United States v. Rosen

487 F. Supp. 2d 703, 2007 WL 4142778, 2007 U.S. Dist. LEXIS 29469

Opinion

MEMORANDUM OPINION

ELLIS, District Judge.

This Espionage Act (18 U.S.C. § 793) prosecution involves a substantial amount of classified information the government contends is information relating to the national defense (“NDI”) 1 and for this reason the government seeks to avoid public disclosure of this material in the course of the trial. To this end, the government, by motion pursuant to § 6 of the Classified Information Procedures Act (“CIPA”), 2 has proposed utilizing a procedure at trial whereby substantial quantities of classified information would be disclosed to the Court, the jury, and counsel, but withheld from the public. This novel proposal, if allowed, would effectively close a substantial portion of the trial. Accordingly, defendants challenge this proposed procedure on the grounds that it is neither authorized by CIPA nor constitutionally permissible.

I.

Both defendants are charged with conspiracy to communicate NDI to persons not authorized to receive it, in violation of 18 U.S.C. § 793(g),(e). Rosen is additionally charged with aiding and abetting alleged co-conspirator Lawrence Franklin’s unauthorized communication of NDI to persons not authorized to receive it, in violation of 18 U.S.C. §§ 793(d) and 2. The superceding indictment generally alleges that over a course of several years, defendants cultivated various sources of information within the United States government, obtained NDI from those sources, and then disseminated that NDI to others not authorized to receive it, including coworkers, journalists, and foreign government officials. For a more complete recitation of the facts alleged in the superced- ing indictment, as well as a more precise delineation of the government’s burden of proof, see United States v. Rosen, 445 F.Supp.2d. 602 (E.D.Ya.2006) (Memorandum Opinion denying motion to dismiss) and see also United States v. Rosen, 447 F.Supp.2d 538 (E.D.Va.2006) (Order clarifying Memorandum Opinion).

II.

Because this prosecution involves classified materials, CIPA governs the use of that material at the trial and pretrial stages. The purpose of CIPA is “to harmonize a defendant’s right to obtain and present exculpatory material ... and the government’s right to protect classified material.... ” United States v. Pappas, 94 F.3d 795, 799 (2nd Cir.1996) (citing United States v. Wilson, 571 F.Supp. 1422, 1426 (S.D.N.Y.1983)); see also Armstrong v. Executive Office of the President, 830 F.Supp. 19, 23 (D.D.C.1993) (same). A brief summary of the CIPA proceedings to date provides useful context for resolution of the issue presented.

At the outset, following a pretrial conference held pursuant to CIPA § 2, a protective order issued pursuant to CIPA § 3 governing the disclosure of classified information to defense counsel and defendants. 3 United States v. Rosen, No. 1:05cr225 (E.D.Va. Sept. 19, 2005) (Protective Orders as to each defendant). Following entry of these orders, discovery proceeded, as authorized by the Federal Rules of Criminal Procedure and CIPA § 4. See also United States v. Rosen, No. 1:05cr225 (E.D.Va. Nov. 8, 2005) (ex parte, in camera, and under seal protective order pursuant to CIPA § 4 restricting discovery of some matters). Then, pursuant to CIPA § 5, defendants gave notice of their intention to introduce at trial a substantial amount of material which they reasonably expected might disclose or cause disclosure of classified material. This notice, it turned out, listed a large volume of material, in part because the indictment alleges a number of distinct elements of NDI covering a range of distinct subjects, and in part because defense counsel cannot know whether the government would assert that certain materials defendants proposed for use at trial were classified or would cause disclosure of classified information.

The next step in the CIPA process was a hearing requested by the government pursuant to CIPA § 6(a), at which the Court would begin the process of making “determinations concerning the use, relevance or admissibility of classified information that would otherwise be made during the trial.... ” In the course of this hearing, some of the potentially-classified material defendants sought to use at trial was ruled irrelevant and inadmissible pursuant to Rules 402 and 403, Fed.R.Evid., while a still quite substantial volume of material considered to be classified or to cause disclosure of classified material was ruled relevant and admissible at trial. United States v. Rosen, No. 1:05cr225 (E.D.Va. Jan. 17, 2007) (sealed CIPA § 6(a) Order). This hearing and these rulings began, but did not complete, the CIPA § 6 process. A further hearing is now required pursuant to CIPA § 6(c) at which the government will be afforded the opportunity to declassify the material ruled relevant and admissible at trial, or alternatively, to move that in lieu of disclosing the classified material at trial, the government be allowed to substitute for such classified information “a statement admitting relevant facts that the classified information would tend to prove” or “a summary of the specific classified informa- tionThe government has filed such a motion with respect to a substantial volume of the classified information ruled relevant and material. The CIPA § 6(c) hearing will be closed to the public, as the government has so requested and CIPA provides that “any such hearing shall be held in camera at the request of the Attorney General.” Defendants, for their part, contest the substitutions advanced by the government for approximately 38 government documents, 8 public source documents, and 22 recordings. Should the government’s motion for substitutions under CIPA § 6(c) fail, the government may object to disclosure of the classified information at issue, and if the government does so, the information may not be disclosed, but in that event the Court may (i) dismiss the indictment in whole or in part, (ii) find against the United States on any issue as to which the classified information relates, or (iii) preclude a witness’ testimony. See CIPA § 6(e)(l)-(2). 4 It bears emphasis that CIPA does not authorize a trial judge to second-guess the government’s decision to classify information. Nor does it authorize a trial judge to find that information is not damaging to national security if revealed. Instead, the CIPA § 6(c) inquiry focuses solely on whether the proposed substitutions afford the defendant substantially the same ability to make his defense as the specific classified information.

III.

