Soon after the start of Ingmar Guan-dique’s trial for the murder of Chandra Levy, The Washington Post (“The Post”) sought access to the jury questionnaires completed by the sixteen empaneled jurors. When informal attempts to gain access were denied, The Post filed a motion to intervene, arguing that both the common law and the First Amendment create a presumption of public access to jury questionnaires used in voir dire. After the trial ended, the court issued an oral ruling denying access, concluding that disclosure of the jury questionnaires would discourage juror candor and intrude on juror privacy. The Post appealed. We hold that The Post, as a surrogate for the public, has a presumptive right of access to the jury questionnaires used in this case, and the trial court erred in not recognizing that right. We further hold that the trial court erred by failing to exercise its discretion in making specific findings about the proteetible privacy interests at stake and considering alternatives to complete closure. We therefore reverse and remand for further proceedings consistent with this opinion.
I. Background
In the spring of 2001, a young congressional intern disappeared after going for a morning jog in Washington’s Rock Creek Park. In October 2010, nearly a decade after Chandra Levy’s highly publicized disappearance, Guandique stood trial for her murder. The circumstances of Chandra Levy’s murder and the ensuing investigation were closely followed by The Post and other national media organizations. The coverage of her case captured the nation’s attention. Not surprisingly, then, the trial court and the parties devoted considerable discussion to the process of selecting a fair and impartial jury in Guandique’s trial. Several factors made the jury selection process more complex than it would have been in a less publicized case. First, of course, there was the volume of press coverage over the years, including an eight-part series in The Post looking into why Chandra Levy’s case remained unsolved for so long. Sari Horwitz, Scott Higham & Sylvia Moreno, Who Killed Chandra Levy? Wash. Post, July 13-27, 2008. Guandique’s defense counsel also identified relevant sensitive issues that might indicate an inability on the part of prospective jurors to be impartial, including attitudes about Latino ethnicity, illegal immigration, and gang affiliation.
Over the course of several months leading up to the trial, the court held at least five status hearings during which the parties discussed the use of comprehensive questionnaires to aid the jury selection process. The court recognized that the case was “unique” in the number of issues that had to be “confronted” during the selection process. At one hearing, the court acknowledged that “questionnaires would be very helpful” because of the complexity of the issues to be considered in selecting a jury for this closely watched murder case. The use of questionnaires was designed to streamline the selection process by eliminating some prospective jurors for cause based on their written answers, which would be reviewed prior to oral voir dire. As the trial court noted, “whatever follow-up questioning there is would be in some ways more limited than it might be if we didn’t use [a] questionnaire because we’ve got all the information and hopefully we can sift through it....” Eventually, the court and the parties agreed to use a questionnaire consisting of fifty-five questions seeking routine demo*883graphic information, such as age, gender, educational background, and employment status, and also asking questions uniquely relevant to Guandique’s trial, such as familiarity with Rock Creek Park, knowledge of local gang activity, and views on illegal immigration. The questionnaire identified prospective jurors only by their juror number; no personally identifiable information such as names, addresses, or social security numbers was sought.
Jury selection began on Monday, October 18, 2010. Members of the press were in attendance during the entire week of voir dire. On the first day, prospective jurors were questioned about preliminary issues that might disqualify them from jury service and were then asked to complete the questionnaire. Those prospective jurors not immediately disqualified returned on Wednesday, October 20, 2010, for individual questioning based on their completed questionnaires.1 The court made no oral promises of confidentiality regarding the questionnaires. However, a cover letter accompanying each questionnaire stated that the questionnaires would be “returned to the Clerk of the Court and kept in confidence, under seal, not accessible to the public or media.” The press was given a blank copy of the questionnaire at some point early in the process. Although it is not clear from the record, the parties do not dispute that the blank copy given to the press included the cover letter with the confidentiality promise. Voir dire continued until Friday, October 22, 2010. The following Monday, October 25, 2010, the jury was sworn and the trial began.
