Jihad Dhiab v. Donald J. Trump

United States Court of Appeals

FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued September 8, 2016 Decided March 31, 2017

No. 16-5011

JIHAD DHIAB, DETAINEE, GUANTANAMO BAY NAVAL

STATION AND SHAKER AAMER, AS NEXT FRIEND OF JIHAD

DHIAB,

APPELLEES

v.

DONALD J. TRUMP, PRESIDENT OF THE UNITED STATES,

ET AL.,

APPELLANTS

HEARST CORPORATION, ET AL.,

APPELLEES

Consolidated with 16-5012

Appeals from the United States District Court

for the District of Columbia

(No. 1:05-cv-01457)

Catherine H. Dorsey, Attorney, U.S. Department of Justice,

argued the cause for respondents-appellants/cross-appellees

(US). With her on the briefs were Benjamin C. Mizer, Principal

Deputy Assistant Attorney General, and Matthew M. Collette,

Attorney.

2

David A. Schulz argued the cause for intervenors-

appellees/cross-appellants. With him on the briefs was Matthew

L. Schafer.

Rachel B. Levinson-Waldman was on the brief for amici

curiae Brennan Center for Justice and Electronic Frontier

Foundation in support of intervenors-appellees.

Hina Shamsi and Arthur B. Spitzer were on the brief for

amici curiae American Civil Liberties Union of the Nation's

Capital and The Reporters Committee for Freedom of The Press

in support of intervenors-appellees/cross-appellants.

Before: ROGERS, Circuit Judge, and WILLIAMS and

RANDOLPH, Senior Circuit Judges.

Opinion for the Court filed by Senior Circuit Judge

RANDOLPH, with whom Circuit Judge ROGERS and Senior

Circuit Judge WILLIAMS join except as to Part II.

Opinion concurring in part and concurring in the judgment

filed by Circuit Judge ROGERS.

Opinion concurring in part and concurring in the judgment

filed by Senior Circuit Judge WILLIAMS.

RANDOLPH, Senior Circuit Judge: The government’s

appeal, and the intervenors’ cross-appeal, are from the district

court’s orders releasing video recordings made at the United

States Naval Base, Guantanamo Bay, Cuba. The recordings are

of military personnel removing a detainee from his cell,

transporting him to a medical unit, and force-feeding him to

keep him alive while he was on a hunger strike.

3

The government classified these recordings as “SECRET”

because disclosing them could damage the national security.

The district court decided that under the Constitution the public

has a right to view the recordings because the detainee’s

attorney filed some of them under seal, at which point the

recordings became part of the court’s record. The government’s

appeal is on the ground that the public has no such constitutional

right. The intervenors’ cross-appeal is on the ground that

several categories of redactions the court approved prior to

public release were too extensive.

I.

The case began when Abu Wa’el (Jihad) Dhiab filed a

petition for a writ of habeas corpus to prevent the government

from force-feeding him. The district court denied Dhiab’s

motion for a preliminary injunction, finding that it lacked habeas

jurisdiction to correct conditions of confinement. Dhiab v.

Obama, 952 F. Supp. 2d 154, 155 (D.D.C. 2013). On appeal, a

panel of this court held that a Guantanamo habeas petitioner

may seek not only relief from confinement, the traditional

remedy in habeas corpus, but also an injunction to alter the

conditions of his confinement. Aamer v. Obama, 742 F.3d 1023,

1033 (D.C. Cir. 2014).1

On remand, Dhiab moved again for a preliminary

injunction, this time challenging particular government force-

feeding practices. He also filed an emergency application for a

temporary restraining order. The district court denied both

1

There is a conflict in the circuits regarding whether complaints

about conditions of confinement are cognizable in habeas cases.

Aamer, 742 F.3d at 1036-38. See also Nettles v. Grounds, 830 F.3d

922, 933 (9th Cir. 2016); Spencer v. Haynes, 774 F.3d 467, 470 n.6

(8th Cir. 2014).

4

motions. Dhiab v. Obama, 74 F. Supp. 3d 16, 19 (D.D.C. 2014);

Order, Dhiab v. Obama, No. 05-01457 (GK), (D.D.C. June 16,

2014), ECF No. 254. In considering Dhiab’s motions, the

district court ordered the government to provide Dhiab’s

attorney, who had been given a security clearance, copies of the

video recordings, the existence of which the government had

disclosed. After the government complied with the order, to

which it objected, Dhiab’s attorney filed some of the recordings

under seal.2

The government recorded Dhiab’s removal from his cell

and his force-feeding in order to train military guards about how

to handle detainees in such circumstances. In classifying each

recording as “SECRET,” we shall assume that the government

complied with Executive Order No. 13,526, 75 Fed. Reg. 707

(Dec. 29, 2009).3 This Executive Order, and those of past

Presidents, see Dep’t of Navy v. Egan, 484 U.S. 518, 527-28

(1988), specified three levels of classified national security

information: “TOP SECRET,” “SECRET,” and

“CONFIDENTIAL.” The “SECRET” classification is reserved

for “information, the unauthorized disclosure of which

reasonably could be expected to cause serious damage to the

national security” of the United States. Exec. Order No. 13,526

2

The classified recordings were sealed as required by the general

protective order governing all Guantanamo Bay detainee litigation. In

re Guantanamo Bay Detainee Litig., 577 F. Supp. 2d 143, 153

(D.D.C. 2008). The district court issued a supplemental protective

order governing the recordings. That order prevented Dhiab’s counsel

from sharing the recordings with counsel representing detainees who

were not challenging the forcible cell-extraction and enteral-feeding

procedures.

3

The intervenors did not claim, let alone establish, that the

classification of the recordings was improper. The district court

expressed no opinion on the subject.

5

§ 1.2(a)(2). Unauthorized disclosure of such classified

information can be a federal criminal offense. See, e.g., 18

U.S.C. § 793.

Press organizations – sixteen of them – sought to intervene

in Dhiab’s habeas case and asked the district court to unseal the

recordings Dhiab’s attorney had filed. Their motion asserted

that under the First Amendment, and common law, the public

had a right to see these recordings because the recordings had

become part of the record of Dhiab’s habeas corpus proceeding.

The government did not oppose their intervention motion but it

did oppose the organizations’ motion to unseal the recordings.

In support of its opposition, the government filed declarations

from military officers explaining how release of the classified

recordings would threaten national security.

In October 2014, the district court granted the

organizations’ motion to intervene and their motion to unseal the

recordings, subject to further proceedings to determine what

portions of the recordings should be redacted to protect

information identifying government personnel. Dhiab v.

Obama, 70 F. Supp. 3d 486, 501 (D.D.C. 2014).

In December of that year, the government released Dhiab

from Guantanamo and transferred him to the control of the

government of Uruguay, thus rendering his habeas petition

moot.

Our court nevertheless heard the government’s appeal from

the district court’s October 2014 order unsealing the recordings

and determined that we lacked appellate jurisdiction. We lacked

appellate jurisdiction because the district court’s ruling was not

an appealable final order or an appealable collateral order in

light of further proceedings the court had scheduled. Dhiab v.

Obama, 787 F.3d 563, 565-66 (D.C. Cir. 2015). We encouraged

6

the district court on remand to consider the government’s

additional declarations, filed in support of a stay of the unsealing

order pending that appeal. Id. at 567. These new declarations

“set out the harm associated with release of the videotapes in

considerably more detail” than the earlier ones. Id.

