TRAXLER, Chief Judge:

Jeffrey Sterling is a former CIA agent who has been indicted for, inter alia, the unauthorized retention and disclosure of national defense information, in violation of the Espionage Act, 18 U.S.C. § 793(d) & (e). The indictment followed the grand jury's probable cause determination that Sterling illegally disclosed classified information about a covert CIA operation pertaining to the Iranian nuclear weapons operation to James Risen, for publication in a book written by Risen, and that he may have done so in retaliation for the CIA's decision to terminate his employment and to interfere with his efforts to publish such classified information in his personal memoirs. Prior to trial, the district court made three evidentiary rulings that are the subject of this appeal. We affirm in part, reverse in part, and remand for further proceedings.

I. Background

A.

According to the indictment, Defendant Jeffrey Sterling was hired as a CIA case officer in 1993, and granted a top secret security clearance. As a condition of his hire, and on several occasions thereafter, Sterling signed agreements with the CIA explicitly acknowledging that he was not permitted to retain or disclose classified information that he obtained in the course of his employment, without prior authorization from the CIA, and that doing so could be a criminal offense.

In November 1998, the CIA assigned Sterling to a highly classified program intended to impede Iran's efforts to acquire or develop nuclear weapons ("Classified Program No. 1"). Sterling also served as the case officer for a covert asset ("Human Asset No. 1") who was assisting the CIA with this program. In May 2000, Sterling was reassigned and his involvement with Classified Program No. 1 ended.

In August 2000, shortly after Sterling's reassignment and after being told that he had not met performance targets, Sterling filed an equal opportunity complaint alleging that the CIA had denied him certain assignments because he was African American. The EEO office of the CIA investigated Sterling's complaint and determined that it was without merit. In August 2001, Sterling filed a federal lawsuit against the CIA alleging that he had been the victim of racial discrimination, and seeking monetary compensation. Several settlement demands were rejected, and the lawsuit was dismissed in March 2004, following the government's invocation of the state secrets doctrine. We affirmed the dismissal. See Sterling v. Tenet, 416 F.3d 338, 341 (4th Cir.2005).

Sterling was officially terminated from the CIA on January 31, 2002, but he had been "outprocessed" and effectively removed from service in October 2001. As part of his termination, Sterling was asked to sign a final acknowledgment of his continuing legal obligation not to disclose classified information. Sterling refused.

On November 4, 2001, James Risen published an article in The New York Times, under the headline "Secret C.I.A. Site in New York Was Destroyed on Sept. 11." J.A. 655. A "former agency official" was cited as a source. J.A. 655. In March 2002, Risen published an article about Sterling's discrimination suit in The New York Times, under the headline "Fired by C.I.A., He Says Agency Practiced Bias." J.A. 156, 725. The article states that Sterling provided Risen with a copy of one of his CIA performance evaluations, which is identified as a classified document. The article also states that Sterling "relished his secret assignment to recruit Iranians as spies." J.A. 156.

In January 2002, in accordance with his non-disclosure agreements with the CIA, Sterling submitted a book proposal and sample chapters of his memoirs to the CIA's Publications Review Board. The Board expressed concerns about Sterling's inclusion of classified information in the materials he submitted.

On January 7, 2003, Sterling contacted the Board and expressed "extreme unhappiness" over the Board's edits to his memoirs, and stated that "he would be coming at ... the CIA with everything at his disposal." J.A. 35-36 (internal quotation marks and alterations omitted). On March 4, 2003, Sterling filed a second civil lawsuit against the CIA, alleging that the agency had unlawfully infringed his right to publish his memoirs. The action was subsequently dismissed by stipulation of the parties. See Sterling v. CIA, No. 1:03-cv-00603-TPJ (D.D.C. July 30, 2004).

The day after he filed his second civil suit, Sterling met with two staff members of the Senate Select Committe e on Intelligence ("SSCI") and raised, for the first time, concerns about the CIA's handling of Classified Program No. 1, as well as concerns about his discrimination lawsuit.1 According to a SSCI staff member, Sterling "threatened to go to the press," although it was unclear "if Sterling's threat related to [Classified Program No. 1] or his lawsuit." J.S.A. 29.

Telephone records indicate that Sterling called Risen seven times between February 27 and March 29, 2003. Sterling also sent an e-mail to Risen on March 10, 2003 — five days after his meeting with the SSCI staff — in which he referenced an article from CNN's website entitled, "Report: Iran has `extremely advanced' nuclear program," and asked, "quite interesting, don't you think? All the more reason to wonder ..." J.A. 37, 726; J.S.A 31.

On April 3, 2003, Risen informed the CIA and the National Security Council that he had classified information concerning Classified Program No. 1 and that he intended to publish a story about it in The New York Times. In response, senior administration officials, including National Security Advisor Condoleezza Rice and Director of the CIA George Tenet, met with Risen and Jill Abramson, then Washington Bureau Chief of The New York Times, to discuss the damage that publication would cause to national security interests and the danger to the personal safety of the CIA asset involved in the operation. Several days later, Ms. Abramson advised the administration that the newspaper would not publish the story.

Approximately three months later, Sterling moved from Virginia to Missouri to live with friends. During this time, 19 telephone calls took place between the New York Times' Washington office and Sterling's friends' home telephone number. Sterling's friends denied any involvement in these calls. A forensic analysis of the computer Sterling used during this time revealed 27 e-mails between Sterling and Risen, several of which indicated that Sterling and Risen were meeting and exchanging information during this time period.

Although The New York Times had agreed not to publish information about Classified Program No. 1, Risen published a book, State of War: The Secret History of the CIA and the Bush Administration ("State of War"), in January 2006, which did disclose the classified information. J.A. 721. Specifically, Chapter 9 of the book, entitled "A Rogue Operation," reveals details about Classified Program No. 1. J.S.A. 219-32. In the book, Risen entitled the program "Operation Merlin" and described it as a "failed attempt by the CIA to have a former Russian scientist provide flawed nuclear weapon blueprints to Iran." J.A. 722. Risen does not reveal his sources for the classified information in Chapter 9, nor has he indicated whether he had more than one source. However, much of the chapter is told from the point of view of a CIA case officer responsible for handling Human Asset No. 1. The chapter also describes two classified meetings at which Sterling was the only common attendee.

B.

On December 22, 2010, a federal grand jury indicted Sterling on six counts of unauthorized retention and communication of national defense information, in violation of 18 U.S.C. § 793(d) and (e); one count of unlawful retention of national defense information, in violation of 18 U.S.C. § 793(e); one count of mail fraud, in violation of 18 U.S.C. § 1341; one count of unauthorized conveyance of government property, in violation of 18 U.S.C. § 641; and one count of obstruction of justice, in violation of 18 U.S.C. § 1512(c)(1). Sterling's trial was set to begin on October 17, 2011.

On May 23, 2011, Attorney General Eric Holder authorized the government to issue a trial subpoena seeking Risen's testimony about the identity of his source for information about Classified Program No. 1 and asking Risen to confirm that statements attributed to sources were actually made by those sources. The government also filed a motion in limine to admit Risen's testimony. Risen moved to quash the subpoena and for a protective order, asserting that he was protected from compelled testimony by the First Amendment or, in the alternative, by a federal common-law reporter's privilege.2

The motions were denied in part and granted in part by the district court. The subpoena was "quashed for Risen's testimony about his reporting and source(s) except to the extent that Risen [would] be required to provide testimony that authenticates the accuracy of his journalism, subject to a protective order." United States v. Sterling, 818 F.Supp.2d 945, 947 (E.D.Va.2011). The district court held that Risen had "a qualified First Amendment reporter's privilege that may be invoked when a subpoena either seeks information about confidential sources or is issued to harass or intimidate the journalist," id. at 951 (emphasis added), and that the government could overcome the privilege only by meeting the three-part test that this circuit established for reporters' claims of privilege in civil cases in LaRouche v. National Broadcasting Co., 780 F.2d 1134 (4th Cir.1986). The district court held that, while the information sought was clearly relevant under the first prong of the LaRouche test, the Government had failed to demonstrate that the information was unavailable from other means and that it had a compelling interest in presenting it to the jury.

In addition to the district court's order quashing Risen's trial subpoena, the district court handed down two other evidentiary rulings that are the subject of this appeal. The district court suppressed the testimony of two government witnesses as a sanction for the government's late disclosure of impeachment material under Giglio v. United States, 405 U.S. 150, 92 S.Ct. 763, 31 L.Ed.2d 104 (1972). The district court also denied the government's motion to withhold from Sterling and the jury, pursuant to the Classified Information Procedures Act ("CIPA"), 18 U.S.C.App. 3, the true names and identities of several covert CIA officers and contractors it intends to call to testify at trial.

In a majority opinion written by Chief Judge Traxler, we now reverse the district court's order holding that Risen has a reporter's privilege that entitles him to refuse to testify at trial concerning the source and scope of the classified national defense information illegally disclosed to him (Issue I). In a separate majority opinion written by Judge Gregory, we reverse the district court's order suppressing the testimony of the two Government witnesses (Issue II), and affirm in part and reverse in part the district court's CIPA ruling (Issue III).

TRAXLER, Chief Judge, writing for the court on Issue I:

II. The Reporter's Privilege Claim

We begin with the government's appeal of the district court order quashing the trial subpoena issued to Risen on the basis of a First Amendment reporter's privilege, and Risen's challenge to our jurisdiction to consider this portion of the appeal.

A. Jurisdiction

Risen contends that we lack jurisdiction to consider the district court's ruling under 18 U.S.C. § 3731, because the district court stated that the limitations on Risen's testimony might be reconsidered under the LaRouche test as the testimony developed at trial. We disagree.

Section 3731 provides for interlocutory appeals by the United States of pretrial orders suppressing or excluding evidence upon certification to the district court that the appeal is not taken for the purpose of delay and that the evidence in question is substantial proof of a fact material to the proceedings. We have held that we have jurisdiction under § 3731 even when the district court "repeatedly indicated that its rulings were preliminary and could change as the trial progressed." United States v. Siegel, 536 F.3d 306, 314 (4th Cir.2008); see also United States v. Todaro, 744 F.2d 5, 8 n. 1 (2d Cir.1984) (finding that a conditional suppression order may be immediately appealed by the government under § 3731); cf. United States v. Horwitz, 622 F.2d 1101, 1104 (2d Cir.1980) ("[W]e do not think that the conditional nature of the district court's ruling, which raises the remote prospect that suppression will not be ordered, necessarily deprives this court of jurisdiction under section 3731 to hear the government's appeal.").

While it is true that the district court left itself some room in its order to adjust the scope of Risen's trial testimony, it also made clear that it did not expect to revisit its decision that Risen was entitled to assert a reporter's privilege under the First Amendment and could not be compelled to reveal his sources. Thus, we hold that we have jurisdiction over the appeal. "To conclude otherwise would insulate the district court's ruling from appellate review" because once jeopardy attaches, the Government cannot appeal, "thus frustrating rather than furthering the purposes of § 3731." Siegel, 536 F.3d at 315.

B. The First Amendment Claim

1.

There is no First Amendment testimonial privilege, absolute or qualified, that protects a reporter from being compelled to testify by the prosecution or the defense in criminal proceedings about criminal conduct that the reporter personally witnessed or participated in, absent a showing of bad faith, harassment, or other such non-legitimate motive, even though the reporter promised confidentiality to his source. In Branzburg v. Hayes, 408 U.S. 665, 92 S.Ct. 2646, 33 L.Ed.2d 626 (1972), the Supreme Court "in no uncertain terms rejected the existence of such a privilege." In re Grand Jury Subpoena, Judith Miller, 438 F.3d 1141, 1146 (D.C.Cir.2006).

Like Risen, the Branzburg reporters were subpoenaed to testify regarding their personal knowledge of criminal activity. One reporter was subpoenaed to testify regarding his observations of persons synthesizing hashish and smoking marijuana; two others were subpoenaed to testify regarding their observations of suspected criminal activities of the Black Panther Party.3 All resisted on the ground that they possessed a qualified privilege against being "forced either to appear or to testify before a grand jury or at trial," unless a three-part showing was made: (1) "that the reporter possesses information relevant to a crime," (2) "that the information the reporter has is unavailable from other sources," and (3) "that the need for the information is sufficiently compelling to override the claimed invasion of First Amendment interests occasioned by the disclosure." Branzburg, 408 U.S. at 680, 92 S.Ct. 2646. "The heart of the [reporters'] claim [was] that the burden on news gathering resulting from compelling [them] to disclose confidential information outweigh[ed] any public interest in obtaining the information." Id. at 681, 92 S.Ct. 2646.

Having so defined the claim, the Court proceeded to unequivocally reject it. Noting "the longstanding principle that the public ... has a right to every man's evidence, except for those persons protected by a constitutional, common-law, or statutory privilege," id. at 688, 92 S.Ct. 2646 (internal quotation marks omitted), the Court held as follows:

Id. at 689-90, 92 S.Ct. 2646 (emphasis added); see id. at 690 n. 29, 92 S.Ct. 2646 (noting that "testimonial privileges [are] disfavor[ed] ... since such privileges obstruct the search for truth" and serve as "`obstacle[s] to the administration of justice'" (quoting 8 J. Wigmore, Evidence § 2192 (McNaughton rev. 1961))).

The First Amendment claim in Branzburg was grounded in the same argument offered by Risen — that the absence of such a qualified privilege would chill the future newsgathering abilities of the press, to the detriment of the free flow of information to the public. And the Branzburg claim, too, was supported by affidavits and amicus curiae memoranda from journalists claiming that their news sources and news reporting would be adversely impacted if reporters were required to testify about confidential relationships. However, the Branzburg Court rejected that rationale as inappropriate in criminal proceedings:

Id. at 691, 92 S.Ct. 2646 (emphasis added); see also id. at 690-91, 92 S.Ct. 2646 (noting that there was "no basis for holding that the public interest in law enforcement and in ensuring effective grand jury proceedings is insufficient to override the consequential, but uncertain, burden on news gathering that is said to result from insisting that reporters, like other citizens, respond to relevant questions put to them in the course of a valid grand jury investigation or criminal trial").4

In sum, the Branzburg Court declined to treat reporters differently from all other citizens who are compelled to give evidence of criminal activity, and refused to require a "compelling interest" or other special showing simply because it is a reporter who is in possession of the evidence. Compare id. at 708, 92 S.Ct. 2646 (holding that government need not "demonstrate[] some `compelling need' for a newsman's testimony"), with id. at 743, 92 S.Ct. 2646 (Stewart, J., dissenting) (advocating adoption of the three-part test that includes demonstration of a "compelling and overriding interest in the information").