Soon after the trial began — the record does not indicate precisely when — The Post made several informal requests2 to access the completed questionnaires of the sixteen empaneled jurors. These requests were denied. On November 3, 2010, in the middle of the government’s case-in-chief, The Post and three other media organizations filed a motion for leave to intervene to access the juror questionnaires. In that motion, The Post argued that it was “presumptively entitled to contemporaneous access” to the jury selection process and that the court had no compelling reason for “the blanket refusal to disclose the questionnaires” when it could have made limited redactions to address any individual privacy concerns. The court did not schedule a hearing on the motion and made no immediate ruling. Nine days later, on November 12 — now into the third day of the defense’s case — the court gave counsel for the media intervenors a “brief, impromptu opportunity to be heard on the record” after it notified them that it would release juror information only on age, gender, education level, and occupation. Rejecting the argument that a blanket closure was constitutionally impermissible absent a compelling interest, the court refused to release the questionnaires mid-trial. The court cited concerns that the usefulness of such questionnaires would diminish if the court allowed “full access” and opined that, as a matter of “decency” and “etiquette” to the jurors who were promised confidentiality, the court could not release the questionnaires without first discussing the matter with them.
On November 16, the media organizations filed a request for formal, on-the-record findings from the court explaining its refusal to provide access to the jury questionnaires. Although neither the *884prosecution nor the defense opposed this request, it was not until November 24, two days after the jurors returned their guilty verdict against Ingmar Guandique, that the court held a formal hearing to explain its rationale for declining to release the completed questionnaires. At that hearing, the court began by noting how “unprecedented” press access had been at the trial, calling it the “broadest of any criminal trial in the Superior Court that I’m aware of.” The court also concluded that it was “atypical” that the press had heard all the individual questioning of prospective jurors. With this “backdrop” in mind, the court went on to justify its closure decision on the basis of promoting juror candor through a promise of confidentiality on written questionnaires and protecting juror privacy and a fair trial. The court stated that because of its “guarantee of privacy,” the questionnaires were more likely to elicit “full candor from the jury.” The court also indicated its belief that written rather than oral questioning would encourage more candid responses. Finally, based on an off-the-record discussion with the jurors in the middle of trial, the court advised the parties that the jurors were opposed, “to a person,” to the release of their questionnaires. This, coupled with the questionnaire’s promise of confidentiality, caused the court to say: “And so having told the jury that, I intend to live up to that promise unless the Court tells me that I have to do otherwise.” The Post appealed the court’s oral ruling denying access to the completed questionnaires.
II. Discussion
A. The Post Made a Timely Request for Access to the Jury Questionnaires
The government concedes that the trial court’s blanket promise of confidentiality was inappropriate, but it urges us to find The Post’s intervention untimely without reaching the merits of the case. Although we ultimately find this argument unconvincing, we note that had The Post intervened earlier, we would not be faced with the dilemma of fashioning a remedy when contemporaneous access to the jury questionnaires is no longer possible.
The government’s waiver argument does not work well in the context of the public’s First Amendment right of access. As this court has explained, “[t]o the extent that [a common law and First Amendment right of access] exists, it exists today for the records of cases decided a hundred years ago as surely as it does for lawsuits now in the early stages of motions litigation.” Mokhiber v. Davis, 537 A.2d 1100, 1105 (D.C.1988). Certainly, it is when the trial is unfolding that the public’s interest is greatest, but that interest does not necessarily end at the close of trial. This court has recognized the “special nature” of the public right of access to matters in litigation, finding that “[o]rdinary principles applicable to [filing a timely] intervention” are not wholly apposite. Id. at 1104. The right of public access is “a right that any member of the public can assert,” whether it is for the purpose of reporting on a trial as it unfolds or researching jury selection ten years later. See id. at 1105.
The government argues that The Post’s intervention was untimely because the trial court had already promised the jurors that their questionnaires would not be released and this “promise of confidentiality could not be remedied mid-trial.” We disagree that once the promise was made there was no going back. The trial court could have explained to the jurors that the guarantee of confidentiality was made in error and that their questionnaires were subject to a constitutional presumption of disclosure unless they had particular privacy concerns, in which case they could request an in camera (but on-the-record) proceeding to discuss those concerns and whether *885they outweighed the presumption of access. That the trial court might have reconsidered making a promise of confidentiality had The Post intervened before the start of voir dire does not convert the court’s promise into an acceptable one or prevent The Post from later asserting its rights.