When the case returned to the district court, it denied the

government’s motion for reconsideration and, over the

government’s objection, ordered the recordings4 released after

the government redacted identifying information such as faces,

voices and names of government personnel. Dhiab v. Obama,

141 F. Supp. 3d 23, 28-29 (D.D.C. 2015). By then Dhiab was

no longer at Guantanamo.5 Even so, the intervenors persisted in

their objections to some of the redactions and sought

reinstatement of some of the deleted audio portions of the

recordings or a transcript and subtitles.6 The district court

4

There is some discrepancy about the number of recordings at

issue. The district court’s initial opinion referred to twenty-eight

recordings already on file, but noted that the government had provided

Dhiab with an additional four recordings. Dhiab, 70 F. Supp. 3d at

492. The opinions on remand refer to thirty-two recordings. Dhiab

v. Obama, 141 F. Supp. 3d 23, 27 (D.D.C. 2015); Dhiab v. Obama,

151 F. Supp. 3d 28, 29 (D.D.C. 2015).

5

Dhiab did not file a brief in this appeal.

6

The government first redacted a mutually selected sample of ten

recordings – eight of the thirty-two recordings produced during

litigation and a compilation recording created by each party. In

addition to covering guards’ faces, uniform patches and other unique

identifiers, the government muted the audio whenever a guard was

speaking. After viewing the set of eight redacted recordings, Dhiab’s

counsel objected to the muting of guards’ voices and requested that the

audio redactions be limited. In the alternative, Dhiab’s counsel

requested the addition of subtitles and the simultaneous release of a

7

denied the intervenors’ motion regarding redaction, Dhiab v.

Obama, 151 F. Supp. 3d 28, 29 (D.D.C. 2015), ordered the

redacted recordings unsealed on or before January 11, 2016, and

granted a stay pending this appeal and cross-appeal, Order,

Dhiab v. Obama, No. 05-01457 (GK), (D.D.C. Jan. 4, 2016),

ECF No. 418.

II.

The intervenors’ claim that the Constitution requires this

national security information, properly classified as “SECRET,”

to be divulged to the world because a lawyer representing a

Guantanamo detainee filed some of the recordings under seal in

his client’s now-moot habeas corpus action is untenable. It is

important to bear in mind that the Constitution gives “the

President as head of the Executive Branch and as Commander

in Chief” the “authority to classify and control access to

information bearing on national security . . ..” Egan, 484 U.S.

at 527.

Through the years our government has been steadfast in

protecting information that, if made public, would jeopardize the

security of the United States. Statutes, longstanding regulations,

comprehensive Executive Orders, rules of the Chief Justice of

the United States, local rules and practices of the federal courts

– and more, enforce and support the President’s constitutional

duty to prevent our government’s secret information from seeing

the light of day, in judicial proceedings or otherwise.

Here the government established that the recordings of

Dhiab were properly classified as “SECRET.” The district court

did not rule otherwise, and the intervenors did not claim, let

transcript. The intervenors, who have not seen the recordings, joined

this motion.

8

alone show, that the classifications were improper. The

government submitted declarations, about which more later,

demonstrating the harm that would result from releasing any of

these recordings, redacted or not.

Yet the intervenors insist that under the First Amendment,

classified information submitted under seal in a judicial

proceeding becomes fair game for a judicial disclosure order,

such as the one the district court issued in this case. Neither the

First Amendment nor any other provision of the Constitution

stands for such a principle.

The intervenors rely heavily on Press-Enterprise Co. v.

Superior Court, 478 U.S. 1, 8-9 (1986).7 This Press-Enterprise

II decision will not bear the weight they place on it. The

Supreme Court framed the question in Press-Enterprise II this

way: whether the public had “a First Amendment right of access

to the transcript of a preliminary hearing growing out of a

criminal prosecution.” Id. at 3. The Court put the question in

terms of the public’s right because the “First Amendment

generally grants the press no right to information about a trial

superior to that of the general public.” Nixon v. Warner

Commc’ns, Inc., 435 U.S. 589, 609-10 (1978) (citing Estes v.

Texas, 381 U.S. 532, 589 (1965) (Harlan, J., concurring); Saxbe

v. Washington Post Co., 417 U.S. 843 (1974); Pell v. Procunier,

417 U.S. 817 (1974); and Zemel v. Rusk, 381 U.S. 1, 16-17

(1965)).

Press-Enterprise II discovered a constitutional right in the

public, although it was a qualified one: such proceedings may be

7

See Richmond Newspapers, Inc. v. Virginia, 448 U.S. 555, 558-

81 (1980) (opinion of Chief Justice Burger, joined by Justices White

and Stevens), id. at 584-98 (opinion of Justice Brennan, joined by

Justice Marshall).

9

sealed but only if “specific, on the record findings are made

demonstrating that ‘closure is essential to preserve higher values

and is narrowly tailored to serve that interest.’” 478 U.S. at 13-

14 (quoting Press-Enterprise Co. v. Super. Ct. (Press-Enterprise

I), 464 U.S. 501, 510 (1984) (considering the closure of voir

dire in a criminal trial)). The district court invoked this

formulation in ordering the classified Guantanamo recordings

unsealed. Dhiab, 70 F. Supp. 3d at 493.

Press-Enterprise II is not comparable to this case. Two

differences are immediately apparent. When the Court wrote of

the importance of public access to evidentiary proceedings it

could not possibly have had in mind classified national security

information. The case came up from a California state court. In

criminal cases in California a preliminary hearing functioned

“much like a full-scale trial.” 478 U.S. at 7. The sealed record

in Press-Enterprise II consisted of testimony and exhibits

relating to murder charges, not classified material. Id. at 4.

The second difference is just as obvious. Unlike Dhiab’s

case, which was civil in nature,8 the underlying action in Press-

Enterprise II was a criminal prosecution. When it comes to

classified national security information the Supreme Court has

decided that the distinction makes a difference. See United

States v. Reynolds, 345 U.S. 1, 12 (1953). In criminal cases, the

government initiates the prosecution. Access and disclosure

rights in criminal cases “do not endanger the government’s

paramount interest in national security. The government’s

interest can be protected by dismissal of the prosecution or less

drastic concessions by the government in a criminal case.”

Bruce E. Fein, Access to Classified Information: Constitutional

8

See, e.g., Fay v. Noia, 372 U.S. 391, 423 (1963), deciding that

the writ of habeas corpus is a “civil remedy for the enforcement of the

right to personal liberty” not “a stage of” a criminal proceeding.

10

and Statutory Dimensions, 26 WM. & MARY L. REV. 805, 828

(1985).9 Matters are quite different in civil cases: “the

Government is not the moving party, but is a defendant . . ..”

Reynolds, 345 U.S. at 12. For this reason, the Court in Reynolds

held that the rationale behind access to national security

information in criminal cases had “no application in a civil

forum.” Id.10 For the same reason, a noted commentator

concluded that “plaintiffs suing the United States enjoy no right

of access to classified information pertinent to the litigation.”

Fein, supra, at 828.

There are additional reasons why Press-Enterprise II does

not apply to this case. To reach its result, the Supreme Court

recounted the English tradition of public criminal trials,

9

The Classified Information Procedures Act, 18 U.S.C. app. 3,

(CIPA) governs the handling of classified evidence in criminal

proceedings. CIPA was enacted to limit the practice of criminal

defendants threatening to disclose classified information in order to

force the government to dismiss the charges. Under CIPA the court

may review the admissibility of classified evidence at a preliminary

hearing held in camera. Id. § 6(a). If the classified information is

admissible, the government can suggest a substitute for the

information or concede the fact the information tends to prove. Id.

§ 6(c). If the court rejects these measures, the government may

declassify the information or dismiss the prosecution. Id. § 6(e).

10

We also have recognized the difference between criminal and

civil proceedings: “Neither the Supreme Court nor this Court has

applied the [First Amendment right of access] outside the context of

criminal judicial proceedings or the transcripts of such proceedings.”

Ctr. for Nat’l Sec. Studies v. U.S. Dep’t of Justice, 331 F.3d 918, 935

(D.C. Cir. 2003).

