dissenting:
Appellant Aaron Ross was arrested in 2008 and again in 2009 after he refused to obey Officer Wayne Early’s orders to stop leafleting in the middle of the sidewalk adjacent to Baltimore’s First Mariner Arena, which was hosting the circus. On both occasions, Ross had sought to exercise his First Amendment right to protest the circus’s treatment of animals by handing out leaflets to passersby, and on both occasions, Ross was conducting himself in a peaceful and unobtrusive manner. The sole basis for Early’s order was that Ross was violating what the majority opinion refers to as Baltimore’s “Policy” pertaining to where circus protestors could stand.
But that “Policy” constituted nothing more than an e-mail — copied, pasted, and resent with minor modifications year after year — from Baltimore’s city attorney to about a dozen members of the police department and city staff. It is undisputed that the staff attorney’s e-mail that formed the only basis of what the majority characterizes as “Policy” was neither adopted by the Baltimore City Council nor disseminated to the public in any systematic manner. And there is little dispute, if any, that the only people who knew about the existence of the e-mail were the unelected city employees who developed and sought to enforce its restrictions.
Simply put, the staff attorney’s e-mail does not constitute Baltimore’s “Policy.” Additionally, the secret nature of Baltimore’s restrictions on First Amendment rights warrants the application of heightened scrutiny because of the potential for abuse and selective enforcement associated with the lack of notice and democratic accountability. But even if we were to allow the parties to agree that our review should be under a lower standard of scrutiny, Baltimore’s restrictions fail the narrow-tailoring analysis.
*563It is axiomatic that our most basic notions of due process are jeopardized when speech restrictions are developed secretly in the back offices of city hall rather than publicly in the council chambers. It seems plausible to me that today’s decision will encourage local governments to avoid the time-consuming and politically costly exercise of adopting speech-restrictive ordinances in favor of developing speech-restrictive “Policies” at the staff level. Accordingly, I must respectfully dissent from the differing view of my fine colleagues in the majority.1
I.
Most of the salient facts in this case are covered in the majority opinion and the district court’s two published opinions, Ross v. Early, 758 F.Supp.2d 313 (D.Md. 2010) (“Ross /”) and Ross v. Early, 899 F.Supp.2d 415 (D.Md.2012) (“Ross II”). A few critical points, however, warrant special emphasis.
First, Ross was arrested in 2008 and 2009 for failing to obey Early’s orders, which were “aimed at enforcing the City’s Protocol.” Ross II, 899 F.Supp.2d at 427. The district court explicitly stated that “[t]he record simply does not support” that Ross was threatening the public safety and that “Ross was in no way blocking or impeding the free flow of patrons attempting to enter or exit the building.” Id. Moreover, when Early was asked during his deposition whether he “ultimately arrested Mr. Ross for violating a law and not for violating an e-mail[,]” Early answered, “No, it was both.” J.A. 324. Thus, Early’s decisions to order Ross to move and to place Ross under arrest were clearly based on Ross’s failure to conform to the restrictions in the e-mail and not on Ross’s interference with pedestrian flow or public safety.
Second, the district court found that the significant government interests that the restrictions were designed to serve were the preservation of “freedom of movement on public streets and sidewalks[,]” Ross I, 758 F.Supp.2d at 322, as well as “pedestrian safety,” Ross II, 899 F.Supp.2d at 425.
Third, when Ross was arrested in 2008 and 2009, the restrictions were in their fifth and sixth years of enforcement. Nonetheless, the restrictions had not been formally adopted, and there is no evidence that the City took measures to inform the public about them. Additionally, the district court noted in both of its opinions that the police officers’ orders were somewhat vague and left the protestors confused. See Ross II, 899 F.Supp.2d at 427-28 (“Further, at one point Ross is simply told ‘red brick,’ a reference to the bricked outer portion of the sidewalk, the Protocol’s designated area for demonstrations.”); Ross I, 758 F.Supp.2d at 325 (“On the video showing [Ross’s] arrest, demonstrators are seen repeatedly expressing confusion about the source of the restrictions and the interaction with their First Amendment rights. The police could not dispel this confusion because they were unable to direct demonstrators to a specific regulation or ordinance and could only instruct them to call the Law Department.”).