The government has proposed a procedure for handling this material at trial and it is this proposal that is at issue here. The government’s proposal is novel; no published opinion has been found or cited in which the precise procedure proposed here was judicially approved or used. Simply put, the government proposes that while the jury, the Court, and counsel will, for the most part, have access to the unredacted classified material, the public will not. Instead, the public, in the course of the trial, will see and hear only the substitutions that have passed through the CIPA § 6(c) process. In other words, and putting to one side the not insubstantial practical problems inherent in conducting a trial pursuant to the government’s proposed procedure, its use would surely exclude the public from substantial and critical parts of the trial. This result is evident from a more detailed description of the proposed procedure.

The government’s proposal is, in effect, a variant and a substantial expansion of the so-called “silent witness rule,” a rule that has been used and judicially approved in certain, but not all circumstances. 5 But unlike in Zettl, the procedure in this case would be used for all classified material in both the government’s case and the defense case. 6 As noted, the effect of using the procedure in this case would be the exclusion of the public from substantial portions of the trial.

The proposed procedure would work as follows: for each classified document discussed at trial, the Court, the witness, counsel, and the jurors would have the unredacted classified document in front of them, either in paper form or via computer screens viewable only by those persons. The public, however, would see only a redacted version. When counsel or a witness wishes to direct the jury to a classified portion of the document, counsel and the witness would refer to the page, paragraph and line numbers, with the Court, opposing counsel, and the jury following along, but members of the public could not follow along because they would be unaware of the specific information referenced by counsel or the witness. And, the witness answering the questions about the document would not be permitted to refer to specific language or information in the document, except by use of certain codes. For example, to rebut the government’s contention that certain material is NDI, defendants will likely wish to call witnesses to compare various public source documents with the alleged NDI. To do so effectively may well require the witness to refer to specific language or contents of both the public source document and the alleged NDI. Anticipating this, the government proposes that the witness would not speak the names of certain specific countries, foreign persons, etc., but would instead use a code (e.g. “Country A,” “Report X,” “Foreign Person Y,” “Foreign Person Z,” etc.) provided also to counsel, the Court, and the jury. Moreover, this code would change with respect to different alleged overt acts, presumably to prevent the public from inferring the meaning of the generic designations or otherwise breaking the code. For example, if a witness discussing a particular alleged disclosure is instructed to refer to Monaco as Country A, a different witness (or even the same witness) discussing another alleged disclosure might use Country B or C to refer to Monaco.

Likewise, when recordings discussing classified information are played, the government proposes that the Court, counsel, witness, and jury listen on special headphones to the entire recording. The public, however, would not hear the full recording; instead, the recorded conversations would be played aloud in the courtroom, but where classified information is discussed in the recording, the public version would revert to static. Also, the public would receive only a redacted transcript of the conversations.

In sum, the novel and distinctive feature of the government’s proposed procedure is that the public is walled off from seeing and hearing everything the jury, the Court, the attorneys and the witnesses see and hear. What the public does not see or hear is the heart of the case, namely the classified material the government claims is the NDI that the defendants allegedly received and distributed without authorization. A further, related novel and distinctive feature of the government’s proposal relates to the jury. Although jurors will see and hear classified material, they, of course, will not have received security clearances for this purpose. Nor has this ever been otherwise in cases involving classified material, for a variety of reasons including the substantial time and effort typically required for such clearances to be completed and the substantial question whether it is constitutionally permissible to exclude jurors because they could not pass a security clearance investigation. The government’s remedy for this anomaly is simply to have the jurors instructed that they cannot disclose to anyone the classified material they will see and hear during the trial. 7

With this description in mind, the analysis now proceeds to address the following questions:

(i) Whether the government’s proposed procedure is explicitly or implicitly authorized by CIPA.

(ii) Whether the government’s proposed procedure violates the right to a public trial, guaranteed to defendants by the Sixth Amendment and to the public by the First Amendment.

IV.

A

The government urges that the use of the silent witness rule, codes, and redacted recordings are “substitutions” authorized by CIPA. Defendants disagree, noting the government’s proposed procedure is nowhere authorized by CIPA, either explicitly or implicitly. Defendants are correct, a conclusion that follows from CIPA’s plain language, but also from the fact that the government’s proposed procedure simply cannot fit within CIPA’s confines even assuming the statute’s plain language would not otherwise preclude it.

Analysis of this statutory argument properly focuses on CIPA § 6(c) and § 8(b), which provisions are set out in full in the margin. 8 As noted, CIPA § 6(c) allows the government to move that in lieu of disclosing classified information at trial, a “substitution” be used, which could take the form of a summary of the information or a presumably sanitized statement of facts the classified statement would tend to prove. Importantly, no substitutions can be used unless there is a judicial finding that the substitution provides defendants with “substantially the same ability to make his defense as would disclosure of the specific classified information.” Consistent with this, CIPA § 8(b) also permits introduction and use at trial of redacted versions of classified materials unless “the whole ought in fairness to be considered.”

Significantly, neither § 6(c) nor § 8(b) explicitly authorize or state that “substitutions” may be made available to the public and the jury on different terms. CIPA is at best silent on this issue. Yet, this silence should not be construed as implicitly authorizing the government’s proposal. To the contrary, it seems clear that CIPA envisions that “substitutions,” if not unfair to defendants, will be used at a public trial in identical form for both the jury and the public. While it is true, as reflected in CIPA’s legislative history, that “Congress expected trial judges to fashion creative solutions in the interests of justice for classified information problems,” 9 there is no evidence that Congress expected this creativity to extend to adopting procedures that effectively close the trial to the public. 10 Indeed, given the strong presumption in the law that trials will be open and that evidence will be fully aired in public, 11 CIPA’s silence about whether “substitutions” and “excisions” can be made available to the public and jury on different terms should be interpreted as a prohibition on doing so. Closing a trial, even partially, is a highly unusual result disfavored by the law. A statute, even one regulating the use of classified information, should not be construed as authorizing a trial closure based on Congress’ mere silence or use of ambiguous language. Rather, because a trial closure implicates important constitutional rights, CIPA should not be read to authorize closure absent a clear and explicit statement by Congress in the statutory language.