11

beginning – the Court wrote – before the Norman conquest.11

Press-Enterprise II, 478 U.S. at 8. Although the Court did not

say as much, the idea apparently was that the Framers of the

First Amendment must have had this history in the back of their

collective minds. See Globe Newspaper Co. v. Super. Ct., 457

U.S. 596, 604-05 (1982).12 The Court cited no historical

11

We wrote in In re Motions of Dow Jones & Co., 142 F.3d 496,

502 (D.C. Cir. 1998):

The Supreme Court ruled in Gannett Co. v. DePasquale, 443

U.S. 368 (1979), that the First Amendment did not entitle

members of the media to attend a pretrial suppression hearing, at

least so long as they could receive copies of the transcript at a

later date after the danger of prejudice to the defendant had

passed. The Court reached the opposite conclusion in

Press–Enterprise Co. v. Superior Court (Press–Enterprise II),

478 U.S. 1, 12, 13 (1986), holding that a “qualified First

Amendment right of access attaches to preliminary hearings in

California”—that is, probable cause hearings “sufficiently like a

trial to justify” the conclusion that they should be open. One of

the main differences between the two cases was that although the

“near uniform practice of state and federal courts has been to

conduct preliminary hearings in open court,” Press–Enterprise II,

478 U.S. at 10; see also id. at 8, the Court in Gannett could

identify no long-standing tradition of public access to pretrial

suppression hearings. 443 U.S. at 384–93.

12

See Richmond Newspapers, Inc. v. Virginia in which the Chief

Justice explained that history was important because it showed that the

“Bill of Rights was enacted against the backdrop of the long history

of trials being presumptively open.” 448 U.S. at 575 (Burger, C.J.,

plurality opinion). The Court gave the same explanation in Press-

Enterprise I, 464 U.S. at 509 n.8. Justice Stevens, in his dissenting

opinion in Press-Enterprise II, pointed out that “in our prior cases

history mattered primarily for what it revealed about the intentions of

the Framers and ratifiers of the First Amendment.” 478 U.S. at 22.

12

evidence supporting that proposition and the wording of the

First Amendment reveals no such understanding, as the Court

itself acknowledged in an earlier opinion. Id. at 604. “With

neither the constraint of text nor the constraint of historical

practice, nothing would separate the judicial task of

constitutional interpretation from the political task of enacting

laws currently deemed essential.” In re Reporters Comm. for

Freedom of the Press, 773 F.2d 1325, 1332 (D.C. Cir. 1985)

(Scalia, J.).

In habeas corpus cases, there is no tradition of public access

comparable to that recounted in Press-Enterprise II with respect

to criminal trials.13 Habeas corpus proceedings do not involve

juries. Since the beginning they have been decided by judges.

Early English courts were in session for only a few months each

year. PAUL D. HALLIDAY, HABEAS CORPUS: FROM ENGLAND TO

EMPIRE 355 n.79 (2010). Yet from the fifteenth to eighteenth

century, English courts regularly adjudicated habeas petitions

between sessions. Id. at 56-57. At such times the English

judges required jailers to make their returns to the writ to the

judge’s private chambers or to the judge’s home. Id. at 54. The

judge then made his habeas decision in private. Id. Between

1500 and 1800, about one-fifth of the writs the judges of

England issued required the jailer make the return to chambers.

Id. Although English judges more frequently requested returns

to chambers during the vacations, the practice also occurred

during terms of court. Id. The Habeas Corpus Act of 1679,

which Blackstone described as the bulwark of English liberties,

1 WILLIAM BLACKSTONE, COMMENTARIES ON THE LAWS OF

ENGLAND 133 (1765), expressly authorized the courts to issue

13

The district court here, Dhiab, 70 F. Supp. 3d at 494, and in In

re Guantanamo Bay Detainee Litigation, 624 F. Supp. 2d 27, 35

(D.D.C. 2009), lamented the lack of any discussion in the caselaw of

this subject.

13

writs of habeas corpus during vacations, thus continuing this

longstanding practice. 31 Car. 2 c. 2.

Of course in this country, proceedings in open court are the

norm, although there are well-established exceptions. See, e.g.,

In re Motions of Dow Jones & Co., 142 F.3d at 502-05. But of

importance here is not just the absence of any “unbroken,

uncontradicted history” of public attendance at habeas corpus

proceedings in eighteenth-century England. Richmond

Newspapers, 448 U.S. at 573 (Burger, C.J., plurality opinion).

More significant is that from the beginning of the republic to the

present day, there is no tradition of publicizing secret national

security information involved in civil cases, or for that matter,

in criminal cases.14 The tradition is exactly the opposite.15

To appreciate this one need only consider what occurs when

classified information might be revealed during an oral

14

See United States v. McVeigh, 119 F.3d 806, 812 (10th Cir.

1997) (“Assuming that the Press-Enterprise II right of access extends

to at least some types of judicial documents, the question remains

whether that right applies to the particular types of documents at issue

in this case.”); In re Grand Jury Subpoena, 103 F.3d 234, 242 (2d Cir.

1996) (holding that under Press-Enterprise II, the court must focus on

the specific motion that the moving party seeks to disclose publicly,

not the broader grand jury investigation or hearings in which the

motion arose).

15

See, e.g., McGehee v. Casey, 718 F.2d 1137, 1147 (D.C. Cir.

1983): “As a general rule, citizens have no first amendment right of

access to traditionally nonpublic government information. See, e.g.,

Houchins v. KQED, Inc., 438 U.S. 1, 8-9 (1978) (plurality opinion);

id. at 16 (Stewart, J., concurring); Saxbe v. Washington Post Co., 417

U.S. 843, 849 (1974); Pell v. Procunier, 417 U.S. 817, 831-32

(1974).”

14

argument in our courtroom. In such cases the court’s security

officer “seals” the courtroom – that is, excludes the public,

including the press. This is done not only in criminal appeals

but also in civil cases. The procedures followed are those the

Chief Justice of the United States has prescribed to protect

classified information in criminal cases pursuant to the

Classified Information Procedures Act. 18 U.S.C. app. 3 § 9

note (Security Procedures Established Pursuant to Pub. L. No.

96-456, 94 Stat. 2025, by the Chief Justice of the United States

for the Protection of Classified Information). One may be

confident that over many years none of the members of our

court, past and present, ever supposed that in complying with the

Chief Justice’s rules, we were somehow violating the

Constitution.

Add to United States v. Reynolds, already mentioned, the

case of Totten v. United States, 92 U.S. 105 (1875).16 Both of

these civil cases are well-known instances in the long history of

protecting national security secrets of the United States.17

Reynolds held that in a suit against the government, the plaintiff

had no right to discover military or state secrets; the privilege

against revealing such information was, the Court wrote, “well

established.” 345 U.S. at 6-7. The Court in Reynolds described

its 1875 decision in Totten as having dismissed a civil suit, the

subject of which was a “state secret.” Id. at 11 n.26. In

Reynolds, even if the plaintiffs’ need for the information had

been “compelling,” no showing of necessity could justify a court

order requiring the government to reveal “military secrets.” Id.

at 11. Relying on Reynolds, our court has held that in order to

16

See generally Daniel L. Pines, The Continuing Viability of the

1875 Supreme Court Case of Totten v. United States, 53 ADMIN. L.

REV. 1273 (2001).

17

See Fein, supra, at 828-31.

15

protect national security information, the district court “need

only be satisfied that there is a reasonable danger that

compulsion of the evidence will expose military matters which,

in the interest of national security, should not be divulged.”

Halkin v. Helms, 598 F.2d 1, 9 (D.C. Cir. 1978) (internal

quotation and emphasis omitted).

The government also discusses cases arising under the

Freedom of Information Act. Although this legislation requires

much government information to be disclosed to the public,

Congress expressly exempted properly classified information.

5 U.S.C. § 552(b)(1). To determine whether the government’s

classification was proper, the court may review the document

itself, but only ex parte and in camera, see 5 U.S.C.

§ 552(a)(4)(B) – that is, in private – as we mentioned in a case

dealing with information regarding detainees at Guantanamo.