Although the Court soundly rejected a First Amendment privilege in criminal proceedings, the Court did observe, in the concluding paragraph of its analysis, that the press would not be wholly without protection:

Id. at 707-08, 92 S.Ct. 2646 (majority opinion)(emphasis added)(footnote omitted). This is the holding of Branzburg, and the Supreme Court has never varied from it. As the Court observed nearly two decades later:

University of Pa. v. EEOC, 493 U.S. 182, 201, 110 S.Ct. 577, 107 L.Ed.2d 571 (1990) (internal quotation marks omitted); see also Cohen v. Cowles Media Co., 501 U.S. 663, 669, 111 S.Ct. 2513, 115 L.Ed.2d 586 (1991) ("[T]he First Amendment [does not] relieve a newspaper reporter of the obligation shared by all citizens to respond to a grand jury subpoena and answer questions relevant to a criminal investigation, even though the reporter might be required to reveal a confidential source.").5

The controlling authority is clear. "In language as relevant to the alleged illegal disclosure of the identity of covert agents as it was to the alleged illegal processing of hashish [in Branzburg], the Court stated that it could not `seriously entertain the notion that the First Amendment protects a newsman's agreement to conceal the criminal conduct of his source, or evidence thereof....'" Judith Miller, 438 F.3d at 1147 (quoting Branzburg, 408 U.S. at 692, 92 S.Ct. 2646); see id. at 1165-66 (Tatel, J., concurring) ("If, as Branzburg concludes, the First Amendment permits compulsion of reporters' testimony about individuals manufacturing drugs or plotting against the government, all information the government could have obtained from an undercover investigation of its own, the case for a constitutional privilege appears weak indeed with respect to leaks [of classified information], which in all likelihood will be extremely difficult to prove without the reporter's aid." (citation omitted)). Accordingly, "if Branzburg is to be limited or distinguished in the circumstances of this case, we must leave that task to the Supreme Court." Id. at 1166.

Notwithstanding the clarity of Justice White's opinion for the Court in Branzburg, and the fact that Justice Powell joined that opinion, Risen argues that Justice Powell's concurring opinion in Branzburg should instead be interpreted as a tacit endorsement of Justice Stewart's dissenting opinion, which argued in favor of recognizing a First Amendment privilege in criminal cases that could be overcome only if the government carries the heavy burden of establishing a compelling interest or need. See Branzburg, 408 U.S. at 739, 743, 92 S.Ct. 2646 (Stewart, J., dissenting).

We cannot accept this strained reading of Justice Powell's opinion. By his own words, Justice Powell concurred in Justice White's opinion for the majority, and he rejected the contrary view of Justice Stewart:

Id. at 709-10, 92 S.Ct. 2646 (Powell, J., concurring)(emphasis added).

Justice Powell's concurrence expresses no disagreement with the majority's determination that reporters are entitled to no special privilege that would allow them to withhold relevant information about criminal conduct without a showing of bad faith or other such improper motive, nor with the majority's clear rejection of the three-part compelling interest test advocated by the Branzburg reporters. To the extent Justice Powell addressed any further inquiry that might take place in a criminal proceeding, he appeared to include within the realm of harassment a request that "implicates confidential source relationships without a legitimate need of law enforcement," id. at 710, 92 S.Ct. 2646 (emphasis added), and he again rejected the dissent's contrary view that the heavy burdens of the three-part, compelling interest test were appropriate:

Id. at 710 n.*, 92 S.Ct. 2646 (emphasis added).

For the foregoing reasons, Justice Powell's concurrence in Branzburg simply does not allow for the recognition of a First Amendment reporter's privilege in a criminal proceeding which can only be overcome if the government satisfies the heavy burdens of the three-part, compelling-interest test. Accepting this premise is "tantamount to our substituting, as the holding of Branzburg, the dissent written by Justice Stewart ... for the majority opinion." Storer Commc'ns. v. Giovan (In re Grand Jury Proceedings), 810 F.2d 580, 584 (6th Cir.1987).6 The Branzburg Court considered the arguments we consider today, balanced the respective interests of the press and the public in newsgathering and in prosecuting crimes, and held that, so long as the subpoena is issued in good faith and is based on a legitimate need of law enforcement, the government need not make any special showing to obtain evidence of criminal conduct from a reporter in a criminal proceeding. The reporter must appear and give testimony just as every other citizen must. We are not at liberty to conclude otherwise.

2.

Although Branzburg alone compels us to reject Risen's claim to a First Amendment privilege, we are also bound by our circuit precedent, for this is not the first time we have passed upon the question of whether and to what extent a reporter's privilege can be asserted in criminal proceedings.

a.

In reaching its decision in this case, the district court relied upon our precedent in LaRouche v. National Broadcasting Co., 780 F.2d 1134 (4th Cir.1986). In LaRouche, we considered a civil litigant's right to compel evidence from a reporter and the First Amendment claim of the press to protect its newsgathering activities. We recognized a reporter's privilege in this civil context that could only be overcome if the litigant met the three-part test that the Branzburg Court rejected in the criminal context. Specifically, we held that district courts, before requiring disclosure of a reporter's source in a civil proceeding, must consider "(1) whether the information is relevant, (2) whether the information can be obtained by alternative means, and (3) whether there is a compelling interest in the information." Id. at 1139.

In LaRouche, we followed the lead of other circuits, including the Fifth Circuit in Miller v. Transamerican Press, Inc., 621 F.2d 721, modified, 628 F.2d 932 (5th Cir. 1980), which held that Branzburg did not preclude recognition of a qualified reporter's privilege or application of the three-part test in civil cases. In such cases, of course, "the public interest in effective criminal law enforcement is absent." Zerilli v. Smith, 656 F.2d 705, 711-12 (D.C.Cir.1981).7

b.

LaRouche, however, offers no authority for us to recognize a First Amendment reporter's privilege in this criminal proceeding. Not only does Branzburg preclude this extension, the distinction is critical, and our circuit has already considered and rejected such "a qualified [reporter's] privilege, grounded on the First Amendment, against being compelled to testify in [a] criminal trial." In re Shain, 978 F.2d 850, 851 (4th Cir.1992) (emphasis added).

The Shain reporters were held in contempt for their refusal to comply with subpoenas to testify in the criminal trial of a former state senator whom they had previously interviewed. At the time, two of our sister circuits had extended the three-part test that had been adopted in civil actions to criminal proceedings, albeit with little to no discussion of the Branzburg opinion. See United States v. Caporale, 806 F.2d 1487, 1503-04 (11th Cir. 1986) (citing Miller, 621 F.2d at 726); United States v. Burke, 700 F.2d 70, 76-77 (2d Cir.1983) (citing Zerilli, 656 F.2d at 713-15).

This court in Shain, however, declined to follow that path. We did not recognize a broad privilege nor did we extend the LaRouche three-part test to criminal proceedings. Instead, we followed Branzburg and held that "absent evidence of governmental harassment or bad faith, the reporters have no privilege different from that of any other citizen not to testify about knowledge relevant to a criminal prosecution." Shain, 978 F.2d at 852. We also considered the effect of Justice Powell's concurring opinion in Branzburg, explaining that Justice Powell "joined in the Court's opinion" and wrote separately only

Id. at 853 (emphasis added) (citation omitted) (quoting Branzburg, 408 U.S. at 710, 92 S.Ct. 2646 (Powell, J., concurring)); see id. (citing United States v. Steelhammer, 539 F.2d 373, 376 (4th Cir.1976)) (Winter, J., dissenting), adopted by the court en banc, 561 F.2d 539, 540 (4th Cir.1977) (per curiam) (noting that "[i]n Steelhammer, we applied Branzburg to compel testimony from the press in a civil contempt trial, recognizing that only when evidence of harassment is presented do we balance the interests involved" (emphasis added)).

To the extent our court has addressed the issue since Shain, we have continued to recognize the important distinction between enforcing subpoenas issued to reporters in criminal proceedings and enforcing subpoenas issued to reporters in civil litigation. Subpoenas in criminal cases are driven by the quite different and compelling public interest in effective criminal investigation and prosecution, an interest that simply is not present in civil cases. See Ashcraft v. Conoco, Inc., 218 F.3d 282, 287 (4th Cir.2000) (applying the LaRouche test to confidential source information in the civil context, but noting Branzburg's "holding that [a] reporter, like [an] ordinary citizen, must respond to grand jury subpoenas and answer questions related to criminal conduct he personally observed and wrote about, regardless of any promises of confidentiality he gave to subjects of stories" (emphasis added)).

There is good reason for this distinction between civil and criminal cases. It has roots in both the majority and concurring opinions in Branzburg, both of which highlight the critical importance of criminal proceedings and the right to compel all available evidence in such matters. As the Court has subsequently observed as well:

Cheney v. United States Dist. Court for the Dist. of Columbia, 542 U.S. 367, 384, 124 S.Ct. 2576, 159 L.Ed.2d 459 (2004) (third alteration in original); see also Judith Miller, 438 F.3d at 1149; Smith, 135 F.3d at 972.

3.

Like the Branzburg reporters, Risen has "direct information ... concerning the commission of serious crimes." Branzburg, 408 U.S. at 709, 92 S.Ct. 2646. Indeed, he can provide the only first-hand account of the commission of a most serious crime indicted by the grand jury — the illegal disclosure of classified, national security information by one who was entrusted by our government to protect national security, but who is charged with having endangered it instead. The subpoena for Risen's testimony was not issued in bad faith or for the purposes of harassment. See id. at 707-08, 92 S.Ct. 2646; id. at 709-10, 92 S.Ct. 2646 (Powell, J., concurring). Risen is not being "called upon to give information bearing only a remote and tenuous relationship to the subject of the investigation," and there is no "reason to believe that his testimony implicates confidential source relationships without a legitimate need of law enforcement." Id. at 710, 92 S.Ct. 2646 (Powell, J., concurring). Nor is the government attempting to "annex" Risen as its "investigative arm." Id. at 709, 92 S.Ct. 2646 (internal quotation marks omitted). Rather, the government seeks to compel evidence that Risen alone possesses — evidence that goes to the heart of the prosecution.

The controlling majority opinion in Branzburg and our decision in Shain preclude Risen's claim to a First Amendment reporter's privilege that would permit him to resist the legitimate, good faith subpoena issued to him. The only constitutional, testimonial privilege that Risen was entitled to invoke was the Fifth Amendment privilege against self-incrimination, but he has been granted immunity from prosecution for his potential exposure to criminal liability. Accordingly, we reverse the district court's decision granting Risen a qualified First Amendment reporter's privilege that would shield him from being compelled to testify in these criminal proceedings.

III. The Common-Law Privilege Claim

Risen next argues that, even if Branzburg prohibits our recognition of a First Amendment privilege, we should recognize a qualified, federal common-law reporter's privilege protecting confidential sources.8 We decline to do so.

A.

In the course of rejecting the First Amendment claim in Branzburg, the Supreme Court also plainly observed that the common law recognized no such testimonial privilege:

Branzburg, 408 U.S. at 685, 92 S.Ct. 2646; id. at 693, 92 S.Ct. 2646 ("[T]he evidence fails to demonstrate that there would be a significant constriction of the flow of news to the public if this Court reaffirms the prior common-law and constitutional rule regarding the testimonial obligations of newsmen" (emphasis added)); id. at 698-99, 92 S.Ct. 2646 ("[T]he common law recognized no such privilege, and the constitutional argument was not even asserted until 1958"); Swidler & Berlin v. United States, 524 U.S. 399, 410, 118 S.Ct. 2081, 141 L.Ed.2d 379 (1998) (noting that "Branzburg dealt with the creation of [a] privilege[] not recognized by the common law" (emphasis added)); see also Judith Miller, 438 F.3d at 1154 (Sentelle, J., concurring) (Branzburg is "as dispositive of the question of common law privilege as it is of a First Amendment privilege"); In re Special Proceedings, 373 F.3d 37, 44 (1st Cir.2004) (Branzburg "flatly rejected any notion of a general-purpose reporter's privilege for confidential sources, whether by virtue of the First Amendment or a newly hewn common-law privilege").

B.

Risen does not take issue with the clarity of Branzburg's statements regarding the state of the common law. Rather, he argues that Federal Rule of Evidence 501, as interpreted by the Supreme Court in Jaffee v. Redmond, 518 U.S. 1, 116 S.Ct. 1923, 135 L.Ed.2d 337 (1996), grants us authority to reconsider the question and now grant the privilege. We disagree.

Federal Rule of Evidence 501, in its current form, provides that:

Fed.R.Evid. 501 (emphasis added).

Congressional enactment of Rule 501 postdates Branzburg, but the Rule effectively left our authority to recognize common-law privileges in status quo. The Rule implemented the previously recognized authority of federal courts to consider common-law privileges "`in the light of reason and experience.'" Jaffee, 518 U.S. at 8, 116 S.Ct. 1923 (footnote omitted). "The authors of the Rule borrowed th[e] phrase from [the Supreme Court's] opinion in Wolfle v. United States, 291 U.S. 7, 12 [54 S.Ct. 279, 78 L.Ed. 617] (1934), which in turn referred to the oft-repeated observation that `the common law is not immutable but flexible, and by its own principles adapts itself to varying conditions.'" Jaffee, 518 U.S. at 8, 116 S.Ct. 1923 (footnote omitted) (quoting Funk v. United States, 290 U.S. 371, 383, 54 S.Ct. 212, 78 L.Ed. 369 (1933)).

Indeed, Rule 501 seems to be more notable for what it failed to do, than for what it did. The proposed Rules originally "defined [nine] specific nonconstitutional privileges which the federal courts [would have been compelled to] recognize (i.e. required reports, lawyer-client, psychotherapist-patient, husband-wife, communications to clergymen, political vote, trade secrets, secrets of state and other official information, and identity of informer)" and "provided that only those privileges set forth [therein] or in some other Act of Congress could be recognized by the federal courts." Fed.R.Evid. 501 advisory committee's note; see also Jaffee, 518 U.S. at 8 n. 7, 116 S.Ct. 1923 This exclusive list of enumerated privileges was ultimately rejected. Instead, Congress "left the law of privileges in its present state and further provided that privileges shall continue to be developed by the courts of the United States under" the "reason and experience" standard. Fed.R.Evid. 501 advisory committee's note.