Finally, in Ross II, the district court granted summary judgment to Early and the municipal defendants on all of Ross’s claims, with the exception of Ross’s See*564tion 1983 claims against the City and the Baltimore City Police Department, which alleged that the speech restrictions were unconstitutional. Ross II, 899 F.Supp.2d at 421, 432-33. The district court determined that the resolution of these claims turned on a disputed factual question, namely whether the restrictions were generally applicable or “targeted toward animal welfare demonstrators specifically[.]” Id. at 422. The district court held that if the restrictions were generally applicable, they would be constitutional by virtue of the more lenient standard of review that applies to ordinances and statutes. Id. at 421-22 (explaining that under intermediate scrutiny, “a time, place, and manner restriction on speech is narrowly tailored ‘so long as the ... regulation promotes a substantial government interest that would be achieved less effectively absent the regulation.’ ” (quoting Ward v. Rock Against Racism, 491 U.S. 781, 799, 109 S.Ct. 2746, 105 L.Ed.2d 661 (1989))). The district court went on to explain that if, on the other hand, the restrictions were targeted toward the circus protestors, then they would be “more analogous to an injunction than a statute of general application, and failing to be sufficiently tailored under heightened scrutiny, would be struck down as unconstitutional.” Id. The parties then stipulated that the restrictions were generally applicable.
II.
The First Amendment prohibits the government from “abridging the freedom of speech ... or the right of the people peaceably to assemble!.]” U.S. Const, amend. I; see also Gitlow v. New York, 268 U.S. 652, 666, 45 S.Ct. 625, 69 L.Ed. 1138 (1925) (incorporating the freedom of speech against the states) and DeJonge v. Oregon, 299 U.S. 353, 364-65, 57 S.Ct. 255, 81 L.Ed. 278 (1937) (incorporating the freedom of assembly against the states). “Leafletting and commenting on matters of public concern are classic forms of speech that lie at the heart of the First Amendment, and speech in public areas is at its most protected on public sidewalks, a prototypical example of a traditional public forum.” Schenck v. Pro-Choice Network of W. N.Y., 519 U.S. 357, 377, 117 S.Ct. 855,137 L.Ed.2d 1 (1997).
But our constitutional speech rights are not unlimited because the First Amendment “does not guarantee the right to communicate one’s views at all times and places or in any manner that may be desired.” Heffron v. Int’l Soc’y for Krishna Consciousness, Inc., 452 U.S. 640, 647, 101 S.Ct. 2559, 69 L.Ed.2d 298 (1981). Thus, “even in a public forum the government may impose reasonable restrictions on the time, place, or manner of protected speech, provided the restrictions ‘are justified without reference to the content of the regulated speech, that they are narrowly tailored to serve a significant governmental interest, and that they leave open ample alternative channels for communication of the information.’ ” Ward v. Rock Against Racism, 491 U.S. 781, 791, 109 S.Ct. 2746, 105 L.Edüd 661 (1989) (quoting Clark v. Community for Creative Now-Violence, 468 U.S. 288, 293, 104 S.Ct. 3065, 82 L.Ed.2d 221 (1984)).
I agree with the majority opinion that Baltimore’s “Policy” is content neutral and leaves open ample alternative channels of communication. Content-neutral speech restrictions are subject to one of two standards of scrutiny: heightened or intermediate. See Madsen v. Women’s Health Center, Inc., 512 U.S. 753, 764-65, 114 S.Ct. 2516, 129 L.Edüd 593 (1994). As discussed below, the main difference between the two standards is the rigor of the narrow-tailoring analysis. Before reaching the narrow-tailoring analysis, however, *565a court must first select the appropriate standard of scrutiny.
A.