ACLU v. Dep’t of Def., 628 F.3d 612, 626-27 (D.C. Cir. 2011).

If the government had properly classified the document, the

litigant, and hence the public, had no right to see it.

In Boumediene v. Bush, the case establishing the right of

Guantanamo detainees to bring habeas actions, the Court

thought the unique proceedings it was authorizing might risk

“widespread dissemination of classified information.” 553 U.S.

723, 796 (2008). To guard against this the Court wrote that the

government “has a legitimate interest in protecting sources and

methods of intelligence gathering; and we expect that the

District Court will use its discretion to accommodate this

interest to the greatest extent possible. Cf. United States v.

Reynolds, 345 U.S. 1, 10 (1953) (recognizing an evidentiary

privilege in a civil damages case where ‘there is a reasonable

danger that compulsion of the evidence will expose military

matters which, in the interest of national security, should not be

divulged’).” Boumediene, 553 U.S. at 796.

16

To that end, Guantanamo habeas proceedings have been

litigated under orders designed to protect classified information.

See, e.g., In re Guantanamo Bay Detainee Litig., 577 F. Supp.

2d 143 (D.D.C. 2008). These protective orders require not only

that classified information be maintained under seal, but also

that counsel (with a security clearance) not disclose classified

information at any hearing or proceeding. Id. at 150, 153. The

government informs us that Guantanamo habeas cases routinely

involve closed sessions to protect classified information from

the public eye. Appellants’ Reply Br. at 20. Dhiab’s case is no

exception: in his habeas proceedings, the district court held an

evidentiary hearing from which the public was excluded. Id. at

21.

As against this, the intervenors are unable to cite a single

case in which a court – other than the district court here – found

that the First Amendment compelled public disclosure of

properly classified national security information in a habeas

proceeding, or in any other type of civil proceeding.

Press-Enterprise II spoke of a need to take into account

“experience and logic” in determining whether the First

Amendment required a record of a judicial proceeding to be

released to the world. 478 U.S. at 9. The “experience” in

habeas corpus cases and in cases involving classified documents

have already been discussed.

As to “logic,” it is important to remember that logic does

not give starting points. First principles do. For this case the

starting point was established at the Founding. The preamble to

the Constitution gives equal billing to the national defense and

“the Blessings of Liberty.” U.S. CONST. pmbl. As the Supreme

Court stated, there is no higher value than the security of the

nation, a value the Court deemed a “compelling interest.” Haig

v. Agee, 453 U.S. 280, 307 (1981) (internal quotation omitted).

17

Press-Enterprise II therefore does not apply to this case and

neither the intervenors nor the public at large have a right under

the First Amendment to receive properly classified national

security information filed in court during the pendency of

Dhiab’s petition for a writ of habeas corpus.

III.

Even if the intervenors had a qualified First Amendment

right of access to the Dhiab recordings, we would still reverse

the district court’s decision. The court’s ruling that the

government failed to show a “substantial probability” of harm

to a higher value was clear error.18 Press-Enterprise II, 478 U.S.

at 14 (internal quotation omitted).

The government identified multiple ways in which

unsealing these recordings would likely impair national security.

Two of these risks – detainees triggering forcible encounters and

developing countermeasures – together and individually, were

enough to prevent these recordings from becoming public. The

government’s declarations explained that the recordings would

enable detainees, assisted by outside militants, to develop

countermeasures to the guards’ cell-extraction and enteral-

feeding techniques. The district court dismissed this prospect

because the government had “already released substantial

18

This court has not decided the proper standard of review for a

district court’s “compelling interest” analysis. The circuits disagree

about whether to apply the ordinary “clear error” standard, FED. R.

CIV. P. 52(a)(6), or “de novo” review in light of Bose Corp. v.

Consumers Union of United States, Inc., 466 U.S. 485, 508-10, 508

n.27 (1984). See, e.g., United States v. Erie Cty., 763 F.3d 235, 238

(2d Cir. 2014); In re Hearst Newspapers, L.L.C., 641 F.3d 168, 174-

75 (5th Cir. 2011). We need not decide this issue because we would

reverse even if the standard were clear error.

18

information” about these procedures and the detainees were

already “intimately familiar” with them. Dhiab, 70 F. Supp. 3d

at 497-98. The government’s declarations contradict the court’s

assessment. The recordings of the feeding process contain

“significantly more information than previously released

imagery,” Declaration of Rear Admiral Kyle J. Cozad, October

15, 2014, ¶ 4, and the publicly released information about cell

extractions was outdated and described techniques not being

used at Guantanamo. Id. ¶¶ 5-6. The recordings also show

elements of the procedures that “take place outside the

detainee’s presence” or “shielded from the detainee or otherwise

obstructed from view.” Declaration of Rear Admiral Richard

W. Butler, ¶ 12; Declaration of Rear Admiral Kyle J. Cozad,

November 26, 2014, ¶ 9.

The government’s expert judgment was that militants could

study the recordings repeatedly and slowly, looking for

“patterns” of “mistakes” not identifiable from first-hand

experience or written descriptions. Id. ¶ 14. Even if the

recordings contained no new information, the government thus

demonstrated good cause for sealing them. Information gleaned

from the recordings could reach current detainees, who

communicate with family members and other outside persons

and have some access to outside media. Id. ¶ 6. Militants could

also use the recordings to train fighters the government may

capture and detain in the future. Id. ¶ 18. When detainees resist

what are already hazardous procedures for the guards, this could

further endanger government personnel at Guantanamo. Guards

have been kicked, grabbed, punched, knocked down, bitten, and

sprayed with bodily fluids. Id. ¶ 7. The government’s interest

in ensuring safe and secure military operations clearly

overcomes any qualified First Amendment right of access.

Rear Admiral Richard W. Butler identified a related danger:

“If video recordings of forced cell extractions, or portions

19

thereof, must be released to the public, detainees would surely

become aware of this, and some likely would respond by

refusing to comply with requests of the guard force in the hope

that such resistance would result in forced cell extractions that

would be recorded by video and released to the public, thus

providing terrorist elements with propaganda to fuel their

continued global hostilities against the United States.”

Declaration of Rear Admiral Richard W. Butler, ¶ 18.

At the time of his declaration, Rear Admiral Butler was the

Commander of the Joint Task Force-Guantanamo. He made his

declaration on personal knowledge. The district court, calling

his statement “speculative,” thought it knew better. Dhiab, 70

F. Supp. 3d at 500. According to the court, detainees will not

act as predicted because the court’s decision will not give all

“detainees the unilateral right to publicize recordings.” Id. at

501. This misses an important point. The concern was that

other detainees will believe courts may disclose recordings of

their behavior. Needless to say, the district court had no day-to-

day experience with the people being detained at Guantanamo

and had no special insight into their mindset. Rear Admiral

Butler did.

The government also explained in detail the risk that

extremists would use the recordings to incite violence against

American troops abroad and as propaganda to recruit fighters.

The recordings are “particularly subject to use” because they

depict “a forcible interaction between . . . personnel and the

detainees.” Declaration of Rear Admiral Sinclair M. Harris,

¶ 12. Images are more provocative than written or verbal

descriptions. Extremists have used Guantanamo Bay imagery

in their propaganda and in carrying out attacks on Americans.

Id. ¶¶ 8,10. For example, the Islamic State beheaded American

journalists wearing orange jumpsuits commonly associated with

Guantanamo Bay detainees. Id. ¶ 8. In his forced final

20

statement before his execution, Steven Sotloff, one of the

journalists, was forced to mention the continued operation of

Guantanamo as a reason why he was about to be murdered. Id.