The government also makes a peculiar plain-error argument, suggesting that The Post’s motion to intervene was not preserved for appeal. As we have noted, The Post’s motion to intervene could be made at any time, and so it was not forfeited when The Post failed to intervene at the earliest possible date. Moreover, the record demonstrates that the trial judge was fully aware of The Post’s request for access to the questionnaires and was not “taken by surprise” that disclosure of the questionnaires was an issue.
Accordingly, we reject the government’s attack on the timeliness of The Post’s challenge. We therefore turn to a de novo review of whether the First Amendment right of public access applies to the jury questionnaires used in this case.
B. The First Amendment Right of Access Applies to Jury Questionnaires Used in Voir Dire
The public’s presumptive First Amendment right of access to criminal trials, including the jury selection process, is well settled. Press-Enterprise Co. v. Superior Court, 464 U.S. 501, 104 S.Ct. 819, 78 L.Ed.2d 629 (1984) (Press-Enterprise I). This broad constitutional right of access is grounded in the history of public trials. Richmond Newspapers, Inc. v. Virginia, 448 U.S. 555, 100 S.Ct. 2814, 65 L.Ed.2d 973 (1980). As the Court observed in Richmond Newspapers, “throughout its evolution, the trial has been open to all who care to observe.” Id. at 564-65, 100 S.Ct. 2814. The value of public trials is undisputed. The presence of the public and the press at criminal trials “historically has been thought to enhance the integrity and quality of what takes place.” Id. at 578, 100 S.Ct. 2814. Open trials contribute to the “proper functioning” of the system by, for example, “discouraging] perjury, the misconduct of participants, and decisions based on secret bias or partiality.” Id. at 569, 100 S.Ct. 2814. Not only does public access enhance just results, it also promotes the “appearance of fairness so essential to public confidence in the system.” Press-Enterprise I, 464 U.S. at 508, 104 S.Ct. 819.
When it extended the First Amendment right of access to the voir dire examination of potential jurors, the Court noted the significance of jury selection both to the parties and to the proper functioning of the criminal justice system. Press-Enterprise I, 464 U.S. at 505, 104 S.Ct. 819. Historical evidence shows that attendance at trial was “virtually compulsory” for free members of the community because it was these members of the public who “rendered] judgment.” Id. Although it is now a six-to-twelve member jury—or a judge— that decides cases, the public-at-large has a valid interest in “learn[ing] whether the seated - jurors are suitable decision-makers.” United States v. Blagojevich, 612 F.3d 558, 561 (7th Cir.2010). Today, when “attendance at court is no longer a widespread pastime,” the public relies on the press for firsthand accounts of the justice system at work. Richmond Newspapers, 448 U.S. at 572, 100 S.Ct. 2814. Indeed, members of the press are treated as “surrogates for the public,” id. at 573, 100 S.Ct. 2814, and their access cannot be “foreclosed arbitrarily.” Id. at 577, 100 S.Ct. 2814.
That a significant part of voir dire in this case was conducted through written questionnaires and not orally is of no constitutional significance. We can think of no principled reason to distinguish written *886questions from oral questions for purposes of the First Amendment right of public access. Jury questionnaires merely facilitate and streamline voir dire; their use does not constitute a separate process. Every court that has decided the issue has treated jury questionnaires as part of the voir dire process and thus subject to the presumption of public access. See In re South Carolina Press Ass’n, 946 F.2d 1037, 1041 (4th Cir.1991) (applying the presumption of access to jury questionnaires); United States v. McDade, 929 F.Supp. 815, 817 n. 4 (E.D.Pa.1996) (finding that Press-Enterprise I “eneompass[es] all voir dire questioning — both oral and written”); In re Washington Post, 1992 WL 233354, at *4 (D.D.C. July 23, 1992) (applying the presumption of access to jury questionnaires); Bellas v. Superior Court, 85 Cal.App.4th 636, 102 Cal.Rptr.2d 380, 386 (2000) (following other courts that “make clear that the content of juror questionnaires [is] publicly accessible” unless the presumption is outweighed by a competing interest; the “limitation on access is tailored as narrowly as possible”; and “the trial court’s findings are articulated with enough specificity that a reviewing court can determine” whether access was properly limited); Copley Press, Inc. v. Superior Court, 228 Cal.App.3d 77, 278 Cal.Rptr. 443, 451 (1991) (“The fact that the questioning of jurors was largely done in written form rather than orally is of no constitutional import.”); Stephens Media, LLC v. Eighth Judicial Dist. Court, 221 P.3d 1240, 1249 (Nev.2009) (“[T]he use of juror questionnaires does not implicate a separate and distinct proceeding.... [It is] merely a part of the overall voir dire process.”); In re Newsday, Inc. v. Goodman, 159 A.D.2d 667, 552 N.Y.S.2d 965, 967 (1990) (“[(Questionnaires completed by the petit jurors in this criminal action were an integral part of the voir dire proceeding.”); Forum Commc’ns Co. v. Paulson, 752 N.W.2d 177, 185 (N.D.2008) (holding that use of jury questionnaires “serves as an alternative to oral disclosure of the same information in open court”); State ex rel. Beacon Journal Publ’g Co. v. Bond, 98 Ohio St.3d 146, 781 N.E.2d 180, 188 (2002) (“Because the purpose behind juror questionnaires is merely to expedite the examination of prospective jurors, it follows that such questionnaires are part of the voir dire process”).