In short, CIPA only authorizes the use of substitutions to avoid disclosure of classified information to the public and the jury, provided defendant’s right to present a defense is not impaired; this authority is not the authority to close a trial to the public.

B.

Even assuming, arguendo, that the CIPA provision allowing “substitutions” might be stretched to encompass the government’s proposed procedure, there is no doubt that the procedure would not pass muster under CIPA’s fairness requirements. Where, as here, a central issue in the case is whether the government’s al leged NDI is indeed genuinely NDI, and the proposed procedure would amount to a wholesale use of the silent witness rule to cover all of the alleged NDI, it cannot be said that the procedure affords defendants “substantially the same ability to make [their] defense as would disclosure of the specific classified information.” CIPA § 6(c). A few examples vividly illustrate this point.

As noted, the government in this case has the burden of proving beyond a reasonable doubt that the alleged NDI is indeed NDI. 12 To this end, the government will likely invite witnesses to compare the substance of certain of defendants’ emails, telephone conversations, or faxes with certain alleged NDI to show that defendants had obtained NDI and were disseminating it without authorization. Defendants, of course, may wish to show, and indeed to emphasize, the dissimilarities between the alleged NDI and the information they obtained. Plainly, they would be significantly hobbled in doing so by use of the government’s proposed procedure, inasmuch as the specific information could not be used in open court. The silent comparison of paragraphs or sentences, even where supplemented by codes, would effectively preclude defense counsel from driving home important points to the jury.

Similarly, it is apparent that defendants intend at trial to rebut the government’s claim that certain material is NDI by having witnesses compare the alleged NDI to contemporaneous public domain material. Once again, the proposed procedure would unfairly impact defendants’ ability to establish this defense. In this context, the silent witness rule, applied across the board as the government proposes here, essentially robs defendants of the chance to make vivid and drive home to the jury their view that the alleged NDI is no such thing, as essentially similar material was abundant in the public domain. Importantly in this respect, it is hard to see how defendants could effectively show, via the silent witness rule, that the details of differences between public source material and the alleged NDI are neither minor nor trivial.

The following hypothetical example puts this point in more concrete terms. .Suppose the government calls Lawrence Franklin as a witness, and he testifies that defendants solicited him to provide particular information about American policy towards “Country A.” Suppose the government then calls an expert witness who testifies that the information defendants solicited from Franklin is NDI, i.e., that the information, inter alia, was closely held by the government and would damage national security if disclosed. Defense counsel would surely seek to contest this testimony by asking this expert on cross examination to compare the alleged NDI to information available contemporaneously in the public domain, perhaps a news or journal article, aiming to show that the information alleged to be NDI was neither closely held, nor damaging to national security, In this event, defense counsel would proffer the news article and, pursuant to the silent witness rule, direct the jurors’ and witness’s attention to the information in the article that defendants claim is essentially similar to the alleged NDI the government claims was solicited by defendants. On being asked whether the two sets of information are the same, the expert would presumably deny that the information is the same. Then defense counsel would surely seek to ask follow-up questions to establish that the differences in the two, if any, are trivial. To do so effectively, defense counsel and the witness must have latitude to explore and contest any minute differences in public domain information and the information defendants allegedly solicited from Franklin.

It seems clear that questions attempting to probe the similarity of putatively classified information and public domain information would require questions and answers in open court about the details and substance of the relevant public source documents and the alleged NDI. Yet, such questions and answers are precisely what the proposed procedure prohibits. At most, coded references would have to be used. But if a difference between the alleged NDI and public domain information is particularly elusive or difficult to describe, it would be impossible for defense counsel to demonstrate effectively via the silent witness rule that any differences between the alleged NDI and the public domain information do not support NDI status for the alleged NDI. Indeed, the code and unspoken comparisons required by the proposed procedure might well gloss over precisely the similarities and insignificance of differences in government-held information and public domain information that may be crucial to defendants’ position. In short, the use of codes would render virtually impossible an effective line of cross-examination vital to the defense.

Moreover, quite apart from compounding the difficulties in comparing public domain information and putative NDI, the proposed procedure — the across the board application of the silent witness rule to all putatively classified information — would make it virtually impossible for defendants to conduct effective cross-examination on why the putative NDI would be damaging to national security if disclosed. It is simply not plausible to suggest that defense counsel could effectively cross-examine witnesses about whether information is potentially damaging to national security by use of coded euphemisms like “the redacted policy assessment” or “the fact on page 10 about Country A’s activities in Country B.” 13 Defense counsel’s closing jury arguments would be similarly limited and hence adversely affected.

Yet another fatal defect in the fairness of the proposed procedure is apparent from the way in which it would hamper defendants should they choose to testify. Clearly, the proposed procedure would unfairly hinder defendants in their effort to explain why they believed any information they sought to obtain, the information they received, and the information they disclosed to others was not NDI. In this regard, defendants must be able to explain precisely what they knew and when, and from whom or what they learned it, and why they did not believe the material was NDI they were not authorized to have, or otherwise lacked the requisite mens rea. See Rosen, 445 F.Supp.2d at 625-26; United States v. Rosen, 240 F.R.D. 204, 209-10 (E.D.Va.2007) (Memorandum Opinion denying motion for depositions and elucidating government’s burden). Yet, the proposed procedure would unfairly hinder defendants from doing so. For example, statements like “I heard from Foreign Person C the fact about Country X, reflected at Exhibit A page 3 paragraph 4 line 2 — well, except for the last clause— and so when I asked Franklin for confirmation of that fact, I thought I was asking for a matter of public record,” may provide some exculpatory description of defendants’ state of mind, assuming the jury could decode the statement quickly enough to follow the questioning. Yet, it would be difficult, if not impossible, for defendants to explain fully why they believed the information they sought or had was in the public domain without revealing details about the information, the identity and reliability of Person C, etc. Because the proposed procedure shackles defendants in this way, it cannot pass muster under CIPA.