The district court disregarded this evidence as legally

irrelevant under the “heckler’s veto” line of cases, which

prevents the government from censoring speech because the

speech may provoke violence or offend others. See Dhiab, 70

F. Supp. 3d at 500 (citing Forsyth Cty. v. Nationalist Movement,

505 U.S. 123, 134-35 (1992)). Those cases are irrelevant. The

government cited this substantial probability of harm to national

security in order to overcome a qualified First Amendment right

of access, not as a justification for suppressing speech. Risks of

violent attacks by third parties against a cooperating defendant

can justify sealing a plea agreement. See Washington Post v.

Robinson, 935 F.2d 282, 291-92 (D.C. Cir. 1991). And the

government unquestionably can classify documents based on the

risk our enemies will use them to incite violence. Judicial

Watch, Inc. v. United States Department of Defense, 715 F.3d

937, 943 (D.C. Cir. 2013), so held.

It bears repeating that the government “has a compelling

interest in protecting . . . the secrecy of information important to

our national security . . ..” McGehee, 718 F.2d at 1143 (quoting

Snepp v. United States, 444 U.S. at 509 n.3 (per curiam)

(emphasis and alteration in original)). See also C.I.A. v. Sims,

471 U.S. 159, 175 (1985); United States v. Yunis, 867 F.2d 617,

623 (D.C. Cir. 1989). The district court did not disagree with

the “SECRET” classification of these recordings, and neither did

the intervenors. By definition, “the unauthorized disclosure of

[the recordings] reasonably could be expected to cause serious

damage to the national security.” Executive Order No. 13,526

§ 1.2 (a)(2). The district court had no basis for ruling that

publicly releasing the recordings could not be expected to cause

such harm.

21

The district court did not reach the intervenors’ common-

law claim because it ruled in their favor on the basis of the First

Amendment. Dhiab, 70 F. Supp. 3d at 492 n.2; see Nixon v.

Warner Commc’ns, Inc., 435 U.S. at 598-99. The law of this

circuit is that the need to “guard against risks to national security

interests” overcomes a common-law claim for access. United

States v. Hubbard, 650 F.2d 293, 315-16 (D.C. Cir. 1980).

Because keeping the recordings sealed is narrowly tailored to

protect the government’s compelling interest in guarding

national security, intervenors cannot prevail on their common-

law claim. See Wash. Post, 935 F.2d at 288 n.7; see also

McVeigh, 119 F.3d at 812.

Because the recordings will remain sealed, the intervenors’

cross-appeal about the extent of the redactions is dismissed as

moot.

Reversed.

ROGERS, Circuit Judge, concurring in part and concurring

in the judgment. Like Judge Williams, I would apply the

experience and logic analysis of Press-Enterprise Co. v.

Superior Court, 478 U.S. 1, 8–9 (1986) (“Press-Enterprise II”),

and so my conclusion about when the government’s interest in

protecting information classified as SECRET will outweigh the

public’s First Amendment interest is more tentative than Judge

Randolph’s. At the same time, I tend to be less tentative than

my colleagues about the nature of the historical background and

the level of generality properly used in the analysis. Still, these

analytical differences aside, the court is in agreement that the

district court’s order making the redacted SECRET videotapes

public must be reversed. See Op. at 1 (Williams, J.); Op. at

16–21 (Randolph, J.). I offer a few observations on my

approach.

Although neither the Supreme Court nor this court has

applied the qualified First Amendment right of access to judicial

civil proceedings, in Press-Enterprise II, the Supreme Court

explained that the access right extends to any judicial

proceeding where there is a “tradition of accessibility” and

“public access plays a significant positive role in the functioning

of the particular process in question.” 478 U.S. at 8. The First

Amendment guarantees the “rights to speak and to publish

concerning what takes place at a trial.” Richmond Newspapers,

Inc. v. Virginia, 448 U.S. 555, 576–77 (1980). The then-Chief

Justice stated that “[w]hether the public has a right to attend

trials of civil cases is a question not raised by this case, but we

note that historically both civil and criminal trials have been

presumptively open.” Id. at 580 n.17. By its terms, the

experience and logic test does not limit the right of access to

criminal proceedings. Every circuit to consider the issue has

concluded that the qualified First Amendment right of public

access applies to civil as well as criminal proceedings. See

Courthouse News Serv. v. Planet, 750 F.3d 776, 786 (9th Cir.

2014); N.Y. Civil Liberties Union v. N.Y.C. Transit Auth., 684

2

F.3d 286, 298 (2d Cir. 2012); Rushford v. New Yorker

Magazine, Inc., 846 F.2d 249, 253–54 (4th Cir. 1988); Publicker

Indus., Inc. v. Cohen, 733 F.2d 1059, 1070 (3d Cir. 1984); In re

Cont’l Ill. Sec. Litig., 732 F.2d 1302, 1308 (7th Cir. 1984); In re

Iowa Freedom of Info. Council, 724 F.2d 658, 661 (8th Cir.

1983); Newman v. Graddick, 696 F.2d 796, 801 (11th Cir.

1983). Speaking for the Second Circuit, Judge Calabresi

explained that the “recognition of the right to attend civil trials

derives from the fact that the First Amendment, unlike the Sixth,

does not distinguish between criminal and civil proceedings, nor

does its distinguish among branches of government.” N.Y. Civil

Liberties Union, 684 F.3d at 298. “[O]nce unmoored from the

Sixth Amendment, there is no principle that limits the First

Amendment right of access” to criminal proceedings. Id.

(internal quotation marks omitted). As Judge Williams

observes, “the fact that habeas proceedings are formally civil is

no obstacle to use of the experience-and-logic framework.” Op.

at 4 (Williams, J.).

Efforts to sow doubt about this conclusion and the

conclusion of our sister circuits are unpersuasive because they

fail, in part, to acknowledge an important distinction between a

criminal defendant’s rights and the public’s role in advancing

observance of those rights. Judge Randolph notes that in United

States v. Reynolds, 345 U.S. 1 (1953), the Supreme Court

distinguished between criminal and civil proceedings in

addressing the government’s invocation of the military secrets

privilege to block the release of information that the government

claimed involved national security. See Op. at 9–10 (Randolph,

J.). But the Court did so in the context of highlighting that the

government could not rely on the state secrets doctrine when it

had prosecuted the defendant: “it is unconscionable to allow [the

government] to undertake prosecution and then invoke its

governmental privileges to deprive the accused of anything

which might be material to his defense.” Reynolds, 345 U.S. at

3

12. The distinction drawn by the Court arose in the context of

protecting the rights of the accused, not to indicate that the

public’s right of access is more robust in criminal than in civil

proceedings. Because the First Amendment right of access is

enjoyed by the public, whether the “government is . . . the

moving party” has little bearing on the First Amendment

analysis. Op. at 10 (Randolph, J.) (quoting Reynolds, 345 U.S.

at 12). The same is true of the Classified Information

Procedures Act (“CIPA”), which Congress passed to “limit the

practice of defendants threatening to disclose classified

information in order to force the government to dismiss the

charges.” Id. at 10 n.9. Because the constitutional right of

access belongs to third parties, laws governing the relationship

between litigating parties are of little consequence to the

application of Press-Enterprise II here.

Furthermore, in raising questions about the nature of the

historical tradition and its level of generality, a word of caution

is in order. See Op. at 5–7 (Williams, J.); Op. at 10–12

(Randolph, J.). At least as early as 1866, habeas applications

were filed in open court in the United States. Ex Parte Milligan,

71 U.S. 2, 5 (1866)). It is true that in holding the right to attend

criminal trials is implicit in the guarantees of the First

Amendment, the Supreme Court in Richmond Newspapers

traced “an unbroken, uncontradicted history” of public

accessibility, 448 U.S. at 573, beginning prior to the Norman

Conquest. But it is also true that the Court subsequently

concluded in Press-Enterprise II that the practice of state and

federal courts in the 19th century onward sufficed to establish a

history of access to preliminary hearings in criminal cases. 478

U.S. at 10 (emphasis added); see id. at 22 (Stevens, J.,

dissenting) (“[I]t is uncontroverted that a common-law right of

access did not inhere in preliminary proceedings at the time the

First Amendment was adopted.”). And in Washington Post v.