Since enactment of Rule 501, the Supreme Court has twice noted that, while not dispositive of the question of whether a court should recognize a new privilege, the enumerated privileges proposed for inclusion in Rule 501 were "thought to be either indelibly ensconced in our common law or an imperative of federalism." United States v. Gillock, 445 U.S. 360, 368, 100 S.Ct. 1185, 63 L.Ed.2d 454 (1980) (declining to recognize under Rule 501 a legislative privilege for state legislators in a federal, criminal prosecution, in part, because it was not one of the nine enumerated privileges recommended by the Advisory Committee); see also Jaffee, 518 U.S. at 15, 116 S.Ct. 1923 (noting that, unlike in Gillock, the inclusion of the psychotherapist-patient privilege was one of the nine, and supported the Court's adoption of the privilege under Rule 501). Notably absent from the nine enumerated privileges was one for a reporter-source relationship.

In Jaffee, the Supreme Court recognized a psychotherapist-patient privilege protecting private communications that took place during counseling sessions between a police officer and a licensed clinical social worker following a fatal shooting. Applying Rule 501, the Court weighed the competing interests and concluded that the plaintiff's interest in obtaining evidence of the confidential communications in the ensuing excessive-force action was outweighed by the patient's private interest in maintaining confidence and trust with his mental health provider and the public's interest in protecting that privacy in order to "facilitat[e] the provision of appropriate treatment for individuals suffering the effects of a mental or emotional problem." Id. at 11, 116 S.Ct. 1923. As noted above, the Court also relied, in part, upon the fact that a psychotherapist-patient privilege was one of the nine, enumerated privileges considered when Rule 501 was adopted and had found near unanimous support in state laws as well.

Contrary to Risen's claim on appeal, Rule 501 and the Supreme Court's use of it to recognize a psychotherapist-patient privilege in Jaffee does not authorize us to ignore Branzburg or support our recognition of a common-law reporter-source privilege today.

Clearly, neither Rule 501 nor Jaffee overrules Branzburg or undermines its reasoning. See In re Scarce, 5 F.3d at 403 n. 3 ("We discern nothing in the text of Rule 501 ... that sanctions the creation of privileges by federal courts in contradiction of the Supreme Court's mandate" in Branzburg.).9

"In rejecting the proposed Rules and enacting Rule 501, Congress manifested an affirmative intention not to freeze the law of privilege," but "rather ... to provide the courts with the flexibility to develop rules of privilege on a case-by-case basis." Trammel v. United States, 445 U.S. 40, 47, 100 S.Ct. 906, 63 L.Ed.2d 186 (1980) (internal quotation marks omitted); see also United States v. Weber Aircraft Corp., 465 U.S. 792, 803 n. 25, 104 S.Ct. 1488, 79 L.Ed.2d 814 (1984) ("Rule 501 was adopted precisely because Congress wished to leave privilege questions to the courts rather than attempt to codify them."); United States v. Dunford, 148 F.3d 385, 390-91 (4th Cir.1998) (same). Rule 501 thus leaves the door open for courts to adopt new common-law privileges, and modify existing ones, in appropriate cases. But nothing in Rule 501 or its legislative history authorizes federal courts to ignore existing Supreme Court precedent.

Even if we were to believe that Jaffee signals that the Supreme Court might rule differently on the existence of a commonlaw reporter's privilege today, we are not at liberty to take that critical step. See Rodriguez de Quijas v. Shearson/Am. Express, Inc., 490 U.S. 477, 484, 109 S.Ct. 1917, 104 L.Ed.2d 526 (1989) ("If a precedent of this Court has direct application in a case, yet appears to rest on reasons rejected in some other line of decisions, the Court of Appeals should follow the case which directly controls, leaving to this Court the prerogative of overruling its own decisions."). Under Risen's view of Rule 501 and Jaffee, inferior federal courts would be at liberty to reconsider commonlaw privileges that have been rejected by the Supreme Court, based upon the passage of time. Rule 501 does not sanction such authority on our part.

Here, "[t]he Supreme Court has rejected a common law privilege for reporters" and "that rejection stands unless and until the Supreme court itself overrules that part of Branzburg." Judith Miller, 438 F.3d at 1155 (Sentelle, J., concurring). Just as the Supreme Court must determine whether a First Amendment reporter's privilege should exist, see Judith Miller, 438 F.3d at 1166 (Tatel, J., concurring), "only the [Supreme Court] and not this one ... may act upon th[e] argument" that a federal common-law privilege should now be recognized under Rule 501, id. at 1155 n. 3 (Sentelle, J., concurring).

C.

Even if we were at liberty to reconsider the existence of a common-law reporter's privilege under Rule 501, we would decline to do so.

As the Supreme Court made clear in Jaffee, the federal courts' latitude for adopting evidentiary privileges under Rule 501 remains quite narrow indeed. Because they "contravene the fundamental principle that the public has a right to every man's evidence," University of Pa., 493 U.S. at 189, 110 S.Ct. 577 (internal quotation marks and alteration omitted), such privileges "are not lightly created nor expansively construed, for they are in derogation of the search for truth," Nixon, 418 U.S. at 710, 94 S.Ct. 3090. "When considering whether to recognize a privilege, a court must begin with `the primary assumption that there is a general duty to give what testimony one is capable of giving, and that any exemptions which may exist are distinctly exceptional, being so many derogations from a positive general rule.'" Virmani v. Novant Health Inc., 259 F.3d 284, 287 (4th Cir.2001) (quoting Jaffee, 518 U.S. at 9, 116 S.Ct. 1923). New or expanded privileges "may be recognized `only to the very limited extent that permitting a refusal to testify or excluding relevant evidence has a public good transcending the normally predominant principle of utilizing all rational means for ascertaining truth.'" Dunford, 148 F.3d at 391 (quoting Trammel, 445 U.S. at 50, 100 S.Ct. 906).

Risen contends that the public and private recognizing a reporter's privilege "are surely as significant public interest at stake in patient and psychotherapist communication." Risen's Brief at 50. But we see several critical distinctions.

1.

First, unlike in the case of the spousal, attorney-client, and psychotherapist-patient privileges that have been recognized, the reporter-source privilege does not share the same relational privacy interests or ultimate goal. The recognized privileges promote the public's interest in full and frank communications between persons in special relationships by protecting the confidentiality of their private communications. Jaffee, 518 U.S. at 10, 116 S.Ct. 1923. A reporter's privilege might also promote free and full discussion between a reporter and his source, but Risen does not seek to protect from public disclosure the "confidential communications" made to him. Id. Risen published information conveyed to him by his source or sources. His primary goal is to protect the identity of the person or persons who communicated with him because their communications violated federal, criminal laws. See e.g., 1 McCormick on Evidence § 72 n.7 (Kenneth S. Broun ed., 7th ed.2013) (requiring for all privileges that "[t]he communications must originate in a confidence that they will not be disclosed" (internal quotation marks omitted)). In sum, beyond the shared complaint that communications might be chilled in the absence of a testimonial privilege, Risen's proffered rationale for protecting his sources shares little in common with the privileges historically recognized in the common law and developed under Rule 501.10

We are also mindful that the Court in Branzburg considered and was unpersuaded by a virtually identical argument that a reporter's privilege was necessary to prevent a chilling effect on newsgathering.

Id. at 698-99, 92 S.Ct. 2646; see also id. at 693, 92 S.Ct. 2646 ("[T]he evidence fails to demonstrate that there would be a significant constriction of the flow of news to the public if this Court reaffirms the prior common-law and constitutional rule regarding the testimonial obligations of newsmen.").

Branzburg also weighed the public interest in newsgathering against the public's interest in enforcing its criminal laws:

Id. at 695-97, 92 S.Ct. 2646; see also id. at 695, 92 S.Ct. 2646 ("Accepting the fact, however, that an undetermined number of informants not themselves implicated in crime will nevertheless, for whatever reason, refuse to talk to newsmen if they fear identification by a reporter in an official investigation, we cannot accept the argument that the public interest in possible future news about crime from undisclosed, unverified sources must take precedence over the public interest in pursuing and prosecuting those crimes reported to the press by informants and in thus deterring the commission of such crimes in the future.").

We fail to see how these policy considerations would differ in a Rule 501 analysis. Unlike the individual privacy interests in confidential communications shared by those protected by a common-law privilege, "[t]he preference for anonymity of those confidential informants involved in actual criminal conduct ..., while understandable, is hardly deserving of constitutional protection." Id. at 691, 92 S.Ct. 2646. The preference is equally undeserving of protection under the common law. Indeed, even those common-law privileges that do protect confidential communications between persons in special relationships have yielded where the communication furthers or shields ongoing criminal activity. See United States v. Zolin, 491 U.S. 554, 562-63, 109 S.Ct. 2619, 105 L.Ed.2d 469 (1989) ("The attorney-client privilege must necessarily protect the confidences of wrongdoers, but the reason for that protection — the centrality of open client and attorney communication to the proper functioning of our adversary system of justice — ceases to operate at a certain point, namely, where the desired advice refers not to prior wrongdoing, but to future wrongdoing") (internal quotation marks omitted); Clark v. United States, 289 U.S. 1, 15, 53 S.Ct. 465, 77 L.Ed. 993 (1933) ("A client who consults an attorney for advice that will serve him in the commission of a fraud will have no help from the law. He must let the truth be told."); Dunford, 148 F.3d at 391 (declining to decide whether parent-minor child testimonial privilege exists in criminal proceedings because, "even if such a privilege were to be recognized, it would have to be narrowly defined and would have obvious limits, ... such as where ... ongoing criminal activity would be shielded by assertion of the privilege").

Just as the First Amendment and the common-law attorney-client privilege do not "confer[] a license to violate valid criminal laws," Branzburg, 408 U.S. at 691, 92 S.Ct. 2646, the common law would not extend so far as to protect illegal communications that took place between Risen and his source or sources in violation of the Espionage Act.

2.

Risen's reliance upon state statutes and decisions that have adopted a reporter's shield also fails to persuade us that we can or should create a federal common-law privilege.

At the time of Branzburg, "[a] number of States ha[d] provided newsmen a statutory privilege of varying breadth." Id. at 689, 92 S.Ct. 2646. And, as Risen argues, nearly all of the remaining states have since "recognized a reporter's privilege in one context or another." Risen's Brief at 55. Generally speaking, such "policy decisions of the States bear on the question whether federal courts should recognize a new privilege or amend the coverage of an existing one." Jaffee, 518 U.S. at 12-13, 116 S.Ct. 1923. However, there is still no "uniform judgment of the States" on the issue of a reporter's privilege or shield, nor was the privilege "among the nine specific privileges recommended by the Advisory Committee in its proposed privilege rules." Id. at 14, 116 S.Ct. 1923. If anything, the varying actions of the states in this area only reinforces Branzburg's observation that judicially created privileges in this area "would present practical and conceptual difficulties of a high order," Branzburg, 408 U.S. at 704, 92 S.Ct. 2646, that are best dealt with instead by legislatures of the state and federal governments. As the Court noted in Branzburg:

Id. at 706, 92 S.Ct. 2646; cf. Judith Miller, 438 F.3d at 1161 (Henderson, J., concurring) (noting that courts "should proceed as cautiously as possible when erecting barriers between us and the truth, recognizing that the Legislature remains the more appropriate institution to reconcile the competing interests — prosecuting criminal acts versus constructing the flow of information to the public — that inform any reporter's privilege to withhold relevant information from a bona fide grand jury" (citation and internal quotation marks omitted)).

The Branzburg Court's observations regarding the practical difficulties of defining and managing a reporter's privilege, and its "unwilling[ness] to embark the judiciary on a long and difficult journey to such an uncertain destination," Branzburg, 408 U.S. at 703, 92 S.Ct. 2646, are well-taken, and we see nothing in "reason [or] experience" that would lead us to a contrary view today, Fed. Rule Evid. 501. Since Branzburg, additional state legislatures have exercised their "free[dom], within First Amendment limits, to fashion their own standards in light of the conditions and problems with respect to the relations between law enforcement officials and press in their own areas." Branzburg, 408 U.S. at 706, 92 S.Ct. 2646. Despite continued efforts, however, Congress has still not provided a reporter's shield by federal statute. See id. at 689 & n. 28, 92 S.Ct. 2686 (noting the earlier federal legislative attempts to provide a privilege).

We decline the invitation to step in now and create a testimonial privilege under common law that the Supreme Court has said does not exist and that Congress has considered and failed to provide legislatively. If Risen is to be protected from being compelled to testify and give what evidence of crime he possesses, in contravention of every citizen's duty to do so, we believe that decision should rest with the Supreme Court, which can revisit Branzburg and the policy arguments it rejected, or with Congress, which can more effectively and comprehensively weigh the policy arguments for and against adopting a privilege and define its scope.

IV. The LaRouche Test

For the foregoing reasons, we hold that there is no First Amendment or federal common-law privilege that protects Risen from having to respond to the government's subpoena and give what evidence he has of the criminal conduct at issue. We note, however, that even if we were to recognize a qualified reporter's privilege and apply the three-part LaRouche test to the inquiry, as the district court did, we would still reverse.

In LaRouche, we recognized a reporter's privilege in civil cases that can be overcome if (1) the information is relevant, (2) the information cannot be obtained by alternative means, and (3) there is a compelling interest in the information. LaRouche, 780 F.2d at 1139. Here, the government has met all three prongs.

A.

There is no dispute that the information sought from Risen is relevant. Moreover, it "can[not] be obtained by alternative means." Id. at 1139. The circumstantial evidence that the government has been able to glean from incomplete and inconclusive documents, and from the hearsay statements of witnesses with no personal or first-hand knowledge of the critical aspects of the charged crimes, does not serve as a fair or reasonable substitute.

1.

The district court held that the government had failed to establish the second factor of the LaRouche test because it has successfully obtained substantial circumstantial evidence that Sterling is the source of the illegally-disclosed information. Fundamentally, the holding appears to be grounded in the premise that circumstantial evidence of guilt should serve as an adequate substitute for a direct, first-hand account of the crime because "`circumstantial evidence is no less probative than direct evidence.'" Sterling, 818 F.Supp.2d at 956 (quoting Stamper v. Muncie, 944 F.2d 170, 174 (4th Cir.1991)). Because the district court believed that the government has uncovered substantial circumstantial evidence that Sterling is guilty, the court's ruling deprives the jury of the best and only direct evidence that supports the prosecution of this crime.

It is true, of course, that a defendant cannot ordinarily overturn a conviction based solely upon the claim that the jury had only circumstantial evidence to consider. See United States v. Bonner, 648 F.3d 209, 213 (4th Cir.2011); Stamper, 944 F.2d at 174. But this does not mean that circumstantial evidence of a fact presented to a jury will always be as convincing as direct evidence of it, particularly where the identity of the perpetrator is contested. See Bonner, 648 F.3d at 214 (reversing conviction because "[w]hile it is possible to convict a defendant solely on circumstantial evidence, in cases where the identity of the perpetrator is in dispute, usually there is some specific `identity' evidence or uncontroverted physical evidence that links the defendant to the scene of the crime"). Nor is it likely that a jury, charged with finding guilt beyond a reasonable doubt, would equate circumstantial evidence of the crucial facts with the direct testimony of the only witness with first-hand knowledge of them. The nature and strength of the evidence is very different. See 1 McCormick on Evidence § 185 (Kenneth S. Broun ed., 7th ed.2013) ("Direct evidence is evidence which, if believed, resolves a matter in issue. Circumstantial evidence also may be testimonial, but even if the circumstances depicted are accepted as true, additional reasoning is required to reach the desired conclusion." (footnote omitted)).