Under the first step of the analysis— selecting the appropriate standard of scrutiny — we are guided by two major Supreme Court decisions: Ward, 491 U.S. at 781, 109 S.Ct. 2746, and Madsen, 512 U.S. at 753, 114 S.Ct. 2516.2 When we review “a content-neutral, generally applicable statute, instead of an injunctive order, its constitutionality [is] assessed under the standard set forth in Ward[.] ” Madsen, 512 U.S. at 764,114 S.Ct. 2516. Injunctive orders, by contrast, “require a somewhat more stringent application of general First Amendment principles....” Id. at 765,114 S.Ct. 2516. Thus, in general, to determine whether the appropriate standard in a case is heightened or intermediate scrutiny, courts must consider whether the restriction is an injunction or an ordinance. But as this case illustrates, not every speech restriction fits neatly into one category or the other. And when that happens, courts must conduct a fact-intensive inquiry to determine whether the restriction is more like an ordinance or more like an injunction.
Madsen’s interpretation of Ward provides a description of the “obvious differences ... between an injunction and a generally applicable ordinance.” Madsen, 512 U.S. at 764, 114 S.Ct. 2516. First, and perhaps most importantly, “[o]rdinances represent a legislative choice regarding the promotion of particular societal interests.” Injunctions, “by contrast, are remedies imposed for violations (or threatened violations) of a legislative or judicial decree.” Id. Second, injunctions “carry greater risks of censorship and discriminatory application than do general ordinances.” Id. “ ‘[T]here is no more effective practical guaranty against arbitrary and unreasonable government than to require that the principles of law which officials would impose upon a minority must be imposed generally.’ ” Id. (quoting Ry. Express Agency, Inc. v. New York, 336 U.S. 106, 112-13, 69 S.Ct. 463, 93 L.Ed. 533 *566(1949)). Third, injunctions “can be tailored by a trial judge to afford more precise relief than a statute where a violation of the law has already occurred.” Id. at 765,114 S.Ct. 2516.
If our analysis of the restriction reveals that it is more like an ordinance, then we would apply the intermediate standard. But if our analysis reveals that the restriction is more like an injunction, then we would undertake a “somewhat more stringent application of general First Amendment principles[.]” Id. To put it simply, although speech-restrictive ordinances and injunctions all must be narrowly tailored, the fit between the speech restriction and the government’s goals must be closer with an injunction than with an ordinance. The next section contains a more detailed description of the differences between Ward’s intermediate-scrutiny standard and Madsen’s, heightened-scrutiny standard.
B.
In Ward, a case involving sound-amplification guidelines that applied to all users of a bandshell in Central Park, the Court held that a “regulation of the time, place, or manner of protected speech must be narrowly tailored to serve the government’s legitimate, content-neutral interests but that it need not be the least restrictive or least intrusive means of doing so.” Ward, 491 U.S. at 798, 109 S.Ct. 2746. The Court went on to explain that the narrow-tailoring requirement is satisfied “ ‘so long as the ... regulation promotes a substantial government interest that would be achieved less effectively absent the regulation.’ ” Id. at 799, 109 S.Ct. 2746 (quoting United States v. Albertini, 472 U.S. 675, 689, 105 S.Ct. 2897, 86 L.Ed.2d 536 (1985)). But the Supreme Court did not stop there. It continued: “To be sure, this standard does not mean that a time, place, or manner regulation may burden substantially more speech than is necessary to further the government’s legitimate interests. Government may not regulate expression in such a manner that a substantial portion of the burden on speech does not serve to advance its goals.” Id. at 799, 109 S.Ct. 2746.
In Madsen, a case in which abortion protestors were enjoined from standing within certain “buffer zones” on public and private property, the Court explained that the “standard time, place, and manner analysis is not sufficiently rigorous.” Madsen, 512 U.S. at 765, 114 S.Ct. 2516. It held that the inquiry for an injunction is whether it burdens “no more speech than necessary to serve a significant government interest.” Id. In other words, injunctions may not burden more speech than necessary to serve a significant government interest, whereas ordinances may not burden substantially more speech than necessary to serve a significant government interest. The difference between the Ward and Madsen standards is that Ward tolerates a degree of overinclusiveness whereas Madsen demands that a restriction burden no more speech than required.
Madsen did not, however, change the narrow-tailoring analysis that courts must conduct for ordinances. Even under the less rigorous intermediate scrutiny, “[g]ov-ernment may not regulate expression in such a manner that a substantial portion of the burden on speech does not serve to advance its goals.” Ward, 491 U.S. at 799, 109 S.Ct. 2746. Put differently, an under-inclusive speech restriction also violates the First Amendment, and in this respect, intermediate scrutiny and heightened scrutiny are the same. Thus, if a regulation burdens speech in such a way that it fails to advance the government’s goals, the regulation violates the First Amendment because it is not narrowly tailored under *567either the Madsen standard for injunctions or the Ward standard for ordinances.