Robinson, 935 F.2d 282, 288 (D.C. Cir. 1991), this court, joining

4

three other circuits, held that a qualified right of access applies

beyond the criminal trial itself to executed plea agreements and

related documents. Because “plea bargaining was probably

nonexistent before 1800” in the United States, Mark H. Haller,

Plea Bargaining: The Nineteenth Century Context, 13 LAW &

SOC’Y REV. 273, 273 (1979); see also George Fisher, Plea

Bargaining’s Triumph, 109 YALE L.J. 857, 1017–24 (2000), this

court apparently relied on post-ratification practices in

concluding that “plea agreements have traditionally been open

to the public.” Robinson, 935 F.2d at 288. Yet my colleague

suggests that pre-ratification practices should be the focus of our

inquiry. See Op. at 10–12 (Randolph, J.).

The Supreme Court has not required there be a history of

absolute accessibility to satisfy the “experience” prong; a “near

uniform practice of state and federal courts” suffices. Press-

Enterprise II, 478 U.S. at 10 (emphasis added); see id. at 10 n.3.

There can be “gaps.” Op. at 7 (Williams, J.). In Press-

Enterprise II, the Court acknowledged an historical “tradition of

accessibility” for state and federal preliminary hearings even

though several states had “no historical counterpart,” 478 U.S.

10 & n.3, and had only recently recognized a right of public

access to preliminary hearings in view of their importance to the

criminal trial. Id. at 10 n.3; see also id. at 24–25 (Stevens, J.,

dissenting). The Court contrasted grand jury proceedings, which

have “traditionally been closed to the public,” id. at 10, because

their “proper functioning . . . depends upon the[ir] secrecy,” id.

at 9. Nonetheless, in relying on English history from the 16th to

18th centuries, my colleagues appear unpersuaded, surprisingly,

that the overwhelming practice of open habeas corpus

proceedings — at least 80% — establishes a sufficient tradition

of accessibility. See Op. at 12 (Randolph, J.); Op. at 7

(Williams, J.). The author on whom they rely suggests that the

in-chambers habeas practice may be explained by the limited

terms of court, the need for prompt issuance of writs to ensure

5

compliance with court orders, and judicial vacations and travel

— all when modern forms of communication were non-existent.

PAUL D. HALLIDAY, HABEAS CORPUS: FROM ENGLAND TO

EMPIRE 53–58 (2010). In other words, there was a well-settled

expectation that habeas proceedings would be open to the public

when the courts were in session.

Neither does this case present the occasion to define the

meaning of “judicial record” in a manner contrary to the

understanding of the district court and the parties, and how this

court has viewed filings in habeas cases. See Op. at 2–3

(Williams, J.); see also Parhat v. Gates, 532 F.3d 834, 836 (D.C.

Cir. 2008). No issue is presented to this court concerning the

“judicial” nature of the SECRET videos filed by Dhiab’s

counsel in the district court. Even so, on its own terms, the

suggested definition would appear to be overly restrictive,

excluding from the category of “judicial records,” for example,

certain pleadings by a party. See FED. R. CIV. P. 7, 79.

As for the logic part of the Press Enterprise II test, it is true

that there is no instruction manual as such for lower courts on

“choosing the level of generality at which to assess the

‘proceeding.’” Op. at 5 (Williams, J.). Supreme Court

precedent is itself a guide, however, and it indicates that a high

level of generality can be appropriate. In Press-Enterprise II,

410 U.S. at 10, the Court applied the experience and logic test

to preliminary hearings in California, concluding that because

“preliminary hearings conducted before neutral and detached

magistrates” in other state and federal courts have traditionally

been open, there was a tradition of access that applied to the

California proceeding. The Court made no mention of the

“substantial variations in the structure of the preliminary hearing

as it is conducted throughout the country.” Jesse H. Choper,

Consequences of Supreme Court Decisions Upholding

Individual Constitutional Rights, 83 MICH. L. REV. 1, 112

6

(1984) (quoting Y. KAMISAR, W. LAFAVE & J. ISRAEL, MODERN

CRIMINAL PROCEDURE: CASES-COMMENTS-QUESTIONS 963 (5th

ed. 1980)); see also Note, The Function of the Preliminary

Hearing in Federal Pretrial Procedure, 83 YALE L.J. 771,

773–74 (1974). For example, the New Hampshire Supreme

Court, in ruling that hearsay was admissible at preliminary

hearings, noted that the highest court in another state had taken

a “contrary position,” and that “[s]uch decisions are based upon

different statutes, [and] a different view of the purpose of a

probable cause hearing.” State v. St. Arnault, 317 A.2d 789, 791

(N.H. 1974) (citing Myers v. Commonwealth, 298 N.E.2d 819

(Mass. 1973)). By lumping together the various state and

federal preliminary hearings when applying the experience and

logic test, the Supreme Court suggests that a lower court may

properly apply the test at a fairly high level of generality. The

writ of habeas corpus has never been a “static, narrow,

formalistic remedy,” Jones v. Cunningham, 371 U.S. 236, 243

(1963), but one whose scope and application has “changed

depending upon the circumstances,” Boumediene v. Bush, 553

U.S. 723, 779 (2008). Taken together, that would mean viewing

Dhiab’s habeas proceeding as falling within the tradition of open

habeas proceedings generally, rather than singling out habeas

petitions filed by Guantanamo detainees for a separate test.

The qualified First Amendment right of access fits well with

the privilege of habeas corpus, which was originally “one of the

few safeguards of liberty specified in [the] Constitution.” Id. at

739. Because criminal trials and habeas proceedings are

designed to protect against abuses of Executive power and guard

individual liberty, why would the First Amendment right of

access apply differently in the two proceedings? “Courts and

commentators have long recognized the centrality of openness

to adjudicatory proceedings: ‘Without publicity, all other checks

are insufficient: in comparison of publicity, all other checks are

of small account.’” N.Y. Civil Liberties Union, 684 F.3d at 296

7

(quoting In re Oliver, 33 U.S. 257, 271 (1948) (quoting 1

Jeremy Bentham, RATIONALE OF JUDICIAL EVIDENCE 524

(1827))). The qualified right of public access plays a significant

positive role in criminal proceedings by ensuring that “standards

of fairness are being observed.” Press-Enterprise Co. v.

Superior Court, 464 U.S. 501, 508 (1984); see also Richmond

Newspapers, 448 U.S. at 569. In habeas proceedings, the

absence of a jury, “long recognized as an inestimable safeguard

against the corrupt or overzealous prosecutor and against the

complaint, biased or eccentric judge[,] makes the importance of

public access . . . significant.” Press-Enterprise II, 478 U.S. at

12–13 (quoting Duncan v. Louisiana, 391 U.S. 145, 156 (1968)).

Also, “[t]o the extent the First Amendment embraces a right of

access to criminal trials, it is to ensure that th[e] constitutionally

protected discussion of governmental affairs is an informed

one.” Globe Newspaper Co. v. Superior Court, 457 U.S. 596,

604–05 (1982) (internal quotation marks omitted); see also

Richmond Newspapers, 448 U.S. at 576–77. Because the writ

of habeas corpus is an important part of our Constitution and a

“vital instrument for the protection of individual liberty,”

Boumediene, 553 U.S. at 743, the public’s qualified right to

informed discussion about its government would apply no less

in these proceedings.

Nor is there reason to conclude that when the Supreme

Court articulated the experience and logic test, “it could not

possibly have had in mind classified national security

information.” Op. at 9 (Randolph, J.). The Court’s test protects

against threats to our nation’s security by prohibiting disclosure

when it will cause a “substantial probability” of harm to an

“overriding interest.” Press-Enterprise II, 478 U.S. at 7, 14.