Finally, it must be noted that the government’s proposed frequently-changing system of coded references not only invites juror confusion, but virtually guarantees it. See Fernandez, 913 F.2d at 162 (affirming district court’s rejection of a similar government proposal on grounds of, inter alia, “artificiality” and potential to “confuse or distract the jury.”). Given the sheer number of substitutions and the proliferation of coded phrases, varying from witness to witness and overt act to overt act, the likelihood of juror confusion would be a sufficient ground, by itself, for rejecting the wholesale proposed substitutions under CIPA § 6(c). Moreover, as explained infra section V, the requested jury instruction prohibiting disclosure of classified information received at trial is also fraught with potential for jury confusion, as well as unfairness to defendants.

There is a paucity of reported cases on the propriety of using the silent witness rule under CIPA, as the rule has been infrequently proposed and even less frequently employed. Some cases have arguably approved of its use, 14 while others have rejected it. 15 Significantly, the rule’s use has never been judicially approved for use (i) where the volume of classified information to which the rule would apply is large, (ii) where defendants objected to its use, and most importantly, (iii) where the effective result of the wholesale use of the rule is to exclude the public from large portions of, and indeed, the heart of the case. Particularly instructive is the contrast between the government’s proposed procedure and Zettl, the principal government case and the only reported case arguably approving use of the silent witness rule. In Zettl, defense counsel did not object to use of the rule as to many documents, and because the issue was not presented on the government’s interlocutory appeal, the Fourth Circuit did not address the question whether the silent witness rule was appropriate for those documents. Significantly, however, the trial judge in Zettl required that the most critical document in the case be admitted in open court rather than via the silent witness rule, an issue raised on appeal, but not reached by the Fourth Circuit for procedural reasons. Zettl, 835 F.2d at 1063, 1067. Thus, even Zettl, closely read, is not an endorsement of the silent witness rule by the Fourth Circuit, but only by a district court whose support for the rule was at best equivocal, since the district court required the critical document to be admitted in open court. In contrast, Fernandez and North, the two most apposite decisions, rejected the silent witness rule, inter alia, on the ground that it would unfairly hinder the defense case by impairing cross-examination and confusing the jury. North noted that the rule is particularly ill-suited to cases involving large quantities of redacted material and numerous substitutions, an apt description of this case. 16

The government protests that the amount of information subject to the silent witness rule, or any other substitution, should not be decisive. In the abstract, this may be correct; the amount of information subject to the silent witness rule is not, by itself, determinative. Accord Pelton, 696 F.Supp. at 159-60. If appropriate substitutions provide defendants substantially the same ability to make their defense as classified documents, then in theory it does not matter that the substitutions are numerous. Yet, it is also true that the quantity of information subject to the silent witness rule does properly inform the CIPA analysis, particularly with respect to juror confusion, which impacts defendants’ ability to effectively make their defense. In practice, the greater the amount of information and documents subject to the silent witness rule, and the more codes involved, the greater the likelihood of juror confusion and the more unwieldy the rule becomes. 17 In the end, however, the most important consideration is whether the procedure impairs defendants from effectively presenting their defenses. The government’s proposed procedure impermissi-bly does so.

In sum, the government’s proposed across the board use of the silent witness rule and use of coded testimony is, at best, an unwieldy inconvenience fraught with potential for confusion. At worst, it unfairly shackles defendants to a script written by the prosecution, bewilders the jury and all but the most well-coached government witnesses, and undermines the right to a public trial. See infra section V. Given this, the government’s proposed procedure must be rejected, as it does not provide defendants with “substantially the same ability to make [their] defense as would the disclosure of the specific classified information” at issue. CIPA § 6(c). 18

V.

Defendants challenge to the government’s proposed procedure rests on constitutional as well as statutory grounds. Specifically, defendants argue that even if CIPA’s language can be stretched to cover the government’s proposed procedure, that procedure nonetheless fails as a violation of defendants’ Sixth Amendment right to a public trial and the public’s First Amendment right to a trial open to public scrutiny. Thus, even if the procedure passed muster under CIPA, constitutional analysis is nonetheless required. 19 This follows, defendants argue, because the proposed procedure effectively excludes the public from essential portions of the trial. 20

Nor is there any doubt that the portions of the trial that would be closed to the public are critical portions of the trial. It is clear that the government’s proposal precludes the public from hearing and evaluating the evidence on a crucial and contested element of the case, namely, whether the information at issue is NDI and whether defendants knew it to be such. Moreover, the quality and quantity of material the government proposes to exclude from public view is significant. Testimony about the putative NDI at issue in seven of the alleged nine disclosures would be partially closed to the public, as would the recordings and documents corresponding to this NDI. Notably, the government’s proposed procedure would treat even certain related public domain documents, including news reports, as if they were classified documents. In short, in the circumstances of this case, the government’s proposal is clearly equivalent to sealing essential aspects of the trial. See Pelton, 696 F.Supp. at 157 (“[T]he court agrees ... that the playing of the tapes to only the court, counsel, defendant, and the jury is a form of ‘closure,’ and may be permitted only in rare cases.”).

The analysis of defendants’ constitutional argument properly begins with the recognition that defendants and the public have a fundamental right to a trial open to the public. The right to a public trial contributes to just adjudication, stimulates public confidence in the judicial system, and ensures that the public is fairly apprised of the proceedings in cases of public concern. A public trial contributes to just adjudication in several ways: (i) requiring witness’ testimony to be public deters perjury; (ii) requiring a judge’s rulings to be made in public deters partiality and bias; and (iii) requiring prosecutors to present their charges and evidence publicly deters prosecutorial vindictiveness and abuse of power. In these ways, the presence of the public encourages accurate factfinding and wise use of judicial and prosecutorial discretion, thereby contributing to public confidence that justice has prevailed at trial. See Bell v. Jarvis, 236 F.3d 149, 164-65 (4th Cir.2000); see also Akhil Amar, Sixth Amendment First Principles, 84 Geo. L.J. 641, 678-80 (1996) (citing, inter alia, Blackstone, Hale, and Wigmore). In short, justice must not only be done, it must be seen to be done.