As the government correctly points out, "no circumstantial evidence, or combination thereof, is as probative as Risen's testimony or as certain to foreclose the possibility of reasonable doubt." Government's Brief at 14. See, e.g., New York Times Co. v. Gonzales, 459 F.3d 160, 170 (2d Cir.2006) ("[A]s the recipients of the disclosures, [the reporters] are the only witnesses — other than the source(s) — available to identify the conversations in question and to describe the circumstances of the leaks.... There is simply no substitute for the evidence they have."); Judith Miller, 438 F.3d at 1181 (Tatel, J., concurring) (noting that while "special counsel appears already to have at least circumstantial grounds for a perjury charge, if nothing else[,] [the reporter's] testimony... could settle the matter"). Risen is the only eyewitness to the crime. He is inextricably involved in it. Without him, the alleged crime would not have occurred, since he was the recipient of illegally-disclosed, classified information. And it was through the publication of his book, State of War, that the classified information made its way into the public domain. He is the only witness who can specify the classified information that he received, and the source or sources from whom he received it.

In any event, the LaRouche test does not ask whether there is other evidence, circumstantial or direct, that the government might rely upon as a substitute to prove guilt; it asks "whether the information [sought from the reporter] can be obtained by alternative means." LaRouche, 780 F.2d at 1139 (emphasis added). Clearly, it cannot be. There are no other witnesses who can offer this testimony, nor is it found in any other form of evidence. Cf. Gonzales, 459 F.3d at 172 n. 5 (noting that such circumstances do not fall within "the paradigmatic case where a newsperson is one of many witnesses to an event and the actions and state of mind of the newsperson are not in issue"). Other than Sterling himself, Risen is the only witness who can identify Sterling as a source (or not) of the illegal leak.

2.

Even if circumstantial evidence could serve as a reasonable alternative to direct evidence, the circumstantial evidence in this case does not possess the strength the district court ascribes to it — particularly when one remembers the prosecution's high burden of proof.

Sterling was not the only CIA agent involved in Classified Program No. 1. Moreover, Sterling met with staff members of the SSCI to voice complaints about the program not more than a month before the government learned that Risen had the classified information, and Sterling claims to be in possession of evidence that an SSCI employee was implicated in a previous unauthorized disclosure of classified information that made its way to Risen.11

During these proceedings, Sterling has often represented that he intends to point his finger at these third parties as the source of the leak.12 The district court's ruling, however, would require the government to compel the testimony of every other possible source, sources who could do little more than assert their own privilege or offer a simple denial of guilt, while allowing Risen, the only person who can identify the perpetrator or perpetrators, to protect his sources from the criminal consequences of their behavior. By depriving the jury of the only direct testimony that can link Sterling to the charged crimes and allowing Sterling to present argument that several others could have been the primary source or sources, the district court would allow seeds of doubt to be placed with the jurors while denying the government a fair opportunity to dispel those doubts. As the government notes, the ruling would open the door for Sterling to mislead the jury and distort the truth-seeking function of the trial.

The telephone records and e-mail messages, and the hearsay statements by witnesses who were in contact with Sterling, which were relied upon by the district court to uphold a reporter's privilege, also fail to serve as reasonable alternatives to Risen's first-hand testimony.

Telephone records, e-mail messages, and the like indicate that Risen and Sterling were communicating with one another. However, it appears that none of the records contain classified information, and the contents of the conversations and communications are otherwise largely unknown. This category of proof is an obviously poor substitute for Risen's direct testimony. See e.g., Judith Miller, 438 F.3d at 1175 (Tatel, J., concurring) ("Insofar as the confidential exchange of information leaves neither paper trail nor smoking gun, the great majority of leaks will likely be unprovable without evidence from either leaker or leakee. Of course, in some cases, circumstantial evidence such as telephone records may point towards the source, but for the party with the burden of proof, particularly the government in a criminal case, such evidence will often be inadequate.").

The proffered hearsay testimony from the former CIA agent and Sterling's then-girlfriend also pales in comparison to Risen's first-hand testimony. Even assuming that the hearsay testimony would be admissible, which we need not decide today, it is not a reasonable equivalent to Risen's testimony.

It is represented to us that Sterling's girlfriend will testify that Sterling told her at some unspecified point that he had a meeting with "Jim" and, during a much later trip to a bookstore, told her that Chapter 9 of State of War was about his work in the CIA. However, it is undisputed that Risen and Sterling had been in contact about other matters, such as his firing by the CIA, and the proffered testimony tells us nothing about the substance of any leak of classified information. Moreover, the persons to whom Sterling points as alternative sources of the leak would have been privy to the same information at about the same time, and Risen has not disclosed whether there is more than one primary source of classified information.

It is also represented to us that a former CIA agent will testify that Risen told him that Sterling was his source. This characterization of the hearsay testimony, however, is much more generous than warranted. The proffered testimony does not establish whether Sterling was the primary or only source of classified information that made its way into State of War, nor does it address the breadth of information found in the book. It too is a poor substitute for Risen's testimony.

Additionally, Sterling has indicated that he will offer another defense to this hearsay testimony, either through cross-examination of Risen or through other expert testimony. Specifically, Sterling has sought to present expert testimony that "[j]ournalists commonly use techniques to disguise their sources," and that "statements made to third parties, including prospective sources, purporting to identify other sources from whom the author has obtained information are inherently suspect and should not be accepted at face value." J.A. 863. Whether or not Sterling can persuade the jury on this point, the argument is not a lost one. Unlike Risen, the former CIA agent simply cannot testify that he knows Sterling to be Risen's source, because he does not know that to be true. He cannot refute the possibility that Risen might have falsely pointed the finger at Sterling to protect his real source from scrutiny, or to entice the former CIA agent to provide similar or confirming information. Only Risen can answer these questions.

Accordingly, even if we were to recognize a reporter's privilege that could deprive a jury of the only direct, firsthand evidence of guilt or innocence, Risen's statement to the former CIA agent would be in violation of the confidentiality agreement that he relies upon to create the privilege. Notwithstanding any evidence of a standard journalistic practice of deception in investigative techniques, Risen has waived any privilege by violating the promise of confidentiality and disclosing the information to a third party. To rule otherwise would not only allow journalists to protect their confidential sources in criminal proceedings, but would also permit journalists to promise confidentiality to those engaged in ongoing criminal conduct, while at the same time disclosing their identities to anyone except law enforcement, grand juries investigating the crimes, and juries called upon to determine innocence or guilt.

Clearly, Risen's direct, first-hand account of the criminal conduct indicted by the grand jury cannot be obtained by alternative means, as Risen is without dispute the only witness who can offer this critical testimony. The information sought from Risen is not reasonably or fairly equaled by the inconclusive records of phone calls and emails, or the hearsay testimony of the other witnesses.

B.

The government has also demonstrated a compelling interest in presenting Risen's testimony to the jury.

"It is `obvious and unarguable' that no governmental interest is more compelling than the security of the Nation." Haig v. Agee, 453 U.S. 280, 307, 101 S.Ct. 2766, 69 L.Ed.2d 640 (1981). This interest extends to "protecting both the secrecy of information to our national security and the appearance of confidentiality so essential to the effective operation of our foreign intelligence service." United States v. Abu Ali, 528 F.3d 210, 247 (4th Cir.2008) (quoting CIA v. Sims, 471 U.S. 159, 175, 105 S.Ct. 1881, 85 L.Ed.2d 173 (1985)). Clearly, the government also has a compelling interest in obtaining direct evidence that Sterling compromised these critical national-security interests by disclosing classified information in violation of validly-enacted criminal laws, and in presenting this evidence to the jury charged with determining his guilt or innocence. See LaRouche, 780 F.2d at 1139.

Risen's testimony is the best evidence to prove Sterling's guilt beyond a reasonable doubt to a jury charged with the search for the truth. He is the only one who can identify Sterling as the perpetrator of the charged offenses, and he is the only one who can effectively address Sterling's expected efforts to point the finger at others. If Risen identifies Sterling as his source, he will have provided unequaled evidence of guilt on this point, yet not deprived Sterling of his defense that the information in Risen's book was not, in fact, national defense information at all. And should Risen identify different or additional sources of national defense information, which could exculpate Sterling, the government maintains an equally compelling interest in obtaining the only available inculpatory evidence against all who jeopardized the security of the United States and at least one of its covert assets.

To date, Sterling has not sought to compel Risen to testify regarding the identity of his source, and he professes to "take[] no position" as to whether Risen has properly invoked a reporter's privilege. Defendant-Appellee's Brief at 5. Sterling has, however, seized upon the government's unsuccessful attempts to compel Risen's testimony to repeatedly point out "how little evidence the Government really has [against him] in this case." J.A. 892. Sterling even goes so far as to point out the absence of direct evidence of his guilt, arguing that:

J.A. 892 (emphasis added); see also J.A. 893 (asserting that "[t]he Government now admits that its case is entirely speculative even as to venue. It admits that it has `no direct evidence, other than Risen's testimony, that establishes where the substantive disclosures of classified information occurred'.... In short, the Government is so fixated on compelling Mr. Risen's testimony — or perhaps jailing him — that it is willing to concede that its case is weak and that it needs Mr. Risen ... to come to the rescue." (emphasis added) (citation omitted)). Hardly a better argument could be made as to why the evidence sought from Risen is unavailable from alternative sources and why the government has demonstrated a compelling need for it.

V.

For the foregoing reasons, we reverse the district court's order granting Risen's motion to quash his trial subpoena and denying the government's motion in limine to admit his testimony, which would allow Risen to protect the identity of the source of the classified, national security information that the grand jury found probable cause to believe was illegally leaked to Risen.

GREGORY, Circuit Judge, writing for the court on Issues II and III:

VI. District Court's Suppression Order

The Government challenges the district court's order excluding two of its witnesses as a sanction for violating a discovery order. The discovery order at issue, entered by the district court with the parties' consent, provided that all Giglio13 material had to be turned over to the defense no later than five calendar days prior to the start of trial. The trial was initially slated to begin on September 12, 2011. However, in early July 2011, Sterling and the Government requested a continuance based on the complexity of the pretrial discovery issues. See 18 U.S.C. § 3161(h)(7)(B)(ii). The district court agreed, rescheduling the trial to begin on October 17, 2011. Thus, the new discovery deadline was October 12, 2011, five days prior to the trial date.

During the months leading up to trial, the Government produced nearly 20,000 pages of discovery material, along with various items in electronic format. As the trial date approached, the Government continued to search the CIA's files, and at the eleventh hour it discovered impeachment materials in the personnel files of six of its witnesses. Due to the risk of classified information being contained in the CIA's files, all of this discovery material had to be presented to the CIA for a line-by-line classification review before the information could be turned over to the defense.

The CIA completed its line-by-line review of the disputed material and provided it to the Government on the evening of October 12, 2011. The Government turned the information over to the defense on the morning of October 13, 2011 — the day after the discovery period expired.

At a pre-trial hearing on October 13, the defense did not object to the late disclosure. At a hearing on October 14, the Friday before the Monday on which the trial was to commence, the district court noted that the Government had not timely complied with the discovery schedule. The Government apologized for the delay and thanked the defense for not objecting — at which point, defense counsel lodged an objection. In addressing a possible remedy, the defense stated the court could grant a brief continuance, but observed that this option would not be particularly palatable to the court. The defense then stated that the court could sanction the Government by striking a witness. At that point the district court decided to strike two witnesses, to "even up the playing field." J.C.A. 577.

The Government objected to the court's order arguing that the delay in production was not in bad faith. As an alternative sanction for the delay, the Government suggested that the court grant a continuance and offered to assist the defense in locating three people whose unfavorable ratings of a CIA colleague comprised a portion of the Giglio material as to that colleague. The court asked the defense about its schedule, seeking to determine whether counsel's other obligations would accommodate a brief continuance. However, the court had already struck two crucial prosecution witnesses, and the defense preferred this sanction to a continuance. Thus, although the court subsequently found the Government did not act in bad faith, it maintained its decision to strike the two witnesses.

We have jurisdiction over the Government's appeal of this order pursuant to 18 U.S.C. § 3731.

The Due Process Clause requires the prosecution to disclose upon request evidence that is favorable to the defense and material to guilt or punishment. United States v. Higgs, 663 F.3d 726, 734-35 (4th Cir.2011). Evidence is favorable if it is exculpatory, Brady v. Maryland, 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963), or if it may be used for impeachment, Giglio v. United States, 405 U.S. 150, 92 S.Ct. 763, 31 L.Ed.2d 104 (1972). The government breaches its duty if it fails to produce evidence that it is obligated to turn over to the defense, or if it fails to timely comply with a discovery order in turning over required evidence. A failure to disclose violates due process only if the evidence in question (1) is favorable to the defendant because it is either exculpatory or impeaching; (2) was suppressed by the government; and (3) is material in that its suppression prejudiced the defendant. Strickler v. Greene, 527 U.S. 263, 281-82, 119 S.Ct. 1936, 144 L.Ed.2d 286 (1999); Vinson v. True, 436 F.3d 412, 420 (4th Cir.2006). Undisclosed evidence is material when its cumulative effect is such that "there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different." Kyles v. Whitley, 514 U.S. 419, 433-34, 115 S.Ct. 1555, 131 L.Ed.2d 490 (1995) (internal quotation marks and citation omitted). A reasonable probability is one sufficient to undermine confidence in the outcome. Id. at 434, 115 S.Ct. 1555.

When the government's contumacious conduct involves a delay in producing discovery, rather than a failure to turn over required materials, the relevant inquiry is "whether the defendant's counsel was prevented by the delay from using the disclosed material effectively in preparing and presenting the defendant's case." United States v. Ingraldi, 793 F.2d 408, 411-12 (1st Cir.1986). "As long as evidence is disclosed before it is too late for the defendant to make effective use of it, there is no due process violation." United States v. Russell, 971 F.2d 1098, 1112 (4th Cir.1992) (discussing allegation of delay in producing exculpatory evidence in violation of Brady).