The right to access judicial proceedings “give[s] meaning to

[the] explicit guarantees” of freedom of speech and press,

Richmond Newspapers, 448 U.S. at 576, and the Court is well

aware that First Amendment rights will often clash with national

8

security concerns, see, e.g., Dennis et al. v. United States, 341

U.S. 494 (1951). Yet the Court crafted a test where the

threshold First Amendment question is whether “the particular

process in question” passes the experience and logic test, Press-

Enterprise II, 478 U.S. at 8; see also Op. at 1, 4 (Williams, J.),

not whether the records submitted in that proceeding contain

classified information. Because the test accounts for the

protection of national security information, the presence of such

information in a judicial proceeding does not crowd out the

decades-old and flexible approach set forth in Press-Enterprise

II.

WILLIAMS, Senior Circuit Judge, concurring in part and

concurring in the judgment: I join Parts I and III of Judge

Randolph’s opinion. I write separately to explain why I view

the First Amendment analysis through a different lens than the

one he applies in Part II of his opinion. While my approach

leads to the same result as his, my conclusions are more

tentative and my doubts more pronounced.

***

The Supreme Court has sketched a two-stage process for

resolving whether the First Amendment affords the public

access to a particular judicial record or proceeding. First the

court must determine whether a “qualified First Amendment

right of public access” exists. Press-Enterprise Co. v.

Superior Court, 478 U.S. 1, 9 (1986). If so, then the potential

qualification comes into play, and the record or proceeding

may be closed only if “closure is essential to preserve higher

values and is narrowly tailored to serve that interest.” Id. at

13-14. Assuming the existence of such a qualified right in

this case, all members of the panel agree that the compelling

national security considerations discussed in Part III of Judge

Randolph’s opinion render that right unavailable here.

But on the question of how to determine if that qualified

right actually exists, we see things somewhat differently. In

analyzing that issue, the Supreme Court has identified two

requirements that it calls the “tests of experience and logic.”

Id. at 9; see also In re Reporters Committee for Freedom of

the Press, 773 F.2d 1325, 1331-32 (D.C. Cir. 1985) (Scalia,

J.). The “experience” inquiry looks to “whether the

proceeding has historically been open.” Reporters Committee,

773 F.2d at 1331. And the somewhat oddly-labeled “logic”

inquiry asks “whether the right of access plays an essential

role in the proper functioning of the judicial process and the

government as a whole.” Id. at 1332. Both tests must be

2

satisfied before we can conclude that the First Amendment

provides a qualified right of access. Id.

Before trying to work through those questions, I note that

the government doesn’t contest the premise that the

documents here are judicial records. Accordingly, I assume

that they are—but it is by no means obvious. Merely filing a

document with the district court isn’t enough to transform it

into a judicial document. See, e.g., SEC v. AIG, 712 F.3d 1,

3-4 (D.C. Cir. 2013). Absent a judicial decision, the

documents filling district court dockets are no more than

litigants’ requests for action (or inaction). United States v. El-

Sayegh, 131 F.3d 158, 162 (D.C. Cir. 1997). To become

judicial records, the files must play some role in the

adjudicatory process—i.e., in the judicial decision at hand.

See AIG, 712 F.3d at 3 (quoting El-Sayegh, 131 F.3d at 163).

Justice Holmes long ago articulated the basic reason in a

somewhat different context:

It is desirable that the trial of causes should take

place under the public eye, not because the

controversies of one citizen with another are of

public concern, but because it is of the highest

moment that those who administer justice should

always act under the sense of public responsibility,

and that every citizen should be able to satisfy

himself with his own eyes as to the mode in which a

public duty is performed.

Cowley v. Pulsifer, 137 Mass. 392, 394 (1884), quoted in

Reporters Committee, 773 F.2d at 1335.

Here it is unclear whether the district court “made [any]

decisions about the[] [disputed recordings] or that otherwise

relied on them.” AIG, 712 F.3d at 4. What we do know is

that Dhiab’s attorneys attached three (of the more than thirty)

3

videos to a brief seeking a temporary restraining order and

that the district court denied that TRO after a classified

hearing. See Notice of Filing, Dhiab v. Obama, No. 05-1457

(GK) (June 14, 2014), ECF. No. 252; Order, Dhiab v. Obama,

No. 05-1457 (GK) (June 16, 2014), ECF No. 254. But we

don’t know if those videos (or others) were played at the

hearing or if they might have been understood as a basis for

the district court’s decision. And by the time the district court

ruled on Dhiab’s request for a preliminary injunction, the

government had stopped performing forced cell extractions on

him and so the district court offered no opinion on their

legality. See Dhiab v. Obama, 74 F. Supp. 3d 16, 21 (D.D.C.

2014). Since videos that show only his extraction were thus

irrelevant to the district court’s adjudication of the remaining

challenges, such recordings (if any) could not be judicial

records. See AIG, 712 F.3d at 3–4; El-Sayegh, 131 F.3d at

163. Videos that include footage of the force feedings pose a

closer question, but the answer is hardly clear. Nowhere in

the district court’s injunction opinion does it rely on the

videos expressly. Indeed, the only mention of those

recordings is in a quick recitation of the procedural history.

74 F. Supp. 3d at 20. While that opinion refers to certain

government exhibits in discussing whether the feeding process

is painful, it identifies those exhibits only by number, so we

don’t know whether they include the disputed videos. See id.

at 25-26. Again, as the government didn’t raise the argument,

I proceed on the assumption that all of the disputed videos are

judicial records and proceed to the task of divining the

possible existence of a qualified First Amendment right of

access.

***

A preliminary concern is posed by our declaration several

years ago that the “experience and logic test . . . has been

limited to judicial proceedings that are part of the criminal

4

trial process” and that “[n]either the Supreme Court nor this

court has applied” it outside that context. Center for Nat’l

Security Studies v. DOJ, 331 F.3d 918, 935 (D.C. Cir. 2003)

(emphasis added). But Reporters Committee proves

otherwise; there we applied the test to documents used in civil

summary judgment proceedings. It’s true that we’ve never

found a qualified First Amendment right outside the criminal

context, but we’ve never categorically ruled it out either (and

many other circuits have concluded that such a right exists in

civil and even administrative matters, e.g., N.Y. Civil Liberties

Union v. N.Y.C. Trans. Auth., 684 F.3d 286, 297-98 (2d Cir.

2011)). And in Center for National Security Studies we noted

that the records sought consisted of “investigatory

information” not “information relating to a government

adjudicative process,” and, anticipating El-Sayegh, we saw

that fact as precluding the application of the First Amendment

access right. 331 F.3d at 936. In short, then, the fact that

habeas proceedings are formally civil is no obstacle to use of

the experience-and-logic framework.

That issue cleared aside, I next turn to the level of

generality at which we should consider the “experience”

bearing upon the proceedings or records in question. The

Court has appeared to focus on the “particular proceeding in

question,” Press-Enterprise, 478 U.S. at 9, but without

explaining whether we look to broad or narrow categories and

without precluding focus on types of documents rather than

proceedings. If proceedings are the subject of analysis, the

likely categories here may range among civil actions

generally, habeas actions, habeas actions relating to conditions

of confinement, and finally habeas actions related to

Guantanamo.

Despite the Supreme Court’s apparent interest in the

“proceedings,” courts have often, where documents were at

issue, turned directly to the documents in dispute, and applied

5

the “experience and logic” ideas to them. See, e.g., United

States v. Erie County, 763 F.3d 235, 241 (2d Cir. 2014); In re

Boston Herald, Inc., 321 F.3d 174, 182 (1st Cir. 2003); United

States v. Corbitt, 879 F.2d 224, 229 (7th Cir. 1989). And in

Reporters Committee, we spoke initially of “proceedings,” but

then slipped seamlessly to addressing the type of documents

sought. 773 F.2d at 1330-41.