Given the important interests at stake, it is now well-settled that defendants in criminal cases and the public both have a right to a trial open to public scrutiny. See Press-Enterprise Co. v. Superior Court of California, 464 U.S. 501, 510, 104 S.Ct. 819, 78 L.Ed.2d 629 (1984) (public’s First Amendment right); Waller v. Georgia, 467 U.S. 39, 44-15, 104 S.Ct. 2210, 81 L.Ed.2d 31 (1984) (defendant’s Sixth Amendment right). It is also well-settled that the standard governing whether the public trial right has been infringed is the same whether the right is asserted by the press under the First Amendment or by a defendant under the Sixth Amendment. Waller, 467 U.S. at 46-47, 104 S.Ct. 2210. Under this constitutional standard, a trial is presumptively open, and may be closed only if certain criteria are met. These criteria are set forth in Press-Enterprise and its progeny, as follows:

(1) an “overriding interest” must exist to close the trial,

(2) the closure is no broader than necessary to protect that interest,

(3) the court considers reasonable alternatives to closure, and

(4) the court makes specific findings on the record concerning the existence of the overriding interest, the breadth of the closure, and the unavailability of alternatives to facilitate appellate review.

Press-Enterprise, 464 U.S. at 510, 104 S.Ct. 819; see also Bell 236 F.3d at 166. As the proponent of closing the proceedings, the government bears a “weighty” burden to establish that closure is permissible. Press-Enterprise, 464 U.S. at 509-10, 104 S.Ct. 819. Decisions to close trials must be made on a case by case basis, with attention to the facts and circumstances of each case; statutes per se requiring closure in certain circumstances are impermissible. Globe Newspaper Co. v. Superior Court, 457 U.S. 596, 611 n. 27, 102 S.Ct. 2613, 73 L.Ed.2d 248 (1982). Finally, before trials may be closed, the public must be given notice and an opportunity to be heard. In re Knight Publishing Co., 743 F.2d 231, 234-35 (4th Cir.1984). An erroneous denial of a public trial is a structural error not amenable to harmless error analysis. Bell, 236 F.3d at 165 (citing cases).

Each of the four Press-Enterprise elements is separately considered.

1. Overriding Interest

The government claims that its interest in protecting classified information is a compelling and overriding one, and has cited numerous cases in support of this claim. 21 While it is true, as an abstract proposition, that the government’s interest in protecting classified information can be a qualifying compelling and overriding interest, it is also true that the government must make a specific showing of harm to national security in specific cases to carry its burden in this regard. The govern ment’s ipse dixit that information is damaging to national security is not sufficient to close the courtroom doors nor to obtain the functional equivalent, namely trial by code. Press-Enterprise and In re Washington Post require more; they require a judicial inquiry into the legitimacy of the asserted national security interest, and specific findings, sealed if necessary, about the harm to national security that would ensue if the request to close the trial is not granted. See Press-Enterprise, 464 U.S. at 510, 104 S.Ct. 819; In re Washington Post, 807 F.2d 383, 391-92 (4th Cir.1986) (rejecting the government’s argument that courts should defer to Executive Branch assertions that trial closures are necessary for national security reasons, and stating that a proceeding cannot be closed merely because the case implicated CIPA at an earlier stage). Moreover, the government’s ipse dixit is insufficient whether it appears by way of classified status, or the bald assertion of counsel that information is damaging to national security. Thus, granting that national security concerns can justify appropriately tailored trial closures, the government nonetheless bears the burden of demonstrating, as a factual matter, that harm to national security would result from failing to close the trial.

Here, the government has not met its burden; instead, it has done no more than to invoke “national security” broadly and in a conclusory fashion, as to all the classified information in the case. Of course, classification decisions are for the Executive Branch, and the information’s classified status must inform an assessment of the government’s asserted interests under Press-Enterprise. But ultimately, trial judges must make their own judgment about whether the government’s asserted interest in partially closing the trial is compelling or overriding. 22 As noted, a generalized assertion of “national security interests,” whether by virtue of the information’s classified status or upon representation of counsel, is not alone sufficient to overcome the presumption in favor of open trials. Here, the government has not proffered any evidence about danger to national security from airing the evidence publicly, let alone an item-by-item description of the harm to national security that will result from disclosure at trial of each specific piece of information as to which closure is sought, as required by Press- Enterprise. 23

Moreover, quite aside from the government’s failure to provide any evidence of harm to national security from an open trial, the government’s assertion of an overriding interest justifying a trial closure is undermined by the substance of the proposed procedure, to wit, that the jury and uncleared witnesses will be permitted to receive unredacted classified information. Given that the government appears willing to trust its confidential information, to jurors, alternates, and uncleared witnesses, including (potentially) defendants, it is difficult to credit fully the government’s claim that the classified information at issue is deserving of rigorous protection. 24

The government urges that this proposed disclosure to the jury does not reflect any inconsistency on its part, because the jurors could be instructed never to disclose the classified information they received at trial. See supra note 7. Yet such an instruction appears patently unfair here, because it would suggest to the jury that the information at issue is NDI. Nor is this a minor consideration given that the status of the information as NDI is a central issue in the case, and indeed an element of the charged crime as to which the government bears the burden of proof beyond a reasonable doubt. An instruction that the jurors must treat the information they receive as a closely-held government secret is not easily reconciled with the instruction that the jury is the sole judge of the facts of the case and that it alone determines whether the information at issue is NDI, i.e., whether it is closely held and potentially damaging to national security if revealed. While the two instructions are not flatly contradictory, they would be beyond the capacity of most jurors to reconcile, and indeed, what would be done, namely the silent witness rule and redactions, coupled with the instruction against disclosure, would certainly make a much greater impression on jurors than what would merely be said to jurors, namely the instruction that the jury determines whether the information is NDI. Nor is it clear that such an instruction against disclosure could be meaningfully enforced in the post-trial future. Moreover, CIPA does not grant any authority to give a non-disclosure instruction to uncleared trial participants other than jurors, such as witnesses or defendants. The proposed instruction creates new problems without solving any others.