The district court is permitted, but not required, to impose sanctions upon the government's failure to timely comply with a discovery order. Fed.R.Crim.P. 16(d)(2); see United States v. Lopez, 271 F.3d 472, 483 (3d Cir.2001). If the court decides to impose a sanction, it may:

Fed.R.Crim.P. 16(d)(2). "A continuance is the preferred sanction." United States v. Hammoud, 381 F.3d 316, 336 (4th Cir. 2004) (en banc) (citing United States v. Gonzales, 164 F.3d 1285, 1292 (10th Cir. 1999)), vacated on other grounds, 543 U.S. 1097, 125 S.Ct. 1051, 160 L.Ed.2d 997 (2005).

When the government fails to timely provide Giglio material, the district court's determination of whether to impose a sanction, and what sanction to impose, is reviewed for abuse of discretion. Hammoud, 381 F.3d at 336. "A district court abuses its discretion only where it `has acted arbitrarily or irrationally[,] has failed to consider judicially recognized factors constraining its exercise of discretion, or when it has relied on erroneous factual or legal premises.'" L.J. v. Wilbon, 633 F.3d 297, 304 (4th Cir.2011) (quoting United States v. Hedgepeth, 418 F.3d 411, 419 (4th Cir.2005)); see James v. Jacobson, 6 F.3d 233, 239 (4th Cir.1993). Likewise, a district court abuses its discretion when it commits an error of law. United States v. Delfino, 510 F.3d 468, 470 (4th Cir.2007); see United States v. Wilson, 624 F.3d 640, 661 n. 24 (4th Cir.2010) ("It is an abuse of discretion for the district court to commit a legal error-such as improperly determining whether there was a Brady violation-and that underlying legal determination is reviewed de novo.").

In fashioning a remedy for a Giglio violation, the district court must consider several factors: the reason for the government's delay, and whether the government acted intentionally or in bad faith; the degree of prejudice, if any, suffered by the defendant; and whether any less severe sanction will remedy the prejudice to the defendant and deter future wrongdoing by the government. Hammoud, 381 F.3d at 336 (citing United States v. Hastings, 126 F.3d 310, 317 (4th Cir.1997)); Gonzales, 164 F.3d at 1292. "When a court sanctions the government in a criminal case for its failure to obey court orders, it must use the least severe sanction which will adequately punish the government and secure future compliance." Hastings, 126 F.3d at 317; see also United States v. Ivy, 83 F.3d 1266, 1280 (10th Cir.1996). Indeed, it "`would be a rare case where, absent bad faith, a district court should exclude evidence.'" Hammoud, 381 F.3d at 336 (quoting United States v. Golyansky, 291 F.3d 1245, 1249 (10th Cir.2002)).

Neither the district court nor Sterling suggests that the Government acted in bad faith, and our review of the record dispels any such notion. It is clear that the sheer volume of materials, along with the inherent delays involved in classification review, was the genesis of the Government's error. The other contributing factor, of course, was the Government's failure to recognize the necessity of reviewing the personnel files of likely witnesses at an earlier stage of the discovery process. We cannot, of course, condone the Government's oversight; as Sterling points out, the Government had many months to examine the relevant records, and the evidence at issue here would have been an obvious source for potential Giglio material. However, other factors guide our decision.

Sterling suggests that because the material was not submitted by the discovery deadline, he "could not possibly have fully investigated and developed the belatedly-disclosed evidence prior to the start of trial, three to four days later."14 (Appellee Sterling's br. at 6). Although we do not take lightly the impact of the Government's delay on Sterling's ability to prepare, it is difficult to imagine that Sterling could have fully prepared with regard to the Giglio material if he received it on the last day of the discovery period, but "could not possibly" have prepared having received the material the next day, four days prior to trial. Sterling alleges that, if he had received the Giglio material at an earlier time, he could have thoroughly investigated the information and the witnesses to which that information pertained. As to the error, the prejudice from the brief delay in disclosure could plainly have been alleviated with a continuance.

Both Sterling and the district court suggest the Government should have produced the Giglio material earlier in the discovery process. Although efforts at earlier review and disclosure of the relevant personnel files might have ameliorated the error, and would certainly have eased the defense's undoubtedly hectic pretrial preparations, the Government was not obligated to accelerate its production to complete discovery in advance of the deadline — a deadline to which the parties and the district court agreed. We can only find error in the Government's one-day delay in production — not in its perhaps ill-advised document review strategy, nor in its failure to produce the materials at an earlier stage of the discovery process.

We are convinced, moreover, that the Government has been adequately chastened, and that it will proceed more judiciously in the future. Further, as the Government is surely aware, any similar future transgression will not be forgiven as easily.

In sum, although the district court did not abuse its discretion by imposing a sanction, the sanction that it chose to impose was simply too severe a response to conduct that was not undertaken in bad faith, that can be remedied with a continuance, and that is unlikely to be repeated. As we said in Hammoud, a continuance is the preferred sanction for a delay in production of Giglio material. Nothing in the record suggests that Sterling would not have been able to make use of the impeachment evidence if given a continuance. See Golyansky, 291 F.3d at 1249-50. We discern no justification for the more severe sanction of striking witnesses. Accordingly, we reverse the district court's order striking the two witnesses.

VII. CIPA Ruling

Prior to trial the Government moved for a protective order, pursuant to the Classified Information Procedures Act ("CIPA"), 18 U.S.C. app. 3 § 6, prohibiting the disclosure of classified and sensitive information. The list of protected information included:

J.C.A. 400. The Government sought to protect the identities of some of its witnesses — as relevant here, current or former CIA operatives — through use of a screen or light disguises (wigs, false beards, half glasses), use of a non-public entrance to the courtroom, and, of critical importance to this appeal, by allowing the witnesses to use last initials rather than their full names (for example, "Mr. D." instead of John Doe).

The district court initially granted in part and denied in part the Government's request for security measures when the CIA operatives testified. The court agreed that the CIA operatives would not have to reveal their names, and allowed that those witnesses could use a non-public entrance to the courtroom. The court stated that no sketch artists would be permitted in the courtroom, but denied the Government's request for the witnesses to testify from behind a screen.15 The Government moved for reconsideration of this ruling, stating that the witnesses needed more protection than was permitted by the district court's prior ruling. Specifically, the Government argued for the use of a portable screen between the witnesses and the public,16 or permitting the witnesses to testify wearing light disguises. Sterling opposed the Government's motion for reconsideration, stating that the Government had offered no new information justifying reconsideration of the court's prior ruling. Sterling also contended that the security measures proposed by the Government would infringe upon Sterling's right to a public trial and to confront the witnesses against him. He contended that the use of screens or disguises was unduly suggestive of the existence of national defense information, problematic because one of his planned defenses was that the information in Risen's book was not, in fact, national defense information. Although Sterling expressed frustration with the security measures previously imposed by the court, he did not ask the court to alter its ruling permitting the CIA operatives to use partial names or pseudonyms.

At the October 14 hearing, the court reversed course as to both the screen and the witnesses' names. The court agreed to permit a screen between the trial participants and the public seating section of the courtroom.17 And although the witnesses could use pseudonyms while testifying, the Government was ordered to provide to defense counsel, Sterling, and the jury a key with the witnesses' true names.18 The Government appealed the portion of the order requiring it to provide a key with the witnesses' true names to Sterling and the jury.

Sterling contends we do not have jurisdiction to review the order requiring disclosure of the witnesses' true identities to Sterling and the jury. The Government raises two bases for its argument that the disclosure order is immediately appealable: 18 U.S.C. § 3731, and CIPA section 7, 18 U.S.C. app. 3, § 7. Section 3731, as recounted at Section II.A, does not confer jurisdiction for an immediate appeal as to this issue because the order is not one suppressing or excluding evidence. Thus, we turn to CIPA.

A.

CIPA provides a framework for determining how to proceed with discovery and admissibility of classified information in criminal cases. See United States v. Moussaoui, 591 F.3d 263, 281-82 (4th Cir. 2010). It was designed to balance the defendant's interest in a fair trial and the government's interest in protecting national security information. United States v. Passaro, 577 F.3d 207, 219 (4th Cir.2009). When classified information may come into play at trial, the government may move for a hearing in the district court "to make all determinations concerning the use, relevance, or admissibility of classified information that would otherwise be made during the trial or pretrial proceedings." 18 U.S.C. app. 3, § 6(a). The district court's order was, we conclude, an order concerning the use of classified information encompassed by CIPA section 6.

It is true, as Sterling contends, that this is not a run-of-the-mill CIPA appeal. CIPA generally comes into play when the defendant seeks to obtain, or plans to disclose, national security information, and the government opposes disclosure. United States v. Moussaoui, 333 F.3d 509, 514 (4th Cir.2003). In Moussaoui, we held that an order permitting a deposition of an enemy combatant witness was not immediately appealable under CIPA. We reasoned that CIPA was concerned with disclosure of classified information at trial, rather than the defendant's pretrial discovery of classified information. Thus, we concluded, CIPA was only applicable by analogy, and in that instance CIPA § 7 did not authorize an interlocutory appeal.

Following Moussaoui, we considered a case in which the government introduced classified information at trial, and relied upon CIPA in protecting that information from disclosure. United States v. Abu Ali, 528 F.3d 210, 255 (4th Cir.2008). There, the government used classified information to which neither Abu Ali nor his counsel was privy. We held that:

Id. The procedural posture of this case is, of course, different from Abu Ali; Abu Ali was an appeal following conviction, not an interlocutory appeal. Nevertheless, it is illustrative; evidence sought to be admitted at trial by the government, like that proffered by the defense, is subject to the protections afforded by CIPA.

The order at issue authorizes disclosure of classified information at trial, unlike the order in Moussaoui, which involved the defendant's pretrial discovery request. Cf. United States v. Moussaoui, 336 F.3d 279, 280 (4th Cir.2003) (Wilkins, C.J., concurring in the denial of en banc rehearing) (noting that CIPA § 6 applies to the use of classified information at trial or in pretrial proceedings, and not to pretrial discovery of classified information). Given our recognition in Abu Ali that CIPA applies to evidence proffered by the government for use at trial, we have jurisdiction over this interlocutory appeal pursuant to Section 7 of CIPA, which provides:

18 U.S.C. app. 3, § 7(a). Having determined that we have jurisdiction to review the district court's order, we turn to the merits, reviewing for abuse of discretion. Abu Ali, 528 F.3d at 253-54 (applying abuse of discretion standard, but striking a balance between the defendant's Confrontation Clause rights and the government's need to protect classified information).

B.

There can be no doubt that the identity of CIA operatives is sensitive information. The identity of CIA operatives is, and always has been, subject to rigorous protection. See, e.g., In re Grand Jury Subpoena, Judith Miller, 438 F.3d 1141 (D.C.Cir.2006). To disclose the identities of CIA operatives, even if not to every spectator in the courtroom, subjects the operatives to targeting by hostile foreign intelligence services and terrorist organizations, and creates a grave danger to the operatives, their families, and the operations in which they are engaged. Cf. United States v. Ramos-Cruz, 667 F.3d 487, 500 (4th Cir.2012) (recognizing that defendant's rights under the Confrontation Clause to identifying information about witnesses is not absolute; if the government shows an actual threat, the district court has discretion to determine whether effective cross-examination is possible if the witness's identity is concealed).

We find no abuse of discretion in the district court's decision to make available to Sterling and his counsel a key to the witnesses' true names. Sterling knows, or may know, some of the witnesses at issue, and depriving him of the ability to build his defense in this regard could impinge on his Confrontation Clause rights. See generally Maryland v. Craig, 497 U.S. 836, 848-49, 110 S.Ct. 3157, 111 L.Ed.2d 666 (1990). Moreover, and unlike the usual cases where witnesses have been permitted to use pseudonyms, the Government in this case has made no showing that Sterling or his counsel pose an actual threat to the safety of these witnesses. See Ramos-Cruz, 667 F.3d at 506; United States v. El-Mezain, 664 F.3d 467, 492 (5th Cir.2011). Thus, we discern no potential for harm from disclosure of their identities to Sterling and his counsel. We cannot, however, take the same approach when it comes to the jury.

Sterling contends that the security measures proposed by the Government will serve to impermissibly heighten the jury's sensitivity to the classified nature of the information Sterling is accused of disclosing, increasing the odds of his conviction. The district court understandably sought to limit to the extent possible the elements of secrecy in this case, and we, too, are mindful of the risk of tainting the jury if unduly suggestive security measures are used at trial. If a security measure is inherently prejudicial, it may be employed "only where justified by an essential state interest specific to each trial." Holbrook v. Flynn, 475 U.S. 560, 568-69, 106 S.Ct. 1340, 89 L.Ed.2d 525 (1986). However, we can discern no real benefit that would inure from providing the jury with the full, true names of the CIA operatives at issue. The court sought to limit the risk of disclosure by proposing to instruct the jurors not to write down the witnesses' true names, but nothing will prevent a juror from remembering the names-and, for that matter, the other classified information presented at trial. Unlike the information Sterling is charged with disclosing to Risen, though, the true names of the CIA operatives at issue will do nothing to enhance the jury's understanding of the facts and legal issues presented at trial. And although we are mindful that the jurors are unlikely to disseminate the names in contravention of the district court's instructions, it simply is not worth the risk to the lives of these operatives (and their families and associates) to disclose the operatives' true names to anyone who does not have a genuine need to know their identities.

Although Sterling may dispute at trial that the information at issue was classified, or that he was the person who passed to Risen the information in Chapter Nine, there is no escaping the fact that Sterling has been charged with disclosing classified information, and the jury will be well aware of that fact from the very outset of the proceedings. The district court has made clear that it will instruct the jury that Sterling's guilt cannot be inferred from the use of security measures in the courtroom. Balancing Sterling's concerns with the very real danger to the CIA operatives if their identities are disclosed, we conclude that a proper jury instruction will alleviate any potential prejudice, and that the district court abused its discretion in taking the more perilous approach of ordering that the jury be given a key with the operatives' true names. Thus, we reverse this portion of the district court's order. We affirm, however, the portion of the order permitting Sterling and his counsel to receive the key with the operatives' true names.

C.

For the foregoing reasons, we reverse the court's exclusion of two Government witnesses, and affirm in part and reverse in part the court's ruling pursuant to CIPA. We remand for further proceedings consistent with this opinion.

TRAXLER, Chief Judge, concurring in part and dissenting in part as to Issues II and III:

I concur in the majority's decision as to Issue II, which reverses the district court's order striking two of the government's witnesses as a sanction for violating the discovery order. With regard to Issue III, I concur in the reversal of the district court's order requiring disclosure of the identities of the covert CIA agents and operatives (the "CIA witnesses") to the jury. I respectfully dissent, however, from the majority's decision to affirm the district court's order requiring disclosure of this information to Sterling.