It is evident that the jury questionnaires here were used to facilitate the jury selection process by exposing any biases relating to, among other issues, Latino ethnicity, illegal immigration, and gang affiliation that otherwise would have been explored through oral questioning. The presumption, then, is that the completed questionnaires, as a part of voir dire, should be available to the press.3
*887Although the right of access is not absolute, the Court explained in Press-Enterprise I that “[t]he presumption of openness may be overcome only by an overriding interest based on findings that closure is essential to preserve higher values and is narrowly tailored to serve that interest.” 464 U.S. at 510, 104 S.Ct. 819. The Court directed specific procedures to be followed in considering whether the presumption has been overcome. The trial court must first “articulate ... with the requisite specificity” the “protee-tible privacy interests” at stake and then consider whether alternatives to complete closure are available to protect those privacy interests. Id. at 513, 104 S.Ct. 819. “Absent consideration of alternatives to closure, the trial court [may] not constitutionally close the voir dire.” Id. at 511, 104 S.Ct. 819. To prevent unnecessary closure, the trial court should inform prospective jurors that if they are concerned about certain sensitive questions, they may affirmatively request “an opportunity to present the problem to the judge in camera but with counsel present and on the record.” Id. at 512, 104 S.Ct. 819. See also Beacon Journal, 781 N.E.2d at 189 (“[A]n individualized examination of each prospective juror’s circumstances is appropriate in considering the privacy interests of such jurors.”). Even then, a juror’s request to keep information private must be balanced against the “constitutional values sought to be protected by holding open proceedings.” Press-Enterprise I, 464 U.S. at 512, 104 S.Ct. 819. In the end, “[t]he trial judge should seal only such parts of the transcript as necessary to preserve the anonymity of the individuals sought to be protected.” Id. at 513, 104 S.Ct. 819.
C. The Trial Court Erred by Failing to Follow the Press-Enterprise I Procedures in This Case
Having determined that the presumption of access applies to the jury questionnaires used in this case, we must now consider whether the trial court proceeded in accordance with Press-Enterprise I. That is, did the trial court articulate specific protectible privacy interests and consider alternatives to complete closure to protect those interests? The Press-Enterprise I procedures call for the “exercise of sound discretion by the court” to “minimize the risk of unnecessary closure” by ensuring that there exists a “valid basis” for concluding that disclosure will infringe a significant privacy interest. Press-Enterprise I, 464 U.S. at 512, 104 S.Ct. 819. It is clear from the record that the trial court failed to exercise this discretion. We have held that “[failure to exercise choice in a situation calling for choice is an abuse of discretion.” Johnson v. United States, 398 A.2d 354, 363 (D.C.1979).
At the hearing on The Post’s motion, the trial court explained that when it asked the jurors — in what appears to have been an off-the-record discussion — whether they would have a problem with disclosing their questionnaires, they advised that they were opposed, “to a person.” Yet the court cited no specific privacy interests raised by any particular juror and considered no alternatives to blanket closure.4 We do not know whether the trial court posed a general question to the jurors *888about thoughts on disclosure or whether it made an individualized inquiry into the privacy interests at stake. The court’s oral findings give us little to analyze. It is worth noting, however, as The Post points out, that several jurors spoke to the press after the verdict and voluntarily appeared on television and in the news. Evidently, the jurors were not insistent “to a person” on maintaining their privacy after the trial.