It is instructive to compare the extent of the limited trial closures approved in Pelton and Abu Marzook, the cases chiefly relied upon by the government, with the extensive trial closure that would result from implementing the government’s proposed procedure. Far from justifying the government’s proposal, these cases support the proposition that any procedure which results in excluding the public from hearing some evidence in the case requires Press-Enterprise analysis in which the trial judge must independently review any proffered government evidence concerning national security interests and whether such interests justify closure. In Pelton, the trial judge closed a mere five minutes of a days-long trial to protect one fact from disclosure: the identity of an ongoing active surveillance target. After squarely rejecting the notion that CIPA authorized the trial closure, the trial judge in Pelton justified the closure under Press-Enter prise because it received and credited a classified affidavit describing the specific harms to national security that would ensue if the identity of the surveillance target were disclosed. See Pelton, 696 F.Supp. at 159. In Abu Marzook, the trial judge partially closed a pre-trial suppression hearing to protect a very limited set of information: the identity of Israeli intelligence agents and the means and methods they employed. See Abu Marzook, 412 F.Supp.2d at 924-26. 25

Moreover, in both Abu Marzook and Pelton, the independent judicial analysis of the asserted interests in trial closure was informed by government affidavits or other evidence describing with particularity the harm to national security that would ensue from failure to close the trial. See Abu Marzook, 412 F.Supp.2d at 926 (national security interest was “more than just an assertion” but “supported by uncontested evidence”); Pelton, 696 F.Supp. at 158-59 (ex parte, in camera classified affidavit filed describing dangers to national security). The same cannot be said here; no affidavit or other evidence has yet been proffered describing with particularity the harm to national security that would result from presenting the evidence in this case publicly. In short, Pelton and Abu Marzook, which permitted sharply limited trial closures based upon a substantiated harm to national security, cannot be said to endorse wholesale use of the silent witness rule on virtually all the classified information in issue based on mere general allegations of harm to national security.

In short, the government’s asserted overriding interest is not treated as such by the government itself, given that the putative NDI will be disclosed to uncleared jurors and witnesses. And while the government’s cited cases support the notion that national security interests can be sufficiently compelling to justify a partial closure of the trial, they also require that the government satisfy Press-Enterprise by adequately supporting its motion with a description of the specific harm to national security that would ensue from disclosure. 26 The government has not done so here. Accordingly, its motion fails under the first prong of Press-Enterprise.

2. Narrowly tailored

For the same reasons, there is no basis to conclude that the government’s proposal is narrowly tailored to protect any national security interests. Since the government has not identified with specificity which of the classified information to be kept from public view would harm national security if disclosed to the broader public, or how the national security would be affected, it is impossible to evaluate whether the proposed closure is as narrowly tailored as possible to protect that asserted interest.

3. Alternatives

Likewise, assuming that an overriding interest exists here because disclosure of some or all of the classified information at issue would damage national security, it is evident that reasonable alternatives to closure exist here. To wit, the government could propose conventional CIPA § 6(c) substitutions that would be given to the jury and the public in the same form. Whether these substitutions would pass CIPA § 6(c) fairness standards is another question not here presented.

4. Findings

Although the conclusions concerning the other Press-Enterprise factors obviate the need for factual findings, the absence of any affidavit describing the ensuing harms to national security should the trial not be closed is noteworthy. A highly detailed explanation of the ensuing harms to national security is especially necessary to make the thorough factual findings required for a closure when, as here, much of the classified information at issue is not self-evidently damaging to national security-

In sum, the government’s proposed procedure effects a closure of the trial, and therefore must satisfy Press-Enterprise. Because no overriding governmental interest justifying a closure is apparent on this record, and because the government’s asserted national security interest is undermined by the substance of its proposed procedure, that procedure must be rejected.

Finally, a word about timing is in order. The government argued that no Sixth Amendment analysis was appropriate until after rulings on specific CIPA § 6(c) substitutions were made. But here, the proposed procedure fails whether analyzed under CIPA or Press-Enterprise. And, because the conclusion reached here is that the government’s proposed procedure is not a substitution authorized by CIPA, but constitutes a partial trial closure, the constitutional analysis is properly undertaken at this time. Indeed, an item-by-item analysis under CIPA would be futile if it is apparent that the proposed procedure, applied to virtually all the classified information in the case, would fail Press-Enterprise.

VI.

To summarize, the wholesale use of the silent witness rule, coded testimony, and redacted recordings effectively closes the trial. This procedure is not a “substitution” authorized by CIPA, and even if it were, it would not afford defendants substantially the same opportunity to present their defense as the specific classified information. Moreover, even if the procedure passed muster under CIPA, it nonetheless would have to satisfy the Press-Enterprise test because it effectively closes portions of the trial. The proposed procedure does not; the government’s showing of an overriding interest in protecting sensitive national security information has been insufficient to this point, and the government’s claim that the information at issue is damaging to the national security if publicly disclosed is belied by its own proposal to release the information to uncleared persons. This opinion does not foreclose the government from proposing specific § 6(c) substitutions that pass CIPA muster. Nor does this opinion foreclose consideration of or narrowly limited use of the silent witness rule where the government provides a specific factual basis for a claim that public disclosure would damage the national security.

For these reasons, defendants’ motion to strike must be granted, and the government’s motion to close the trial, styled a CIPA § 6(c) motion, must be denied. At the government’s request and with no ob jection from defendants, entry of an appropriate Order will be delayed to afford the government time to consider how to proceed with respect to the CIPA process.