In sum, a properly conducted narrow-tailoring analysis examines both whether the restriction is over- or underinclusive. The test for overinclusiveness is more stringent for an injunction than it is for an ordinance, but the test for underinclusiveness is identical for both types of speech restrictions. I turn now to an analysis of the facts of the case to explain why I would apply heightened scrutiny and why, even under intermediate scrutiny, Baltimore’s restrictions would fail a properly conducted narrow-tailoring analysis.
III.
A.
Baltimore’s restrictions were imposed neither via an injunction nor via an ordinance. However, it is clear to me that the City’s unadopted and secret speech restrictions more resemble an injunction than an ordinance.
In deciding to apply the less-stringent intermediate standard, both the district court and the majority found the generally applicable nature of the restrictions to be dispositive. This was error because general applicability is only one characteristic of ordinances. And nothing suggests that it is somehow a dispositive one. As the Supreme Court noted in Madsen, and as the majority opinion recognizes here, ordinances “represent a legislative choice” and carry fewer “risks of censorship and discriminatory application” than do injunctions. Madsen, 512 U.S. at 764, 114 S.Ct. 2516; ante at 552-53.
By stark contrast, Baltimore’s restrictions involved absolutely no legislative choice regarding the promotion of societal interests. Instead, they were simply made up by an unelected city lawyer. That unelected city employee wrote the restrictions without notice to the public and without the opportunity for public input that is generally required for ordinances passed pursuant to Maryland state law and pursuant to the Charter for the City of Baltimore. See Md. Code Ann., Local Gov’t § 9-105 (precluding Maryland counties from adopting acts, ordinances, or resolutions until ten days after a public hearing and requiring the publication of advance notice of the hearing and a summary of the proposed enactment in a newspaper of general circulation once each week for two successive weeks); Charter of Baltimore City art. Ill, § 14 (requiring legislative acts to “be by ordinance or resolution” and precluding ordinances from taking effect until after three separate readings). The fact that the public never knew — or even could have known — about the existence of the restrictions poses risks of censorship and discriminatory application that are even greater than those risks with injunctions. After all, when a court issues an injunction, it is clear who is bound and what conduct is proscribed. Moreover, the enjoined party’s ability to appeal provides an avenue of relief that is not available with a secret regulation.
The majority cites several cases in which other circuits have applied intermediate scrutiny to generally applicable, but una-dopted, restrictions on speech. Obviously, this Court is not bound by the decisions of other circuits. But to the extent that the cited cases are offered to guide our analysis, they are easily distinguishable from the facts here because they involve a legislative delegation of policymaking authority,3 a one-of-a-kind security situation,4 or *568obvious actual notice of the speech restriction.5
Here, by contrast, Ross had no idea of the existence of the restrictions until he deposed Early while taking discovery in this lawsuit — a full six months after his second arrest, and a year-and-a-half after his first arrest. Indeed, no evidence suggests that anyone beyond the drafter and a dozen or so recipients of the e-mail containing the restrictions — all of whom were people responsible for enforcing the restrictions — had any knowledge of their existence. There is also no evidence that the restrictions were enforced' only in emergency or otherwise unique security situations. Moreover, even if it could be argued that there was no time to officially adopt the restrictions initially, the City certainly had time to adopt and publicize the restrictions at some point during the five years between their creation and Ross’s first arrest.
Today’s ruling has troubling implications. After today, generally applicable, albeit secret, speech restrictions are afforded the same level of scrutiny in the Fourth Circuit as duly adopted ordinances. The potential for abuse is great.
Local governments will be able to develop and enforce speech-restrictive “Policies” without having to provide even a whisper of advance notice regarding the existence or content of the restrictions. In the event that the public becomes aware of the secret speech restrictions, there will be no electoral accountability for the unelected employees who developed the restrictions. And, perhaps most troubling, judicial relief will be more difficult to obtain for the person whose speech is restricted via enforcement of a secret “Policy” than it would be for a person whose speech is restricted via the enforcement of an injunction entered against him. The irony, of course, is that the person restricted by the injunction knows exactly what speech or conduct is proscribed, whereas the person restricted by the secret “Policy” does not — and cannot — know the same until it is too late.