Prior to trial, the government filed a motion under Section 6 of the Classified Information Procedures Act ("CIPA"), see 18 U.S.C.App. III, requesting permission to substitute pseudonyms for the true names of the CIA witnesses. The government also asked that a screen be used to shield the witnesses from the public's view, but not the view of Sterling or the jury. The motions were accompanied by CIA and FBI declarations explaining in detail that public disclosure would jeopardize the personal safety of the witnesses, their families, and associates, and would jeopardize the effectiveness of the CIA witnesses as agents and operatives. Additionally, foreign intelligence and terrorist organizations have a significant interest in identifying CIA agents and operatives, and use information gleaned from trials to expose their activities, sources, and methods.

The district ruled that the CIA witnesses would be allowed to testify using pseudonyms and from behind a screen, but that their true identities would have to be disclosed to Sterling and the jury. The majority reverses the district court's ruling as to the jury, but affirms as to Sterling. Because disclosure of the identities of the CIA witnesses endangers the personal safety of the witnesses and others associated with them, and jeopardizes the witnesses' effectiveness as agents and operatives, and there has been no demonstration that Sterling cannot effectively cross-examine the witnesses without this information, I would reverse the disclosure ruling as to both the jury and Sterling.

A.

As a general rule, "the Confrontation Clause guarantees a defendant the right to question an adverse witness about identifying information, including his full name and address." United States v. Ramos-Cruz, 667 F.3d 487, 500 (4th Cir.2012) (citing Smith v. Illinois, 390 U.S. 129, 131, 88 S.Ct. 748, 19 L.Ed.2d 956 (1968)). However, "th[e] right is not absolute," and "a trial court may limit cross-examination if the information sought could endanger the witness." Id. (internal quotation marks omitted). "When the government seeks to withhold a witness's true name, address, or place of employment, it bears the burden of demonstrating that the threat to the witness is actual and not a result of conjecture." Id. (internal quotation marks and alteration omitted). Once the government meets this burden, the court must "review relevant information and determine whether disclosure of the witness's identifying information is necessary to allow effective cross-examination." Id.

B.

There is "no governmental interest ... more compelling than the security of the Nation," and "[m]easures to protect the secrecy of our Government's foreign intelligence operations plainly serve these interests." Haig v. Agee, 453 U.S. 280, 307, 101 S.Ct. 2766, 69 L.Ed.2d 640 (1981); see also Snepp v. United States, 444 U.S. 507, 509 n. 3, 100 S.Ct. 763, 62 L.Ed.2d 704 (1980). "[T]he Government must tender as absolute an assurance of confidentiality as it possibly can" to intelligence officers and sources, C.I.A. v. Sims, 471 U.S. 159, 175, 105 S.Ct. 1881, 85 L.Ed.2d 173 (1985), and courts should exercise particular caution before "order[ing] [their] identit[ies] revealed," id. at 176, 105 S.Ct. 1881. Protecting the classified identities of covert CIA agents and operatives is of particular concern because disclosure places not only our national security at risk, but also the personal safety of those who have committed their lives to the service of our country. Indeed, Congress has criminalized such disclosure, see 50 U.S.C. § 421, given the "behavior's `intolerable' consequences: `[t]he loss of vital human intelligence which our policymakers need, the great cost to the American taxpayer of replacing intelligence resources lost due to such disclosures, and the greatly increased risk of harm which continuing disclosures force intelligence officers and sources to endure.'" In re Grand Jury Subpoena, Judith Miller, 438 F.3d 1141, 1179 (D.C.Cir. 2006) (Tatel, J., concurring) (quoting S.Rep. No. 97-201, at 10-11 (1981)); see also 50 U.S.C. § 403g (noting that "the interests of the security of the foreign intelligence activities of the United States" require that the names of CIA personnel be protected).

The actual threat to CIA witnesses has been well documented in this case, and it appears that we all agree on this point. As the majority notes: "To disclose the identities of CIA operatives, even if not to every spectator in the courtroom, subjects the operatives to targeting by hostile foreign intelligence services and terrorist organizations, and creates a grave danger to the operatives, their families, and the operations in which they are engaged." Majority op. at 75. Accordingly, we unanimously conclude that the district court abused its discretion in requiring disclosure of the identifying information to the jury.

I depart from the majority's view, however, that disclosure to Sterling is nevertheless required because there has been no showing that Sterling poses an actual threat to the safety of the witnesses. "[T]he appropriateness of using pseudonyms to protect witnesses does not depend on whether the threat to the witness comes directly from a defendant or from another source." Ramos-Cruz, 667 F.3d at 501 (internal quotation marks omitted). But, in any event, the grand jury in this case has found probable cause to believe that Sterling has already revealed classified information about a covert operation and a covert CIA asset for publication in the public domain. In my opinion, no more needs to be shown to demonstrate that disclosure of the true identities of the CIA witnesses to Sterling poses an actual and specific risk, sufficient to require serious inquiry into the necessity of the disclosure for purposes of confrontation.

Because the government seeks to protect the confidentiality of the CIA witnesses' identities to minimize the actual threat disclosure poses to them, Sterling was required to demonstrate that disclosure is necessary to conduct an effective cross-examination. See id. at 500; see also United States v. El-Mezain, 664 F.3d 467, 492, 493 (5th Cir.2011) (holding that the defendants' Confrontation Clause rights were not violated by allowing Israeli security officers to testify using pseudonyms, due to the "serious and clear need to protect the true identities of [the witnesses] because of concerns for their safety" and the defendants' adequate opportunity "to conduct effective cross-examination"); United States v. Lonetree, 35 M.J. 396, 410 (C.M.A.1992) (rejecting argument that Confrontation Clause was violated by allowing a United States intelligence agent to testify without disclosing his true name because it endangered the agent and "was not essential to a fair resolution of the cause").

I have much respect for the district court, which has dealt with difficult questions arising from the classified nature of this case. On this particular point, however, I am constrained to find an abuse of discretion. Given the dangers involved, the district court should have granted the government's motion to withhold disclosure of the witnesses' identifying information because there had been no showing that the disclosure was "necessary to allow effective cross-examination." Ramos-Cruz, 667 F.3d at 500. Instead, the district court merely ruled that the identities of the CIA witnesses should be revealed because "the defendant may know things about [a] witness," and could "turn to counsel and say: Hey, ask him about such-and-such on cross-examination." J.C.A. at 487. The majority similarly concludes only that failure to disclose the identifying information might "depriv[e] [Sterling] of the ability to build his defense" and, "in this regard could impinge on his Confrontation Clause rights." Majority op. at 516-17. In my opinion, this is too speculative a basis upon which to require disclosure of the identities of the CIA witnesses to Sterling.

Sterling has been provided with discovery on all of the witnesses by their pseudonyms, including prior statements, interview reports, cables, and other documents. Sterling therefore appears to already know the factual connection that each witness has to his case. See Ramos-Cruz, 667 F.3d at 501 (noting that "because the government disclosed to the defense details of the[] witnesses before the trial, the defendants were able to effectively cross-examine the witnesses without threatening their safety" (internal quotations marks omitted)). Because disclosure of the identities of the covert CIA witnesses endangers their safety, and Sterling has not made the required demonstration that he needs this information in order to conduct a meaningful cross-examination of the witnesses, I would reverse the district court's order requiring disclosure of the identities of the CIA witnesses to Sterling as well.

GREGORY, Circuit Judge, dissenting as to Issue I:

Today we consider the importance of a free press in ensuring the informed public debate critical to citizens' oversight of their democratically elected representatives. Undoubtedly, the revelation of some government secrets is too damaging to our country's national security to warrant protection by evidentiary privilege. Yet the trial by press of secret government actions can expose misguided policies, poor planning, and worse. More importantly, a free and vigorous press is an indispensable part of a system of democratic government. Our country's Founders established the First Amendment's guarantee of a free press as a recognition that a government unaccountable to public discourse renders that essential element of democracy — the vote — meaningless. The majority reads narrowly the law governing the protection of a reporter from revealing his sources, a decision that is, in my view, contrary to the will and wisdom of our Founders.

The district court ruled that under Branzburg v. Hayes, 408 U.S. 665, 92 S.Ct. 2646, 33 L.Ed.2d 626 (1972), and subsequent precedent from this Circuit, the Government could not compel Risen to reveal his source for chapter nine of his book, State of War. We review de novo the district court's legal determination that the reporter's privilege exists in the criminal context, and we examine the district court's application of that privilege to the instant facts under a deferential abuse-of-discretion standard.1 Church of Scientology Int'l v. Daniels, 992 F.2d 1329, 1334 (4th Cir.1993); LaRouche v. Nat'l Broad. Co., 780 F.2d 1134, 1139 (4th Cir.1986).

A.

The freedom of the press is one of our Constitution's most important and salutary contributions to human history. See U.S. Const. amend. I ("Congress shall make no law ... abridging the freedom of speech, or of the press[.]"). Reporters are "viewed `as surrogates for the public,'" United States v. Criden, 633 F.2d 346, 355 (3d Cir.1980) (quoting Richmond Newspapers, Inc. v. Virginia, 448 U.S. 555, 573, 100 S.Ct. 2814, 65 L.Ed.2d 973 (1980)), who act in the public interest by uncovering wrongdoing by business and government alike. Democracy without information about the activities of the government is hardly a democracy. The press provides "a constitutionally chosen means for keeping officials elected by the people responsible to all the people whom they were selected to serve." Mills v. Alabama, 384 U.S. 214, 219, 86 S.Ct. 1434, 16 L.Ed.2d 484 (1966). A citizen's right to vote, our most basic democratic principle, is rendered meaningless if the ruling government is not subjected to a free press's "organized, expert scrutiny of government." Justice Potter Stewart, Or of the Press, 26 Hastings L.J. 631, 634 (1975).

The protection of confidential sources is "necessary to ensure a free and vital press, without which an open and democratic society would be impossible to maintain." Ashcraft v. Conoco, Inc., 218 F.3d 282, 287 (4th Cir.2000). If reporters are compelled to divulge their confidential sources, "the free flow of newsworthy information would be restrained and the public's understanding of important issues and events would be hampered in ways inconsistent with a healthy republic." Id.; see also Zerilli v. Smith, 656 F.2d 705, 711 (D.C.Cir.1981) ("Compelling a reporter to disclose the identity of a source may significantly interfere with this news gathering ability" and threaten "a vital source of information," leaving citizens "far less able to make informed political, social, and economic choices.").

Yet if a free press is a necessary condition of a vibrant democracy, it nevertheless has its limits. "[T]he reporter's privilege ... is not absolute and will be overcome whenever society's need for the confidential information in question outweighs the intrusion on the reporter's First Amendment interests." Ashcraft, 218 F.3d at 287. And we must be mindful of the "fundamental maxim that the public ... has a right to every man's evidence." Jaffee v. Redmond, 518 U.S. 1, 9, 116 S.Ct. 1923, 135 L.Ed.2d 337 (1996) (quoting United States v. Bryan, 339 U.S. 323, 331, 70 S.Ct. 724, 94 L.Ed. 884 (1950)).

The public, of course, does not have a right to see all classified information held by our government. But public debate on American military and intelligence methods is a critical element of public oversight of our government. Protecting the reporter's privilege ensures the informed public discussion of important moral, legal, and strategic issues. Public debate helps our government act in accordance with our Constitution and our values. Given the unprecedented volume of information available in the digital age — including information considered classified — it is important for journalists to have the ability to elicit and convey to the public an informed narrative filled with detail and context. Such reporting is critical to the way our citizens obtain information about what is being done in their name by the government.

A reporter's need for keeping sources confidential is not hypothetical. The record on appeal contains affidavits proffered by Risen detailing the integral role of confidential sources in the newsgathering process. Scott Armstrong, executive director of the Information Trust and former Washington Post reporter, points to three ways in which investigative journalism uses confidential sources: "developing factual accounts and documentation unknown to the public," "tak[ing] a mix of known facts and new information and produc[ing] an interpretation previously unavailable to the public," and "publiciz[ing] information developed in government investigations that has not been known to the public and might well be suppressed." Joint App'x (J.A.) 531. "It would be rare," Armstrong asserts, "for there not to be multiple sources — including confidential sources — for news stories on highly sensitive topics." Id. In turn, "[m]any sources require such guarantees of confidentiality before any extensive exchange of information is permitted." J.A. 350. Such guarantees of confidentiality enable sources to discuss "sensitive matters such as major policy debates, personnel matters, investigations of improprieties, and financial and budget matters." Id. Even in ordinary daily reporting, confidential sources are critical. "[O]fficial government pronouncements must be verified before they are published," and this is frequently done through discussion with officials not authorized to speak on the subject but who rely on assurances of confidentiality. J.A. 352. These discussions can often lead to "unique and relevant, contextual comments" made by the confidential source, comments that deepen the story. Id.

The affidavits also recount numerous instances in which the confidentiality promised to sources was integral to a reporter's development of major stories critical to informing the public of the government's actions. See, e.g., J.A. 378-80 (affidavit of Dana Priest) (noting, among many stories, her reporting on the existence and treatment of military prisoners at Guantanamo Bay, Cuba; the abuse of prisoners in Abu Ghraib, Iraq; the existence of secret CIA prisons in Eastern Europe; and the "systematic lack of adequate care" for veterans at Walter Reed Army Medical Center relied upon confidential sources). Carl Bernstein, who has worked for the Washington Post and ABC News, writes that without his confidential source known as "Deep Throat," the investigation into the Watergate scandal — the break-in of the Democratic National Committee's offices in the Watergate Hotel and Office Building that led to the resignation of President Nixon — would never have been possible. J.A. 361-62. "Total and absolute confidentiality" was essential for Bernstein to cultivate the source. J.A. 362.

For all that the record establishes, common sense tells us the value of the reporter's privilege to journalism is one of the highest order. See Riley v. City of Chester, 612 F.2d 708, 714 (3d Cir.1979) ("The interrelationship between newsgathering, news dissemination and the need for a journalist to protect his or her source is too apparent to require belaboring."). Indeed, reporters "depend[] upon an atmosphere of confidentiality and trust" to carry out their mission, a mission critical to an informed and functioning democracy. Jaffee, 518 U.S. at 10, 116 S.Ct. 1923.

B.

Any consideration of the reporter's privilege must start with Branzburg, where the Supreme Court upheld, by a vote of five to four, the compulsion of confidential source information from reporters. Branzburg v. Hayes, 408 U.S. 665, 92 S.Ct. 2646, 33 L.Ed.2d 626 (1972). The majority opinion highlighted the "longstanding principle that `the public ... has a right to every man's evidence,' except for those persons protected by a constitutional, common law, or statutory privilege." Id. at 688, 92 S.Ct. 2646 (citations omitted). The opinion also stated that "news gathering is not without its First Amendment protections," id. at 707, 92 S.Ct. 2646, but the Court did not specify exactly what those protections might encompass, although it indicated that "[o]fficial harassment of the press" and bad faith investigations might fall within the parameters of the First Amendment's protection of reporters. Id. at 707-08, 92 S.Ct. 2646.