The trial court also expressed concerns about the effect of disclosure on juror candor.5 The court believed that public disclosure would diminish the value of jury questionnaires as a tool to elicit candid responses and would result in decreased use of jury questionnaires in the future. The court had no problem keeping oral voir dire open to the public, however, and it cited no authority to suggest that disclosure of written answers would somehow affect a juror’s response in a way that it would not in the context of oral voir dire. We agree with other jurisdictions that a generalized concern about juror candor is not enough to overcome the presumption of open access. See, e.g., ABC, Inc. v. Stewart, 360 F.3d 90, 102 (2d Cir.2004) (rejecting the trial court’s “concluso-ry ... findings” that “the presence of reporters at voir dire .proceedings would have ... chilled juror candor”); Stephens Media, 221 P.3d at 1253 (“[W]e caution district courts from hastily closing voir dire proceedings because of the possibility that the presence of the press would inhibit juror candor.”).
This is not to say that concerns about juror candor may never justify closure. It may be that in some cases the need for a “modest limitation on access” will override the presumption of openness. See United States v. King, 140 F.3d 76, 83 (2d Cir.1998) (upholding the closure of jury questionnaires where “in this particular case of a defendant of unusually high visibility, already subject to extraordinarily hostile publicity, the airing of jurors’ responses will significantly inhibit the candor necessary to assure a fairly selected jury and therefore a fair trial”). The Second Circuit upheld the limitations on access to jury questionnaires in the trial of boxing promoter Don King, id., because the trial court made detailed findings that such a limitation was necessary to ensure candor in “the delicate area of possible racial bias.” United States v. King, 911 F.Supp. 113, 117 (S.D.N.Y.1995). Although that “delicate area” also was at play in Guan-dique’s trial, the court made no specific *889individualized findings that publicity would impact the impartiality of prospective jurors and it considered no alternatives to blanket closure.
Finally, we reject the government’s argument that the trial court’s promise of confidentiality, although improper “as a matter of policy,” cannot be undone and that it would be unfair to disclose the questionnaires at this point. Promises of confidentiality in this context are not merely inappropriate; they are constitutionally unsound. Such a promise does not trump the First Amendment right of access. See Beacon Journal, 781 N.E.2d at 190 (“Constitutional rights are not superseded by the mere promise of a trial judge to act contrary to those rights.”).
III. Remedy
Having determined that the trial court’s blanket closure was improper, we now consider the appropriate remedy. The Post originally sought contemporaneous access to the jury questionnaires. We can do nothing about that. But, as we noted above, the public’s interest in the case does not end with the verdict. The values underlying the First Amendment right of access — for example, the public trial as a check on the fair functioning of the criminal justice system — are served even after the verdict is in. This is especially true where, as here, the defendant’s direct appeal of his conviction is pending in this court. Moreover, The Post, as a surrogate for the public, has an ongoing interest in the jury selection process as a general matter of civie interest.
Although literal compliance with Press-Enterprise I is not possible after the fact, we see no reason why, on remand, the trial court cannot rectify the mistake made when it promised the jurors that the completed questionnaires would be kept confidential. We therefore remand the case to the trial court to “unscramble the egg” broken when the court made a promise to the prospective jurors that could not be kept consistent with the constitutional command of Press-Enterprise I.
Before discussing the procedures to be followed on remand, we add a note about the trial court’s promise of confidentiality. It is apparent that the trial judge was concerned with keeping his word (“I intend to live up to that promise unless the Court tells me that I have to do otherwise.”), and we empathize with his plight. At the same time, we see no reason why the trial court may not, if it so chooses, recall the jurors and advise them of this court’s decision on appeal, holding that The Post’s request for access to the completed questionnaires was wrongly denied. The trial court may then proceed with the procedures mandated by Press-Enterprise I, explaining to the jurors that this court has ordered those procedures to be followed in this case.