For the foregoing reasons, I would apply the more rigorous narrow-tailoring analysis described in Madsen to Baltimore’s unadopted, speech restrictions in this case. I turn now to an explanation of why, even under intermediate scrutiny, I would hold that Baltimore’s restrictions are not narrowly tailored and, thus, fail.
B.
Though I believe that heightened scrutiny is the correct standard to apply in this case, even under the lower intermediate scrutiny standard applied by the majority, Baltimore’s restrictions fail the narrow-tailoring analysis. As I described above, a *569properly conducted analysis of a statute or ordinance under intermediate scrutiny requires courts to analyze, at a minimum, both (1) whether the restriction burdens substantially more speech than necessary; and (2) whether the restriction operates “in such a manner that a substantial portion of the burden on speech does not serve to advance [the government’s] goals.” Ward, 491 U.S. at 799, 109 S.Ct. 2746. If the restriction is overinclusive, it fails the first part of the narrow-tailoring analysis. If the restriction is underinclu-sive, it fails the second part of the narrow-tailoring analysis.6 See City of Ladue v. Gilleo, 512 U.S. 43, 51, 114 S.Ct. 2038, 129 L.Ed.2d 36 (1994) (discussing underinclu-siveness in the context of content neutrality and noting that “[wjhile surprising at first glance, the notion that a regulation of speech may be impermissibly underinclu-sive is firmly grounded in basic First Amendment principles.” (emphasis in original)).
Here, Baltimore’s goals are simple and legitimate; they are to ensure “freedom of movement on public streets and sidewalks[,]” Ross I, 758 F.Supp.2d at 322, and to ensure “pedestrian safetyf,]” Ross II, 899 F.Supp.2d at 425. The record contains ample evidence regarding the congestion on the sidewalks as circus-goers queue up to enter First Mariner Arena. Additionally, the restrictions do not ban leafleting altogether. Rather, they require leafleters to stand in specified locations, all of which are on the same block as the Arena. Therefore, I take no issue with the majority’s and district court’s conclusion that under Ward, the restrictions do not burden substantially more speech than necessary and are not overinclusive.
However, it seems plain to me that the restrictions are underinclusive because they “regulate expression in such a manner that a substantial portion of the burden on speech does not serve to advance [the government’s] goals.” Ward, 491 U.S. at 799, 109 S.Ct. 2746. It bears repeating both the burden imposed by the restrictions as well as Baltimore’s goals: the burden on speech is the prohibition against leafleting on the sidewalks adjacent to First Mariner Arena, and the goals are to maintain pedestrian flow and safety. The apparent mismatch between the burden on speech and Baltimore’s goals is caused by the secret nature of the restrictions.
The record contains still images and videos of Ross’s 2008 and 2009 arrests. It is evident that, as the district court found, Ross posed no threat to public safety and did not impede pedestrians while distributing leaflets. Ross II, 899 F.Supp.2d at 427. It is equally evident that the police caused pedestrian congestion while they attempted to explain the restrictions to the (understandably) confused protestors. The protestors’ reluctance to stop engaging in peacefully conducted protected speech is understandable because they had neither actual nor constructive knowledge of the restrictions on their First Amendment rights. Indeed, an arrest video shows one protestor calling her lawyer to ask whether the First Amendment protected her right to distribute leaflets on the public sidewalk. Unfortunately for that protestor, not even the world’s best First Amendment lawyer could have given a client proper advice in this situation because the only people who knew about Baltimore’s First Amendment restrictions were those few recipients of the city attorney’s e-mail.
*570As the majority correctly notes, it is the City’s burden to demonstrate that the speech restrictions meet the applicable level of scrutiny. See Bd. of Trustees of State Univ. of N.Y. v. Fox, 492 U.S. 469, 480, 109 S.Ct. 3028, 106 L.Ed.2d 388 (1989). The City has not met that burden here because the secret restrictions did not, in fact, advance the legitimate goals of maintaining pedestrian flow and safety. Accordingly, I would hold that Baltimore’s restrictions fail even Ward’s narrow-tailoring analysis and are, therefore, unconstitutional.