One is tempted to dismiss all of this as immaterial, on the

theory that we have a case of “pay me now or pay me later.”

If disclosure risks inflicting serious harm, it will emerge either

in the assessment of experience and logic (here, for example,

by focusing on Guantanamo habeas cases, dominated as they

are by classified information), or as a trigger of the right’s

qualification (because the classified character of the

documents meets the government’s burden of showing a

compelling need for secrecy).

Not so fast. Intervenors point to Globe Newspaper Co. v.

Superior Court, 457 U.S. 596 (1982), where the Court applied

its experience-and-logic tests to criminal trials generally,

rejecting the state’s effort to make the classification at the

level of the testimony in question—that of a minor child in a

sexual abuse case. Id. at 605 n.13. The Court made no claim

that experience and logic dictated openness for that segment

of the case. But, turning to the government’s burden, it

rejected the claim that the interests for excluding the public

were compelling. Thus the parties claiming access had a

fairly easy ride to showing experience and logic, and the

government faced an uphill battle on its justification for

privacy. Resolving the level of generality affected not only

who bore the burden of persuasion but also the severity of that

burden.

Apart from choosing the level of generality at which to

assess the “proceeding” (plus the choice between proceeding

6

and documents), the “experience” test requires a decision on

what “history” is relevant. Yet we have no more guidance

here. Reporters Committee tells us that a “historical tradition

of at least some duration is obviously necessary,” 773 F.2d at

1332, but doesn’t tell us how long is long enough. As Judge

Randolph notes, pre-ratification history is surely relevant,

Randolph Op., Part II, at 11 n.12, but courts have plainly not

seen the relevant history as limited to proceedings before

1791. Press-Enterprise itself relied exclusively on history

that post-dated the First Amendment to some extent or

another. Compare 478 U.S. at 10-12 & n.3 (considering 1807

treason trial of Aaron Burr and modern cases showing that

most states required preliminary hearings be held in open

court), with id. at 22-25 (Stevens, J., dissenting) (arguing that

the inquiry should look to history “at the time the First

Amendment was adopted,” not those “recent common-law

developments”). And unlike the plurality opinion in

Richmond Newspapers, Inc. v. Virginia, which traced the

history of trial access back to “the days before the Norman

Conquest,” 448 U.S. 555, 565 (1980), our decision in

Reporters Committee relied heavily on late-nineteenth and

early-twentieth century case law, 773 F.2d at 1332-36; see

also Detroit Free Press v. Ashcroft, 303 F.3d 681, 700 (6th

Cir. 2002) (collecting cases from the Third, Sixth, and Ninth

Circuits relying on more recent history). Admittedly none of

those decisions explicitly grappled with this question, but they

at least raise doubt about stopping the historical clock at 1791.

Here the parties have neither provided a comprehensive

history or sought to explain why only certain history is

relevant. The government’s historical analysis amounts to

nothing more than a few citations to relatively recent cases

and a brief discussion of the Classified Procedures

Information Act of 1980. See App. Br. at 48-49. Intervenors

were similarly cursory, citing a handful of mostly recent cases

before jumping into their competing interpretation of CIPA.

7

See Intervenors’ Br. at 27-28, 31-35 & nn.4, 7. But neither

side tells us why those cases or CIPA might be relevant to the

historical tradition. And—other than intervenors’ throwaway

citation to Ex Parte Milligan, 71 U.S. 2 (1886)—neither gives

a hint of what the tradition of access was like pre-1970, much

less pre-1791. Judge Randolph’s opinion identifies evidence

that a substantial portion (20%) of early English habeas cases

were heard and decided privately. See Randolph Op., Part II,

at 12 (citing PAUL D. HALLIDAY, HABEAS CORPUS: FROM

ENGLAND TO EMPIRE 54, 56-57 (2010)). And that evidence

appears persuasive. Besides, focus on such evidence has the

advantage of being consistent with the idea that those who

adopted the First Amendment meant to embody a pre-existing

right of access to judicial proceedings and records. See

Richmond Newspapers, 448 U.S. at 575-76. If that were the

historical inquiry, all we’d need to know would be whether

the access sought was available before 1791. See Press-

Enterprise, 478 U.S. at 22 (Stevens, J., dissenting). But the

cases don’t demand any such focus or read the right so

narrowly. Nor do they give any guidance on how to choose

among potentially relevant time spans; we’re left simply to

guess at what history might be relevant.

We’re similarly in the dark in terms of how consistent a

tradition of openness must be within a given time span. The

Court has told us that an “unbroken, uncontradicted history”

will do the trick, Richmond Newspapers, 448 U.S. at 573, but

has also found a qualified First Amendment right when there’s

a “near uniform [historical] practice,” Press-Enterprise, 478

U.S. at 10 (emphasis added). But while its use of the word

“near” tells us that the tradition can have gaps, we don’t know

how close to uniform the tradition needs to be. Perhaps, as

Judge Rogers suggests, it’s enough that the public historically

had access at least 80% of the time. See Rogers Op. at 4.

(Though most teachers would likely doubt the wisdom of

calling a score of 80% “overwhelming.”) On the other hand,

8

as Judge Randolph argues, it’s certainly hard to say that

there’s “an unbroken, uncontradicted history” of public access

when 20% of the cases were heard privately. See Randolph

Op., Part II, at 12. Both positions seem reasonably grounded

in the Court’s precedents, but obviously both can’t be right.

Luckily, however, nothing in this case requires us to guess at

how to resolve that troublesome issue, so it is enough for us to

note the questions left unanswered by the Court’s precedents.

Putting those doubts aside for a moment, at least one

thing is clear: If the experience-and-logic framework is to be

applied to Guantanamo habeas cases, at least the “logic”

part—whether public access would play a significantly

positive role in these proceedings—seems relatively easy.

Boumediene v. Bush (and the standing protective orders for

the Guantanamo cases) recognize that detainees litigating

these cases have a practical need for classified information to

contest the legality of their detention, see 553 U.S. 723, 784-

86 (2008), a premise that in light of Aamer v. Obama, 742

F.3d 1023 (D.C. Cir. 2014), extends to conditions of

detention. But as the intervenors themselves recognize, the

government could “oppose the disclosure of classified

material as privileged under the state secrets doctrine,”

Intervenors’ Br. at 35, and would have every reason to do so

in light of its national-security concerns. Judge Randolph’s

discussion of United States v. Reynolds, 345 U.S. 1 (1953),

and Totten v. United States, 92 U.S. 105 (1875), drives home

the point that the government could short-circuit the entire

Guantanamo habeas process by invoking that privilege and

thereby depriving detainees (and also the courts) of potentially

critical information. See Randolph Op., Part II, at 9-10, 14-

15. (Indeed, since the criteria to classify information as

“SECRET” appear to be more stringent than the privilege test

we articulated in Halkin v. Helms, 598 F.2d 1, 9 (D.C. Cir.

1978), it seems likely that the government could withhold any

information properly classified at that level.) It therefore

9

seems safe to say that disclosure would not play a positive

role in those proceedings; to the contrary, it might

substantially hollow them out, perhaps to the point of raising a

Suspension Clause question.

***

In short, then, under current understandings, choices as to

level of generality for the relevant proceedings (and between

proceedings and documents), and the scope of the relevant

historical inquiry, can easily be decisive, both in shifting the

burden of persuasive and in its rigor. Yet we have little

guidance from the Supreme Court, or indeed any other, as to

how to make those choices.

In this case, however, we can avoid these questions.

Even if we are to apply a higher level of generality (perhaps

habeas generally or even just civil matters) or to look to

relatively recent history, and even if doing so would show

experience and logic to lie on the intervenors’ side, it is of no

consequence—in view of our conclusion that the security

interests invoked by the government are compelling (and no

lesser remedy is available than preserving them from public

access). I therefore join with my colleagues in reversing the

district court.