in their possession and predict how its Brady relevance may be affected by the defense presented at trial. For example, each of the one hundred and three predicate acts for just two of the eighty six counts in the indictment is presumably supported by a variety of witnesses, physical evidence, and other material, as well as police reports and investigatory documents prepared by the local police. Without some back and forth with the defense, the prosecution will not even know where to look for Brady material. As a result, avoiding Brady error will depend solely on prose-cutorial epiphany when the defense puts on its case. 9

This is a recipe for mistrial or reversal on appeal — a result that delays justice, and wastes prosecutorial resources, judicial resources, and jury time, particularly in complex criminal cases (the very cases most affected by the panel majority’s opinion). These expanded Brady obligations exacerbate the potential for error, since federal prosecutors must also disclose Brady material in the possession of local authorities now deemed “government agents.”

2. Stripping Case Management Power from District Courts

Rule 16(d) gives district courts the authority to control criminal discovery and to enter protective orders as needed. That is precisely what the able and experienced district court judge sought to do here. That power has proven adequate to allow for the fair trial in this circuit of members of other organizations that may have threatened witness safety, including violent criminal gangs such as the Aryan Brotherhood and the Mexican Mafia. But here, the government refused to cooperate on the novel and incorrect theory of privilege now endorsed by the panel majority. This is a classic example of over-aggressive prosecutors seeking a short-term advantage without fully contemplating the potential negative impact on the success of their own future prosecutions. 10 The ma *1108 jority opinion whisks away from the district court the discretion to decide what material is important to a fair trial and vests it in the prosecution. In this ease, the panel majority’s intervention halted a complex criminal trial in its tracks and thwarted the ability of the trial court to manage its own trial.

3. Broad Impact

The panel majority’s opinion will affect a sizeable number of criminal defendants in this circuit. For most judicial districts in the Ninth Circuit, between thirty and sixty percent of criminal prosecutions assigned to the Federal Public Defender’s office originated with or involve local law enforcement. 11 In those cases, federal prosecutors can now assert a work product privilege over any and all investigatory materials acquired from local authorities and refuse to turn them over to the defense. This will force defense counsel to dedicate more resources to tracking down that information through means other than Rule 16. Ultimately, the increased time and effort spent in criminal discovery will affect the quality and cost of criminal defense. The effectiveness of defense counsel will be decreased by the reduced access to information. Costs under the Criminal Justice Act, paid from the public treasury, will increase as counsel are forced to pursue other avenues to acquire material already in the hands of prosecutors.

V. Conclusion

In addition to deciding an exceptionally important issue in a way that directly conflicts with every controlling authority, the panel majority’s opinion is just plain wrong. It gets the textual interpretation wrong. It gets the policy analysis wrong. It deviates from Supreme Court and circuit law on agency and work product privilege. It hamstrings district court judges. It condones trial by ambush in a capital case. It increases the costs of defending and trying complex criminal cases. It hurts the prosecution, the defense, and the quest for truth. For those reasons this appeal merited rehearing by our en banc court and an ultimate ruling that affirms the district court.

. Nine Federal Public Defenders representing the Central, Eastern, and Northern Districts of California, the District of Montana, the District of Nevada, the Eastern and Western Districts of Washington, the District of Idaho, the District of Guam, and the District of Oregon jointly filed an amicus brief in favor of rehearing en banc, citing the effect of the panel majority's opinion on a significant percentage of all pending and future criminal prosecutions in the Ninth Circuit.

. The panel majority insists that the documents include no exculpatory information. This is simply not true. As District Court Judge Alsup noted in one of his discovery orders in this case, "[T]he names of all witnesses ... have been redacted [by the prosecutors] — even as to witnesses wholly or largely exculpatory to one or more defendants." See United States v. Diaz, No. 05-0167 (N.D.Cal. May 18, 2006) (Order re Discovera-bility of Local Police Reports under Rule 16 and Schedule for Further Submissions) (Docket No. 367) at 1.

. Witness protection was the government’s primary justification for its recalcitrant conduct; it is unclear why, for non-testifying civilians, the government did not simply assert its privilege under Roviaro v. United States, 353 U.S. 53, 59 , 77 S.Ct. 623 , 1 L.Ed.2d 639 (1957).

. The Jencks Act requires prosecutors to turn over to the defense statements made by testifying witnesses if those statements are in the prosecutor’s possession. See 18 U.S.C. § 3500 ; Fed R.Crim. P. 26.2; United States v. Polizzi, 801 F.2d 1543, 1552 (9th Cir.1986).

. The interplay between these rules is thoughtfully explored by Judge Matsch in his discovery order from the capital trial of Timothy McVeigh.

. There is no criminal discovery analog to the "substantial need” safety valve to compel discovery of “fact work product.” See Fed. R.Civ.P. 26(b)(3). In the absence of a Jencks Act or Brady compulsion to divulge such materials, the only way a defendant can acquire them is through prosecutorial largesse.

. The panel majority’s focus on physical possession is misguided, and even if it were correct, would lead to the bizarre result that the local police reports become privileged work product only if they are physically handed to federal prosecutors, but remain unprivileged if they are not.

. There is no reason under the panel majority's opinion that agents of another sovereign, such as Mexico or Canada, could not also retroactively become federal agents for the purpose of invoking the work product privilege.

. This is invariably true for some Brady material related to wholly unanticipated defense theories, but typically Brady material will be found before trial if there is meaningful discovery.

. The district court made a number of highly critical findings about the conduct of the prosecutors, which the panel majority fails to address. See United States v. Diaz, No. 05-0167 (N.D.Cal. July 20, 2006) (Testimony-Preclusion Order as Sanction for Noncompliance with Rule 16 Order) (Docket No. 578) at 6 ("[Djefense counsel still remain in the dark' — and are intended by the government to remain in the dark — as to the source of the potential evidence.”) (emphasis in original); *1108 id. at 12 ("Despite the Court's considerable respect for government counsel herein, this order must find that they have acted willfully to abridge Rule 16 rights to gain trial advantage over the defense.”).

. This estimate of potential cases affected throughout the Ninth Circuit is set forth in the Federal Public and Community Defenders's amicus brief in support of rehearing en banc.