1 . It is important to recognize that NDI and classified material may not be coextensive sets. The Executive Branch, pursuant to regulations, may classify material as Confidential, Secret, or Top Secret, and may take appropriate protective measures, such as compartmentation, to help ensure that there is no unauthorized disclosure of information essential to national security. See Exec. Order No. 13,292, 68 Fed.Reg. 15,315 (March 25, 2003) amending Exec. Order No. 12,958, 60 Fed.Reg. 19,825 (April 17, 1995) (establishing classification regime). In contrast, NDI is a statutory term that is an element of an Espionage Act violation. To qualify as NDI, information must be closely held by the government and potentially damaging to national security if disclosed. See United States v. Rosen, 445 F.Supp.2d 602, 618-25 (E.D.Va. 2006). Thus, where, as here, the indictment alleges unauthorized receipt and disclosure of NDI, the government must prove beyond a reasonable doubt that the material involved was NDI. See In re Winship, 397 U.S. 358, 364, 90 S.Ct. 1068, 25 L.Ed.2d 368 (1970); Rosen, 445 F.Supp.2d at 625. It is possible, therefore, that the set of classified material may not be coextensive with the set of NDI. In other words, not all material the Executive Branch classifies is, ipso facto, NDI; the government must prove beyond a reasonable doubt that it is NDI. See Rosen, 445 F.Supp.2d at 625. While classified status may be probative of information’s NDI status, it is not conclusive.

2 . 18 U.S.C.App. 3.

3 . Defense counsel received government security clearances to view this material. Defendants are not so cleared.

4 . Judicial rejection of the government's proposed substitutions need not lead immediately to the CIPA § 6(e) sanctions stage. Instead, the CIPA § 6(c) process, in practice, may appropriately be iterative, with the government afforded an opportunity to propose new substitutions in place of those that have been judicially rejected, or alternatively, to declassify material and withdraw the CIPA § 6(c) motion. See United States v. Libby, 467 F.Supp.2d 20, 23-24 (D.D.C.2006) (accepting government’s revised substitutions after having rejected parts of earlier attempts in, inter alia, United States v. Libby, Criminal No. 05-394, 2006 WL 3262446 (D.D.C. Nov.13, 2006), vacated by United States v. Libby, Criminal No. 05-394, 2006 WL 3333059 (D.D.C. Nov. 16, 2006)).

5 . See infra notes 14-15. It appears the Fourth Circuit in United States v. Zettl, 835 F.2d 1059, 1063 (4th Cir.1987) coined the term "silent witness rule” and described it as follows:

[U]nder such a rule, the witness would not disclose the information from the classified document in open court. Instead, the witness, would have a copy of the classified document before him. The court, counsel, and jury would also have copies of the classified document. The witness would refer to specific places in the document in response to questioning. The jury would then refer to the particular part of the document as the witness answered. By this method, the classified information would not be made public at trial but the defense would be able to present that classified information to the jury.

Importantly, defense counsel in Zettl did not object to using the rule with respect to most of the documents at issue there. Id. at 1063.

6 . In a few instances, the government proposes more conventional substitutions, namely a redacted document or a summary that will be provided to the jury and the public in the same form. These substitutions are not addressed here, but will be considered in the course of the closed CIPA § 6(c) hearing.

7 . The only authority for such an instruction is § 6 of 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 (1981) (reprinted as a note to CIPA § 9), which states that "after a verdict has been rendered by a jury, the trial judge should consider a government request for a cautionary instruction to jurors regarding the release or disclosure of classified information contained in documents they have reviewed at trial.” (emphasis added). As the emphasized language implies, the decision to give such an instruction comes after a presumably public trial and, in any event, is discretionary. Importantly, the instruction conveys to the jury, with little subtlety, that the information at issue deserves protection from dissemination, from which the jury may infer that the information is indeed NDI. The silent witness rule may likewise have this effect during the trial.

Not addressed in the government’s proposed procedure is whether a similar instruction would be necessary for uncleared witnesses and for defendants.

8 . CIPA § 6(c)(1) provides that

Upon any determination by the court authorizing the disclosure of specific classified information under the procedures established by this section, the United States may move that, in lieu of the disclosure of such specific classified information, the court order

(A) the substitution for such classified information of a statement admitting the relevant facts that the specific classified information would tend to prove; or

(B) the substitution for such classified information of a summary of the specific classified information.

The court shall grant such a motion of the United States if it finds that the statement or summary will provide the defendant with substantially the same ability to make his defense as would disclosure of the specific classified information.

CIPA § 8(b) provides that

The court, in order to prevent unnecessary disclosure of classified information involved in any criminal proceeding, may order admission into evidence of only part of a writing, recording, or photograph, or may order admission into evidence of the whole writing, recording, or photograph with excision of some or all of the classified information contained therein, unless the whole ought in fairness to be considered.

9 . United States v. North, 713 F.Supp. 1452 (D.D.C.1989) (citing H. Rep. No. 96-1436, 1980 U.S.C.C.A.N. 4294).

10 . United States v. Pelton, 696 F.Supp. 156, 157 (D.Md.1986) (“The court does not agree with the government that CIPA provides a statutory basis for the proposed closure.... There is nothing in the legislative history to suggest that the government may close all or part of a public trial.”).

11 .See infra section V and cases cited therein.

12 . See supra note 1 and cases cited therein.

13 . The impaired cross-examination may also rise to the level of a Confrontation Clause violation, a question neither reached nor decided here.

14 . See United States v. Zettl, 835 F.2d 1059, 1063 (4th Cir.1987) (noting that district court approved the silent witness rule as to some documents, but not analyzing the issue on appeal).

15 . See United States v. Fernandez, 913 F.2d 148, 161-62 (4th Cir.1990) (rejecting a “key card proposal” similar to the silent witness rule because, inter alia, it was an "artificial means of presenting evidence” which “might confuse or distract the jury”); United States v. North, Criminal No. 88-0080-02, 1988 WL 148481 at *3 (D.D.C. Dec.12, 1988) (rejecting the rule and noting that "this technique for denying public access to the full proof in the interests of protecting national security cannot serve the requirements of this particular case which will involve thousands of pages of redacted material and numerous substitutions. Cross-examination would be stultified and confusion would undoubtedly increase.”).