IV.
For the foregoing reasons, I would apply the heightened standard described in Madsen and conclude that Baltimore’s speech restrictions are unconstitutional because they burden more speech than necessary to achieve the City’s goals. But even under the less rigorous standard described in Ward, I would conclude that Baltimore’s restrictions are unconstitutional because they burden speech in a manner that does not advance the City’s goals. Accordingly, I respectfully dissent.
. The majority and the district court concluded that Baltimore’s "Policy” is content neutral and leaves open ample alternative channels of communication. Because I take issue only with (1) the level of scrutiny applied to the restrictions, and (2) the narrow-tailoring analysis that the district court and the majority conducted, I do not discuss either content neutrality or alternative channels of communication in this dissent.
. The majority notes that Ross “accepts intermediate scrutiny as the applicable standard of review and challenges only the district court's determination that, under that standard, the Policy is facially constitutional as a reasonable time, place, and manner restriction on speech.” Ante at 552. But the district court’s conclusions of law pertaining to the First Amendment claim at issue here are reviewable by this Court, regardless of whether Ross "accepts” those conclusions or not. See Kamen v. Kemper Fin. Servs., Inc., 500 U.S. 90, 99, 111 S.Ct. 1711, 114 L.Ed.2d 152 (1991) ("When an issue or claim is properly before the court, the court is not limited to the particular legal theories advanced by the parties, but rather retains the independent power to identify and apply the proper construction of governing law.”).
Additionally, although the majority states that "the parties have stipulated to a set of facts warranting the application of intermediate scrutiny,” ante at 555, the parties' stipulations simply cannot convert the e-mails sent by an unelected city lawyer into an ordinance. Because there is neither an ordinance nor an injunction, this case does not fit neatly into either the Ward or the Madsen analysis. It, therefore, remains our duty to ensure that the appropriate level of scrutiny is applied. See Marbury v. Madison, 1 Cranch 137, 5 U.S. 137, 177, 2 L.Ed. 60 (1803) (“It is emphatically the province and duty of the judicial department to say what the law is.”). Moreover, even if the parties attempted to stipulate to the standard of review, it should go without saying that "[w]e are not bound to accept, as controlling, stipulations as to questions of law.” Sanford’s Estate v. Comm’r Internal Revenue, 308 U.S. 39, 51, 60 S.Ct. 51, 84 L.Ed. 20 (1939). See also Swift & Co. v. Hocking Valley Ry. Co., 243 U.S. 281, 289-90, 37 S.Ct. 287, 61 L.Ed. 722 (1917) ("If the stipulation is to be treated as an agreement concerning the legal effect of admitted facts, it is obviously inoperative; since the court cannot be controlled by agreement of counsel on a subsidiary question of law.”).
. Saieg v. City of Dearborn, 641 F.3d 727, 730 (6th Cir.2011) (involving a no-leafleting policy at the Arab International Festival that was developed by the Dearborn police department *568pursuant to a resolution passed by the City Council that subjected the Festival to "the rules and regulations of the Police Department”).
. Marcavage v. City ofN.Y., 689 F.3d 98, 105 (2d Cir.2012) (involving the City of New York’s demonstration policy that pertained to the Republican National Convention at Madison Square Garden, which presented "extraordinary” security challenges in the wake of the 2001 terrorist attacks). The demonstrators in Marcavage also had actual notice of the restrictions.
. Faustin v. City and County of Denver, Colo., 423 F.3d 1192 (10th Cir.2005) (involving an unwritten total ban on signs and banners on highway overpasses); Int'l Caucus of Labor Comms. v. City of Montgomery, 111 F.3d 1548 (11th Cir.1997) (involving Montgomery’s ban on the placement of information tables on city sidewalks and landscaping strips); Potts v. City of Lafayette, Ind., 121 F.3d 1106 (7th Cir.1997) (involving the posted prohibition against entering a KKK rally with any item that could be used as a weapon).
. The majority and the district court simply fail to undertake this second part of the narrow-tailoring analysis.