Further complicating matters is Justice Powell's "enigmatic concurring opinion," id. at 725, 92 S.Ct. 2646 (Stewart, J., dissenting), which is in part at odds with the majority opinion he joined. In the concurrence, Justice Powell emphasized "the limited nature of the Court's holding," and endorsed a balancing test, according to which "if the newsman is called upon to give information bearing only a remote and tenuous relationship to the subject of the investigation," then courts should consider the applicability of the reporter's privilege on a "case-by-case basis" by "the striking of a proper balance between freedom of the press and the obligation of all citizens to give relevant testimony with respect to criminal conduct." Id. at 709-10, 92 S.Ct. 2646 (Powell, J., concurring).

The full import of Justice Powell's concurrence continues to be debated. Some analogize the Branzburg majority opinion to a plurality opinion, and therefore assert Justice Powell's concurrence as the narrowest opinion is controlling. See In re Grand Jury Subpoena, Judith Miller, 438 F.3d 1141, 1148 (D.C.Cir.2006) (describing appellants' argument that in a five-to-four decision, "the opinion of the least encompassing justice [] determines the precedent set by the decision"); cf. McKoy v. North Carolina, 494 U.S. 433, 462 n. 3, 110 S.Ct. 1227, 108 L.Ed.2d 369 (1990) (arguing that a separate opinion "cannot add to what the majority opinion holds, binding the other four Justices to what they have not said; but it can assuredly narrow what the majority opinion holds, by explaining the more limited interpretation adopted by a necessary member of that majority") (Scalia, J., dissenting). Others, like my good friends in the majority, treat Justice Powell's concurrence as ancillary, see ante 495-96, and simply rejoin that "the meaning of a majority opinion is to be found within the opinion itself." McKoy, 494 U.S. at 448 n. 3, 110 S.Ct. 1227 (Blackmun, J., concurring).

Given this confusion, appellate courts have subsequently hewed closer to Justice Powell's concurrence — and Justice Stewart's dissent — than to the majority opinion, and a number of courts have since recognized a qualified reporter's privilege, often utilizing a three-part balancing test. See, e.g., United States v. Caporale, 806 F.2d 1487, 1504 (11th Cir.1986) (applying the reporter's privilege in the criminal context); United States v. Burke, 700 F.2d 70, 76-77 (2d Cir.1983) (recognizing the qualified privilege in criminal cases); Zerilli v. Smith, 656 F.2d 705, 711-13 (D.C.Cir.1981) (applying the reporter's privilege in a civil case). Indeed, a mere five years after Branzburg, a federal court of appeals confidently asserted that the existence of a qualified reporter's privilege was "no longer in doubt." Silkwood v. Kerr-McGee Corp., 563 F.2d 433, 437 (10th Cir.1977). In short, Justice Powell's concurrence and the subsequent appellate history have made the lessons of Branzburg about as clear as mud.

The Fourth Circuit, like our sister circuits, has applied Justice Powell's balancing test in analyzing whether to apply a reporter's privilege to quash subpoenas seeking confidential source information from reporters. We first explicitly adopted Justice Powell's balancing test in an en banc opinion in United States v. Steelhammer, 539 F.2d 373, 376 (4th Cir. 1976) (Winter, J., dissenting), adopted by the court en banc, 561 F.2d 539, 540 (4th Cir.1977). Then in LaRouche, we applied the reporter's privilege doctrine to a civil case, again citing Justice Powell's concurrence in Branzburg for authority. 780 F.2d at 1139. Following the lead of the Fifth Circuit, we applied a three-part test to help us balance the interests at stake in determining whether the reporter's privilege should be applied; that is, we considered "(1) whether the information is relevant, (2) whether the information can be obtained by alternative means, and (3) whether there is a compelling interest in the information." Id. (citing Miller v. Transamerican Press, Inc., 621 F.2d 721, modified, 628 F.2d 932 (5th Cir.1980)). We went on to find that there was no abuse of discretion when the district court denied LaRouche's motion to compel discovery of a reporter's sources because LaRouche "had not exhausted reasonable alternative means of obtaining [the] same information." LaRouche, 780 F.2d at 1139.

In a subsequent case in the criminal context, In re Shain, four reporters in South Carolina asserted the reporter's privilege to protect information gleaned from interviews with a state legislator. 978 F.2d 850, 851-52 (4th Cir.1992). But applying Justice Powell's principles, we rejected the reporters' claim on the ground that none of the reporters asserted that the interviews were confidential, that there were agreements to refuse revealing the identity of the interviewee, or that the government sought to harass the reporters. Id. at 853. Thus, although the reporter's privilege was not recognized in "the circumstances of this case," see id. at 854, it is clear to me that we have acknowledged that a reporter's privilege attaches in criminal proceedings given the right circumstances.

The most recent federal appellate court decision to address the reporter's privilege at length is In re Grand Jury Subpoena, Judith Miller, 438 F.3d 1141, 1145-49 (D.C.Cir.2006). In that case, the court rejected the reporter's privilege claim asserted by Judith Miller of The New York Times, stating that the Branzburg decision was dispositive. The majority there — as in this case — reasoned that the Supreme Court had not revisited the question of a reporter's privilege under the First Amendment after Branzburg, and that Justice Powell's concurrence did not detract from the precedential weight of the majority's conclusion that there was no First Amendment reporter's privilege, at least when there was no suggestion that the reporter was being pressed for information as a means of harassment or intimidation. Id. at 1145-49. In a thoughtful concurrence, though, Judge Tatel pointed to the ambiguities of the Branzburg decision, and noted that nearly every state and the District of Columbia has recognized a reporter's privilege. Nevertheless, Judge Tatel concluded that "if Branzburg is to be limited or distinguished in the circumstances of this case, we must leave that task to the Supreme Court." Id. at 1166 (Tatel, J., concurring). And although he felt constrained to deny applying a First Amendment privilege, Judge Tatel would have held that Rule 501 of the Federal Rules of Evidence provides for a reporter's privilege (though on the facts of that case, the privilege would have given way due to the extraordinary national security issue involved). See id. at 1177-78 (Tatel, J., concurring).

C.

On this background, I turn to the question now before the court: Are there circumstances in which a reporter may refuse to testify as to the identity of one of his confidential sources, when the government seeks this information as part of a criminal investigation, and there is no evidence of prosecutorial bad faith or harassment? Some appellate courts have used a three-part test, essentially identical to the test we announced in LaRouche in the civil context, to help determine whether to apply the reporter's privilege in criminal cases. See, e.g., United States v. Caporale, 806 F.2d 1487, 1504 (11th Cir.1986); United States v. Burke, 700 F.2d 70, 76-77 (2d Cir.1983). They require the moving party, i.e. the government, "to make a clear and specific showing" that the subpoenaed information is "highly material and relevant, necessary or critical to the maintenance of the claim, and not obtainable from other available sources." Burke, 700 F.2d at 77 (internal citations and quotation marks omitted). Cf. 28 C.F.R. § 50.10 (policy in regards to the issuance of subpoenas to members of the news media).

I, too, would recognize a qualified reporter's privilege in the criminal context, and evaluate the privilege using the three-part test enunciated in LaRouche as an "aid" to help "balance the interests involved." 780 F.2d at 1139. I would add a caveat to this general rule, however; in cases involving questions of national security, if the three-part LaRouche test is satisfied in favor of the reporter's privilege, I would require consideration of two additional factors: the harm caused by the public dissemination of the information, and the newsworthiness of the information conveyed.2 Cf. id. at 1139 (establishing a balancing test for the reporter's privilege in the civil context); In re Grand Jury Subpoena, Judith Miller, 438 F.3d at 1175 (Tatel, J., concurring) (stating that courts must "weigh the public interest in compelling disclosure, measured by the harm the leak caused, against the public interest in newsgathering, measured by the leaked information's value"). Thus, even when the LaRouche test favors recognizing the reporter's privilege, in matters of national security this privilege can still be overridden by pressing government interests. It is important to note that such a test does not depart from established precedent, to the contrary, it adheres to Justice Powell's concurrence in Branzburg that "[t]he asserted claim to privilege should be judged on its facts by the striking of a proper balance between freedom of the press and the obligation of all citizens to give relevant testimony with respect to criminal conduct." 408 U.S. at 710, 92 S.Ct. 2646 (Powell, J., concurring).

D.

Whatever the limits of who may claim reporter's privilege, it is clear that Risen — a full-time reporter for a national news publication, The New York Times — falls into the category of people who should be eligible to invoke the privilege. I also note that Risen has been offered immunity by the Government, so there is no Fifth Amendment issue with regard to compulsion of his testimony. The threshold inquiries having been satisfied, I turn to the question of whether the reporter's privilege should apply in this case, applying the test I announced herein.3

1.

The inquiry when applying the first LaRouche factor is the relevance of Risen's testimony to the Government's case. Unlike the Branzburg case, where the reporters had knowledge of suspected crimes that could be seriously damaging to individuals and the government, the Government here seeks a conviction for the very act of disclosure. The Government claims that Risen's testimony is valuable to its case against Sterling for revealing national defense secrets for two reasons: establishing venue and supporting the Government's case on the merits. With respect to the former, the Government bears the burden of proving by a preponderance of the evidence that "the essential conduct elements" of the charged offenses occurred within the Eastern District of Virginia. United States v. Ebersole, 411 F.3d 517, 524 (4th Cir.2005) (internal quotation marks omitted).

The record suggests the Government can show that Risen made phone calls from the Eastern District of Virginia to Sterling's Missouri residence. Furthermore, emails exchanged with Sterling used a server located in the Eastern District of Virginia. Of course, in order to prove venue, the Government must show that classified information was disclosed during these communications. It appears venue can be established without requiring Risen to disclose his confidential sources, limiting the relevance of his testimony. And as addressed below with regard to the value of Risen's testimony to the Government's case-in-chief, the circumstantial evidence that classified information was discussed appears to be strong,4 indicating that Risen's testimony regarding his confidential sources is by no means pertinent to the Government proving Sterling guilty.

2.

Turning to the second LaRouche factor, whether the information sought — the identity of the source of the leak — is available by other means, the Government claims Risen's testimony is a critical part of its case against Sterling largely because Risen is the only eyewitness to the crime; the other evidence is circumstantial.5 The Government's demonstration of its good-faith effort to obtain similar evidence through other means is a necessary part of its showing. See United States v. Cuthbertson, 651 F.2d 189, 195-96 (3d Cir.1981) (requiring a demonstration that the party seeking to overcome the reporter's privilege "demonstrate that he has made an effort to obtain the information from other sources") (quoting Criden, 633 F.2d at 358-59). But it is precisely because of the Government's diligence that it doth protest too much. An analysis of the circumstantial evidence shows the Government's case is not as weak as it or the majority claims, limiting the need for Risen's testimony.

First, the Government can demonstrate that Sterling showed Risen's book to Sterling's then-girlfriend in a bookstore and, without so much as opening it, Sterling told her that chapter nine discussed his work at the CIA.6 The book itself reveals details about Classified Program No. 1 that tend to link Sterling to chapter nine. For example, sections of the chapter are told from the point of view of the case officer responsible for Human Asset No. 1 — which was Sterling's responsibility — and the Government asserts that the chapter describes two classified meetings at which Sterling was the only common attendee.

Second, the Government has the aforementioned phone records demonstrating that Sterling and Risen called each other seven times between February 27 and March 31, 2003. The Government also has evidence that Sterling attempted to delete emails referencing meetings and shared information between Sterling and Risen, and parts of the emails were indeed obliterated. In one email that was not fully deleted, Risen asks Sterling, "Can we get together in early January?" J.A. 40. In another, Risen tells Sterling "I want to call you today[.] I'm trying to write the story.... I need your telephone number again." J.A. 40. Risen sent another email to Sterling, this time stating "I'm sorry if I failed you so far but I really enjoy talking to you and would like to continue," J.A. 41, an apparent reference to The New York Times's refusal to publish Risen's story on Classified Program No. 1.

Third, the prosecution expects to elicit at trial the testimony of a former United States intelligence official. Risen allegedly told this official, who occasionally discussed Risen's reporting with him, that Sterling was involved in recruiting a source for "an important operation" that "targeted [] the Iranian nuclear program," and that Sterling was frustrated by the perceived lack of recognition he received within the CIA for his efforts. Joint Classified App'x (J.C.A.) 622, 624-25. This official, the district court wrote, "told the grand jury that Risen had told him that Sterling was his source for information about the Iranian nuclear weapons operation."

Finally, the Government can also link Risen and Sterling in the reporting of classified information on a prior occasion: Risen's March 2002 New York Times article entitled "Fired by the C.I.A., He Says Agency Practiced Bias" noted that Sterling provided Risen with one of Sterling's classified performance evaluations. In short, the Government has made "[a]ll reasonable attempts ... to obtain information from alternative sources" as recommended by the Department of Justice's internal guidelines on subpoenas for testimony by news media, see 28 C.F.R. § 50.10. The Government's efforts have yielded multiple evidentiary avenues that, when presented together, may be used to establish what the Government sought to establish solely with testimony from Risen — that Sterling leaked classified information, rendering Risen's testimony regarding his confidential sources superfluous.

3.

The third LaRouche factor is whether the Government has a compelling interest in the information it seeks from Risen. Suffice it to say, the prosecution's body of evidence without Risen's testimony is strong.7 The frequency of the phone calls between Risen and Sterling, the forensically retrieved emails, the stories published in The New York Times, the testimony of a former United States intelligence official, and the bookstore eyewitness provide extensive circumstantial evidence of the crime and the court's venue. While Sterling may argue that other staff members who had access to national security information could have been the source of the leak, the Government, as it acknowledges, may simply call to the stand those staff members to ask whether they were Risen's source.

While the prosecution would undoubtedly be better off with Risen's testimony — none of the remaining pieces of evidence is a smoking gun — the balancing test cannot mean that the privilege yields simply because "no circumstantial evidence, or combination thereof, is as probative as Risen's testimony or as certain to foreclose the possibility of reasonable doubt."8 Brief for the United States at 14. The specificity of the information contained in chapter nine of Risen's book, coupled with the limited universe of individuals who had access to the information, the circumstantial evidence, and proof by negative implication, compose a reasonably strong case for the Government. As we have stated before, "circumstantial evidence is no less probative than direct evidence." Stamper v. Muncie, 944 F.2d 170, 174 (4th Cir.1991). I would therefore conclude that the Government has failed to demonstrate a sufficiently compelling need for Risen's testimony.