Beyond that, we leave it to the trial court’s discretion, in the first instance, to fashion the remand proceedings, although we must make clear what we believe is required. First and foremost, and in keeping with the teachings of Press-Enterprise I and its progeny, the trial court must start with the presumption that the completed jury questionnaires should be disclosed in their entirety. If the court finds that responses to any of the questions “touch[ ] on deeply personal matters” that may warrant redaction, the court should recall those jurors to provide them with an opportunity to raise any concerns they might have in camera and on the record. See Press-Enterprise I, 464 U.S. at 511, 104 S.Ct. 819. Ultimately, however, the decision to withhold any of the questionnaire responses remains the responsibility of the court. Accordingly, the questionnaires must be disclosed with any court-ordered redactions supported by *890specific individualized findings capable of appellate review.6 These procedures will adequately protect both the public’s right of access and any special privacy interests of the jurors.
Reversed and remanded.
. Tuesday, October 19, was a "gap day” when the parties reviewed the completed questionnaires.
. The Post "made both oral and e-mail requests” through the Superior Court’s public affairs officer, which The Post describes as "the customary procedure by which the news media seek court records in Superior Court.”
. Relying on our decision in Mokhiber, the government urges us to decide this case on common law grounds, without reaching the First Amendment question. The principle of constitutional avoidance is inapposite here. It is well settled after Press-Enterprise I that, under the First Amendment, access to the jury selection process is presumed, and competing privacy interests or fair trial concerns do not warrant complete closure without considering alternatives. In Mokhiber, this court's holding relied on the common law in part because it was dealing with a civil proceeding. Mokhiber, 537 A.2d at 1107 n. 4. The Supreme Court has emphasized that "a right of access to criminal trials in particular” deserves special First Amendment protection. Globe Newspaper Co. v. Superior Court, 457 U.S. 596, 605, 102 S.Ct. 2613, 73 L.Ed.2d 248 (1982). We also noted in Mokhiber a distinction between judicial records, for which there is a common law right of access, and judicial proceedings, for which there is both a common law and a First Amendment right of access. Mokhiber, 537 A.2d at 1107 (declining to rely on the line of Supreme Court cases dealing with a First Amendment right of access to criminal proceedings because "we deal here with a question of access to court records, not to court proceedings”). That the jury questionnaires themselves are technically *887documents and not proceedings to attend does not alter application of the First Amendment right of access because, as we have said, the questionnaires are part of the voir dire proceeding, not separate from it. Having so concluded, we cannot ignore clear authority on the application of the First Amendment.
. The court earlier had agreed to a very limited disclosure of the jurors’ age, gender, education, and occupation.
. The issue of juror candor is treated as an interest that implicates a defendant's Sixth Amendment right to a fair trial. See, e.g., ABC, Inc. v. Stewart, 360 F.3d 90, 100-02 (2d Cir.2004); In re South Carolina Press Ass'n, 946 F.2d at 1042 (accepting the trial court’s finding that "[i]f the voir dire is to serve its function as the safeguard of the defendant’s sixth amendment rights, then clearly candor must be the hallmark of such a proceeding”). When a defendant’s right to a fair trial is the competing interest, the proceeding at issue "shall be closed only if specific findings are made demonstrating that, first, there is a substantial probability that the defendant's right to a fair trial will be prejudiced by publicity that closure would prevent and, second, reasonable alternatives to closure cannot adequately protect the defendant’s fair trial rights.” Press-Enterprise Co. v. Superior Court, 478 U.S. 1, 14, 106 S.Ct. 2735, 92 L.Ed.2d 1 (1986) (Press-Enterprise II). We think it important in this case that Guan-dique, represented by able trial counsel, expressed no objection to The Post's request for access to the jury questionnaires and reiterated that position in a statement filed with this court on appeal. With no objection from Guandique, it is difficult to conclude that disclosure would prejudice Guandique’s right to a fair trial because, as the Second Circuit noted in Stewart, in which the government, not the defendant, sought closure, "[i]f openness would truly have jeopardized the fair trial rights of the defendants in this case, we imagine that the defendants, represented by experienced counsel, would have initiated the request for closure.” 360 F.3d at 102.
. We note, however, that the United States acknowledges that privacy interests are not implicated in all, or even most, of the questionnaire responses in this case, and our own reading of the questions leads us to the same belief. While we do not presume to prejudge the answer, we cannot help noting that it seems unlikely that the answers to the questions will raise serious privacy concerns of the magnitude needed to override the public’s interest in a completely open voir dire process.