16 . See supra note 15.

17 . Indeed, as defendants argue, the government itself may be less than clear as to the silent witness rule’s application in certain circumstances here. With respect to some conversations alleged as overt acts, the government has proposed redacted transcripts for public consumption using codes or summaries different from the codes proffered for use by witnesses with respect to that overt act. In some instances, the government has redacted the transcripts, yet proposes to play the entire recorded conversation in open court. The government’s own inconsistent treatment of the classified information reinforces the likelihood that the silent witness rule and coded references will produce juror confusion.

18 . It is important to note that nothing in this Opinion or in the result reached is intended to suggest that the silent witness rule is per se impermissible. To the contrary, as Zettl arguably reflects, there are circumstances where use of the rule might be appropriate. Likewise, Pelton and Abu Marzook, discussed infra section V, approved procedures that shielded some information seen by the factfinder from public view, but those courts did so only after appropriate analysis of the right to a public trial. See Pelton, 696 F.Supp. at 159; Abu Marzook, 412 F.Supp.2d at 924-26. And indeed, in the course of the remaining CIPA § 6(c) process in this case, this Court may conclude, after a further in camera hearing, that the silent witness rule may appropriately be used in conjunction with a specific substitution proposed for specific classified information pursuant to CIPA § 6(c). But because the effect of the silent witness rule is to close the trial in substantial part, the rule's use, if proposed as part of a CIPA substitution, must satisfy not just CIPA's fairness requirement, but also the constitutional requirements for a trial closure, discussed more fully infra section V. Of course, the result reached here does not eliminate the need for the Court, as required by CIPA § 6(c), to consider, rule, and make findings with respect to specific substitutions for classified information proposed by the government.

19 . Conversely, Pelton recognized that a trial closure, though not authorized by CIPA, may be permissible if it passes constitutional muster. Pelton, 696 F.Supp. at 157-59.

20 . It is noteworthy in this respect that the government sensibly appears to have abandoned its original position that the proposed use of the silent witness rule, coded testimony, and redacted recordings does not close the trial because the public would be present in the courtroom. This argument, if credited, leads to the absurd result that a trial unintelligible to the public is still "open” to the public simply because the public is physically present to see and hear what they cannot understand. The public's physical presence, by itself, does not guarantee that a trial is public; it is also necessary that the trial be reasonably comprehensible to the physically present public.

21 . See, e.g., Dept. of Navy v. Egan, 484 U.S. 518, 527, 108 S.Ct. 818, 98 L.Ed.2d 918 (1988) (noting, outside Press-Enterprise context, that government has "compelling interest" in withholding national security information from unauthorized persons); Richmond Newspapers, Inc. v. Virginia, 448 U.S. 555, 598 n. 24, 100 S.Ct. 2814, 65 L.Ed.2d 973 (1980) (Brennan, J., concurring in the judgment); United States v. Ressam, 221 F.Supp.2d 1252, 1263 (W.D.Wash.2002) (national security justified partial sealing of CIPA protective orders); United States v. Pelton, 696 F.Supp. at 159 (potential disclosure of surveillance target justified partial trial closure); United States v. Abu Marzook, 412 F.Supp.2d at 925 (closure justified to protect intelligence means and methods, and identity of operatives).

22 . See In re Washington Post, 807 F.2d at 391-92; see also Pelton, 696 F.Supp. at 159 ("While the court would not find a mere assertion of 'national security’ sufficient to overcome the important first amendment values at issue in this case, the court has conducted its own analysis of the classified affidavit and the unredacted transcripts, and finds that there are serious national security concerns that would be affected” by refusing to close the trial.) (emphasis added).

23 . The government states that it will proffer affidavits demonstrating the harm to national security from disclosure of the information at issue during the course of the sealed CIPA § 6(c) hearing. Yet the burden is on the proponent of a trial closure to justify the closure. Press-Enterprise, 464 U.S. at 509-10, 104 S.Ct. 819. Nor can the government claim to have been surprised that such affidavits would be required at this stage. Once defendants’ motion to strike was filed and the Court set separate schedules for the government's CIPA § 6(c) motion and the motion to strike, the government was on notice that its proposed procedure, even though styled by the government as CIPA substitutions, would be challenged as a trial closure and would be analyzed as such. Indeed, the government's own cited cases, Pelton and Abu Marzook, demonstrate precisely the need for the government to adduce evidence supporting a closure. In any event, even assuming the government submitted affidavits for each of the pieces of classified information it seeks to introduce via the silent witness rule, its effort would nonetheless found because defendants' right to present their case would be imper-missibly impaired. See supra section IV.B.

24 . The government points out that pursuant lo CIPA § 8(a), classified information may be introduced at trial without changing its classification status. But that provision governs for purposes of CIPA only; it does not control the constitutional analysis. In particular, it does not prevent the Court from taking into account the government’s willingness to disclose classified information to uncleared individuals when assessing the sincerity or legitimacy of the government’s proffered interest in closing portions of the trial.

25 . The trial judge in Abu Marzook intimated that the government’s interest in closing the proceeding was overriding merely because the hearing testimony was classified and governed by CIPA, though it also concluded that the asserted national security interests were supported by undisputed evidence. Id. at 926. To whatever degree the court relied on the fact that CIPA governed the testimony, such reliance was inconsistent with the rule in this circuit that courts cannot defer to the government’s desire to close trials on the basis of a bare assertion of national security interest by the government, nor may courts conclude that the interest in trial closure is compelling merely because CIPA governs some aspects of the case. Rather, trial judges must conduct their own analysis to determine if the proffered national security interest is compelling or overriding. See In re Washington Post, 807 F.2d at 391-92; Pelton, 696 F.Supp. at 159.

26 . If the harm to national security were patently obvious and incontrovertible, no affidavits or other evidence in this regard might be necessary. That is not the case here.