4.

Satisfied that the LaRouche factors weigh in favor of Risen's privilege from testifying as to his confidential sources, I turn next to newsworthiness and harm, the two additional factors I suggest should apply in a case involving national security information. On the present record, the newsworthiness of the leaked information appears to be substantial. The information contained in chapter nine of State of War covers the United States intelligence community's efforts concerning the development of the Iranian nuclear program. The chapter questions the competence of the CIA's management of Classified Program No. 1. Chapter nine discusses a plan to have a former Russian scientist give Iranian officials incorrect nuclear weapon design specifications in an attempt to determine the status of the Iranian nuclear weapons program, and to stall or thwart the progress of that program, perhaps for years. The blueprints were so deficient, the chapter opines, that the Russian scientist spotted a flaw almost immediately. Although the scientist explained this flaw to the CIA, Risen writes, the CIA proceeded with the plot. In a letter accompanying the blueprints, the Russian scientist disclosed to the Iranians the flaw he spotted in the plans. Because the Iranians had received scientific help from Russian and Chinese scientists, the chapter continues, and because Iran already had black market nuclear blueprints, Iranian scientists could likely differentiate the good from the flawed in the American blueprints. In other words, Risen asserts, Classified Operation No. 1 may have helped Iran advance its nuclear program. The chapter also describes the inadvertent disclosure to an Iranian double-agent of the identities of every spy the CIA had within Iran — information that was then turned over to Iranian security officials, who in turn arrested a number of those agents. Finally, the chapter recounts the CIA's inability to obtain more than "fragmentary information about Iran's nuclear program." J.S.A. 208.

This information is not extraneous. Quite the opposite, it portends to inform the reader of a blundered American intelligence mission in Iran. Since the United States' invasion of Iraq in 2003, our nation's focus has shifted to the nuclear capabilities of Iran, specifically whether Iran is attempting to build a nuclear bomb and how soon it can achieve the technical capabilities to do so. State of War was released in 2006 — three years after the Iraq invasion. The Iraq invasion was undertaken in part based on concerns that Iraq had developed weapons of mass destruction, possibly including nuclear weaponry. See J.S.A. 182. The apparent lack of weapons of mass destruction in Iraq, it has been argued, highlights a significant failure of United States intelligence. See J.A. 381. Risen himself contributed to our understanding of this alleged failure. See James Risen, "C.I.A. Held Back Iraqi Arms Data, U.S. Officials Say," The New York Times, July 6, 2001, at A1; J.S.A. 218-232 (chapter nine of State of War).

In a similar vein, Risen's investigation into the methods and capabilities of the United States foreign intelligence community with respect to the Iranian nuclear program is surely news of the highest import, particularly given the apparent contretemps made in the National Intelligence Estimate of 2007. See National Intelligence Council, National Intelligence Estimate, Iran: Nuclear Intentions and Capabilities (Nov. 2007), http://www.odni.gov/press_releases/20071203_release.pdf (asserting with "high confidence" that Iran in 2003 halted its nuclear weapons program, despite 2005 intelligence estimate noting that Iran is "determined to develop nuclear weapons"). Significant public speculation about the possibility of a conflict with Iran has repeatedly surfaced in recent years. See Seymour M. Hersh, "Iran and the Bomb," The New Yorker, June 6, 2011, http://www.newyorker.com/reporting/2011/06/06/110606fa.facts.hersh ("There is a large body of evidence ... including some of America's most highly classified intelligence assessments, suggesting that the United States could be in danger of repeating a mistake similar to the one made with Saddam Hussein's Iraq eight years ago — allowing anxieties about the policies of a tyrannical regime to distort our estimations of the state's military capabilities and intentions."). Risen's reporting on Iran's nuclear capabilities is also particularly relevant given the criticism of the national press for its perceived failure to scrutinize United States intelligence regarding Iraq's weapons capabilities. See James Risen, "C.I.A. Held Back Iraqi Arms Data, U.S. Officials Say," N.Y. Times, July 6, 2004, at A1. Indeed, it is hard to imagine many subjects more deserving of public scrutiny and debate.9

As a final step in the First Amendment inquiry, I would require the district court to balance the newsworthiness of the information against the harm caused by the leak.10 The present record is not well developed on this point. The district court understandably declined to conduct fact-finding on this issue because this factor had not been identified in prior case law. Moreover, the Government has not clearly articulated the nature, extent, and severity of the harm resulting from the leak.11 Without such evidence, it is impossible for a reviewing court to determine whether the First Amendment interest in presenting newsworthy information to the public — if indeed the district court finds the information newsworthy — is outweighed by the consequences of the leak. Moreover, although I recognize the difficultly of evaluating the government's interests in a case involving national security information, I am also mindful of the fact that "[t]he First Amendment interest in informed popular debate does not simply vanish at the invocation of the words `national security.'" United States v. Morison, 844 F.2d 1057, 1081 (4th Cir.1988) (Wilkinson, J., concurring). With all things considered, the district court was correct in holding that Risen was protected from disclosing his confidential sources by a First Amendment reporter's privilege.

I find it sad that the majority departs from Justice Powell's Branzburg concurrence and our established precedent to announce for the first time that the First Amendment provides no protection for reporters. Ante 496. Under the majority's articulation of the reporter's privilege, or lack thereof, absent a showing of bad faith by the government, a reporter can always be compelled against her will to reveal her confidential sources in a criminal trial. The majority exalts the interests of the government while unduly trampling those of the press, and in doing so, severely impinges on the press and the free flow of information in our society. The First Amendment was designed to counteract the very result the majority reaches today. In sum, I would affirm the district court's ruling as to Risen's assertion of a First Amendment reporter's privilege, albeit using the three-part LaRouche test and balancing the two additional factors identified herein: newsworthiness of the leaked information and the harm resulting from the leak.

E.

Even if I were not inclined to recognize a First Amendment privilege for a reporter in the criminal context given Branzburg, I would recognize a common law privilege protecting a reporter's sources pursuant to Federal Rule of Evidence 501.12 Rule 501 was promulgated three years after the Supreme Court's decision in Branzburg. See Pub.L. No. 93-595, 88 Stat. 1926 (1975). The Rule authorizes federal courts to create new evidentiary privileges using the "common law ... in the light of reason and experience." Fed. R.Evid. 501. The Rule "did not freeze the law governing the privileges of witnesses in federal trials at a particular point in our history, but rather directed federal courts to `continue the evolutionary development of testimonial privileges.'" Jaffee v. Redmond, 518 U.S. 1, 9, 116 S.Ct. 1923, 135 L.Ed.2d 337 (1996) (quoting Trammel v. United States, 445 U.S. 40, 47, 100 S.Ct. 906, 63 L.Ed.2d 186 (1980)). By adopting Rule 501, Congress has given authority to the courts to use case-by-case adjudication to find new evidentiary privileges. United States v. Weber Aircraft Corp., 465 U.S. 792, 803 n. 25, 104 S.Ct. 1488, 79 L.Ed.2d 814 (1984) ("Rule 501 was adopted precisely because Congress wished to leave privilege questions to the courts rather than attempt to codify them."). In light of Branzburg's insistence that "Congress has freedom to determine whether a statutory newsman's privilege is necessary and desirable and to fashion standards and rules as narrow or broad as deemed necessary to deal with the evil discerned," 408 U.S. at 706, 92 S.Ct. 2646, a full discussion of the reporter's privilege must reckon with Rule 501.

Testimonial privileges "are not lightly created nor expansively construed, for they are in derogation of the search for truth." United States v. Nixon, 418 U.S. 683, 710, 94 S.Ct. 3090, 41 L.Ed.2d 1039 (1974). But the Supreme Court and the circuit courts, using Rule 501, have recognized a number of testimonial privileges. See, e.g., Jaffee, 518 U.S. at 15, 116 S.Ct. 1923 (recognizing psychotherapist-patient privilege); Upjohn Co. v. United States, 449 U.S. 383, 386-90, 101 S.Ct. 677, 66 L.Ed.2d 584 (1981) (recognizing attorney-client privilege); Trammel v. United States, 445 U.S. 40, 51-53, 100 S.Ct. 906, 63 L.Ed.2d 186 (1980) (recognizing marital communications privilege); Goodyear Tire & Rubber Co. v. Chiles Power Supply, Inc., 332 F.3d 976 (6th Cir.2003) (recognizing settlement communications privilege); Riley v. City of Chester, 612 F.2d 708, 715 (3d Cir.1979) (recognizing a qualified reporter's privilege). All of these privileges are "distinctly exceptional," and have only been recognized because they serve a "public good transcending the normally predominant principle of utilizing all rational means for ascertaining truth." Jaffee, 518 U.S. at 9, 116 S.Ct. 1923 (internal quotation marks and citations omitted). In my view, the reporter-source privilege meets this high bar.

The Supreme Court has stated that "the policy decisions of the States bear on the question [of] whether federal courts should recognize a new privilege or amend coverage of an existing one," and "[i]t is of no consequence that recognition of the privilege in the vast majority of States is the product of legislative action rather than judicial decision." Id. at 12-13, 116 S.Ct. 1923. When the Branzburg decision issued, only seventeen states had recognized some protection for a reporter regarding his or her confidential sources. Branzburg, 408 U.S. at 689 n. 27, 92 S.Ct. 2646. Today, only one state, Wyoming, has not enacted or adopted a reporter's privilege. Thirty-nine states and the District of Columbia have shield laws for reporters, whether those shields are absolute or qualified. See Ala.Code § 12-21-142; Alaska Stat. § 09.25.300; Ariz.Rev.Stat. Ann. § 12-2237; Ark.Code Ann. § 16-85-510; Cal. Const. Art. I, § 2(b); Cal. Evid.Code § 1070; Colo.Rev.Stat. §§ 13-90-119, 24-72.5-101; Conn. Gen.Stat. Ann. § 52-146t; Del.Code Ann. tit. 10, § 4320; D.C.Code § 16-4701; Fla.Stat. § 90.5015; Ga.Code Ann. § 24-9-30; Haw.Rev.Stat. § 621, as amended by 2011 Haw. Sess. Laws ch. 113 (June 14, 2011); 735 Ill. Comp. Stat. 5/8-901; Ind.Code Ann. §§ 34-46-4-1, -2; Kan. Stat. Ann. § 60-480; Ky.Rev.Stat. Ann. § 421.100; La.Rev.Stat. Ann. § 45:1451; Md.Code Ann. Cts. & Jud. Proc. § 9-112; Mich. Comp. Laws § 767.5a; Minn.Stat. § 595.021; Mont. Code Ann. § 26-1-901; Neb.Rev.Stat. § 20-144; Nev.Rev.Stat. Ann. § 49.275; N.J. Stat. Ann. § 2A:84A-21; N.M. Stat. Ann. § 38-6-7; N.Y. Civ. Rights Law § 79-h; N.C. Gen.Stat. § 8-53.11; N.D. Cent.Code § 31-01-06.2; Ohio Rev.Code Ann. § 2739.12; Okla. Stat. Ann. tit. 12, § 2506; Or.Rev.Stat. § 44.510; 42 Pa. Cons.Stat. Ann. § 5942; R.I. Gen. Laws § 9-19.1-1; S.C.Code Ann. § 19-11-100; Tenn.Code Ann. § 24-1-208; Tex. Civ. Prac. & Rem.Code Ann. §§ 22.021-22.027; Utah Order 08-04 [Utah R. Evid. 509]; Wash. Rev.Code Ann. § 5.68.010; 2011 W. Va. Acts 78 (to be codified at W. Va.Code § 57-3-10); Wis. Stat. Am. § 885.14. In ten states without statutory shield laws, the privilege has been recognized in some form or another by the courts. See State v. Salsbury, 129 Idaho 307, 924 P.2d 208 (1996); Winegard v. Oxberger, 258 N.W.2d 847 (Iowa 1977), cert. denied, 436 U.S. 905, 98 S.Ct. 2234, 56 L.Ed.2d 402 (1978); In re Letellier, 578 A.2d 722 (Me.1990); In re John Doe Grand Jury Investigation, 410 Mass. 596, 574 N.E.2d 373 (1991); Sinnott v. Boston Retirement Bd., 402 Mass. 581, 524 N.E.2d 100, cert. denied, 488 U.S. 980, 109 S.Ct. 528, 102 L.Ed.2d 560 (1988); State ex rel. Classic III v. Ely, 954 S.W.2d 650, 653 (Mo.Ct.App.1997); State v. Siel, 122 N.H. 254, 444 A.2d 499 (1982); Hopewell v. Midcontinent Broad. Corp., 538 N.W.2d 780, 782 (S.D.1995), cert. denied, 519 U.S. 817, 117 S.Ct. 69, 136 L.Ed.2d 30 (1996); State v. St. Peter, 132 Vt. 266, 315 A.2d 254 (1974); Brown v. Commonwealth, 214 Va. 755, 204 S.E.2d 429 (1974); Hawkins v. Williams, No. 29,054 (Hinds County Circuit Court, Mississippi, Mar. 16, 1983) (unpublished). A number of these jurisdictions — Alabama, Arizona, California, Delaware, the District of Columbia, Indiana, Kentucky, Maryland, Montana, Nebraska, Nevada, New York, Ohio, Oklahoma, Oregon, and Pennsylvania — make the privilege an absolute bar to compelling a reporter to divulge his sources. On the basis of "the uniform judgment of the States," the Supreme Court recognized the psychotherapist-patient privilege. Jaffee, 518 U.S. at 14, 116 S.Ct. 1923. The landscape in regards to the reporter's privilege has changed drastically since Branzburg. The unanimity of the States compels my conclusion that Rule 501 calls for a reporter's privilege.

F.

The paramount importance of the free press guaranteed by our Constitution compels me to conclude that the First Amendment encompasses a qualified reporter's privilege. Using the factors identified herein and given the facts at hand, Risen must be protected from disclosing the identity of his confidential sources. This is consistent with Branzburg and the need for courts to balance "freedom of the press" against "the obligation of all citizens to give relevant testimony with respect to criminal conduct." 408 U.S. at 724, 92 S.Ct. 2646 (Powell, J., concurring). Moreover, given the near unanimity of the states with regard to a reporter's privilege, I would recognize the privilege under Federal Rule of Evidence 501. Thus, I would affirm the district court's order quashing the trial subpoena and denying the Government's motion to admit Risen's testimony as to the source relied upon by Risen for Chapter Nine of State of War. As to Issue I, then, I respectfully dissent from the majority's holding.