dissenting from the denial of rehearing en banc.
The panel’s decision in this case will restrict speech on a matter of vital public concern. The columns at issue urged government action on a question of grave national import and life-or-death conse*254quence. It is unclear, to say the least, that Virginia law would ever find these columns to be defamatory, and the panel pushes state law in a direction that not only portends liability for valuable public commentary but aggravates, rather than alleviates, the constitutional tensions inherent in the defamation field.
It is worth remembering the context in which the columns at issue were published. In the aftermath of the September 11 attacks, the nation was alerted to the fact that someone was sending letters laced with anthrax through the mails. The letters were not simply directed at public officials but apparently at private individuals as well. Those who handled mail on a regular basis were concerned for their safety, and even ordinary residents were advised to take special precautions when opening their mail. At least five people died from anthrax exposure. There was, in addition, worry that law enforcement was ineffectual in locating the source of the anthrax production and distribution. In other words, both the problem and the steps necessary to resolve it were matters of public, indeed national, concern.
In the spring and summer of 2002, Nicholas Kristof, a columnist for the New York Times, published a series of pieces urging the federal government to step up its investigative and preventive efforts and take control of the situation. The columns appeared in the Times on five different dates: May 24, July 2, July 12, July 19, and August 13. They included descriptions first of a “Mr. Z,” later identified in one column as Dr. Steven Hatfill, whom circumstantial evidence suggested was a person of interest in the anthrax investigations. In plaintiffs view, these discussions contained serious factual errors and unfairly implicated him as the perpetrator. In defendant’s view, the columns repeatedly disavowed any such conclusion and urged that the government conduct a thorough inquiry that would either inculpate plaintiff or exonerate him.
As a result of these events, this action for defamation and intentional infliction of emotional distress ensued. The district court dismissed it because, inter alia, it found that the columns were not actionable under state law. A divided panel of this court reversed. Hatfill v. New York Times, 416 F.3d 320 (2005). The majority found that the columns were fairly read as accusing plaintiff of the anthrax murders, and therefore held that defendant could face state tort liability for publishing them. Id. at 337. Judge Niemeyer dissented, “find[ing] nothing in the letter or spirit of the columns that amount[ed] to such an accusation.” Id.
The panel’s opinion appeared to downplay the importance of any First Amendment concerns in two important ways. First, it noted that the case was before the court on a motion to dismiss under Federal Rule 12(b)(6). E.g., id. at 334. Second, it characterized the case as wholly one of state law. E.g., id. at 330. I shall address in turn why these two rationales were not proper bases for reversing the district court. And I shall lastly take up what I regard as the deeply unfortunate free speech implications of the panel’s ruling.
I.
The panel viewed its inquiry as limited to consideration of whether plaintiff had “adequately pled the elements of his claims under Virginia law.” 416 F.3d at 324. This, I think, is much too simple. The procedural posture of this case—a motion to dismiss for failure to state a claim under state law—should not obscure its constitutional importance.
A defamation case does not putter along as a state law case in its earliest stages, *255only to suddenly acquire First Amendment implications upon the tender of an affirmative defense. Defamation actions by their very nature seek to punish past speech and raise the specter of chilling future speech. For this reason, the Supreme Court has waged a lengthy “struggle to define the proper accommodation between the law of defamation and the freedoms of speech and press.” Phila. Newspapers, Inc. v. Hepps, 475 U.S. 767, 768, 106 S.Ct. 1558, 89 L.Ed.2d 783 (1986) (internal quotation marks and alterations omitted). In New York Times Co. v. Sullivan, 376 U.S. 254, 84 S.Ct. 710, 11 L.Ed.2d 686 (1964), and its progeny, the Court infused the state common law of defamation with a constitutional dimension, creating not only procedural protections but also limiting “the type of speech which may be the subject of state defamation actions.” Milkovich v. Lorain Journal Co., 497 U.S. 1, 16, 110 S.Ct. 2695, 111 L.Ed.2d 1 (1990) (emphasis omitted).
The Constitution provides states some latitude to protect a citizen’s good name and reputation, but that leeway is limited by the Founders’ injunction that law shall not “abridg[e] the freedom of speech, or of the press.” U.S. Const. amend. I. It makes little sense to acknowledge the special sensitivity of speech to defamation actions and then to say that speech interests matter little or not at all because of the procedural posture of the action. While a heightened pleading standard in defamation cases may be inappropriate, see, e.g., Swierkiewicz v. Sorema N.A., 534 U.S. 506, 514-15, 122 S.Ct. 992, 152 L.Ed.2d 1 (2002) (rejecting a heightened standard in employment discrimination cases), there is no reason why an action of this kind cannot frequently be resolved on a motion to dismiss. The most critical part of the record—the speech itself—is available pri- or to any discovery. Whether the statements are defamatory as a matter of law will therefore be ripe for decision.
The panel majority in this case has read Virginia law aggressively to permit a wide array of defamation suits against news organizations. A court that reads state law so expansively when deciding a motion to dismiss creates a “threat ... of pecuniary liability” that “may impair the unfettered exercise of ... First Amendment freedoms.” Greenbelt Coop. Publishing Ass’n v. Bresler, 398 U.S. 6, 12, 90 S.Ct. 1537, 26 L.Ed.2d 6 (1970).
Even if liability is defeated down the road, the damage has been done. The defendant in this case may well possess the resources necessary for protracted litigation, but smaller dailies and weeklies in our circuit most assuredly do not. The prospect of legal bills, court appearances, and settlement conferences means that all but the most fearless will pull their punches even where robust comment might check the worst impulses of government and serve the community well. To allow litigation to impose large costs will dull democracy at the local level, because the monetary impacts of litigation for all but the largest media organizations will prove unacceptably high. Federal courts must maintain fidelity to state law, but here a federal court sitting in diversity has created constitutional problems by pushing state defamation law in new directions without unambiguous direction from the state courts or legislature. This has compounded the sort of constitutional tensions which the exercise of judicial restraint has long sought to avoid. See Ashwander v. Tenn. Valley Auth., 297 U.S. 288, 348, 56 S.Ct. 466, 80 L.Ed. 688 (1936) (Brandéis, J., concurring).
II.
In allowing actions for defamation per se and intentional infliction of emotional dis*256tress to proceed, the panel’s decision stretches the bounds of liability beyond where the state courts have placed them.
A.
The first two counts of plaintiffs complaint state claims for common law defamation. In the first, plaintiff alleges that Kristofs columns were defamatory because they amounted to “identification of [plaintiff] with the anthrax mailer.” Compl. ¶ 42. In the second, he alleges that various individual statements contained within the columns are independently defamatory, because even in isolation they “would tend to incriminate [plaintiff].” Id. ¶ 43. The panel agreed with plaintiffs theory, finding that the columns and most of the statements “imputed to [plaintiff] the commission of a crime involving moral turpitude,” 416 F.3d at 334, 335, and were therefore defamation per se under Virginia law, see Food Lion, Inc. v. Melton, 250 Va. 144, 458 S.E.2d 580, 584 (1995).
That conclusion is both factually and legally incorrect. From a factual standpoint, even giving plaintiff the full benefit of Federal Rule 12(b)(6), the panel misinterprets the language of the statements and columns. The Supreme Court of Virginia instructs us that “[a] statement imputes the commission of a crime when it refers to matters that would naturally and presumably be understood by those [reading] them as charging a crime.” Id. In other words, the statements and columns are not defamatory per se unless they can fairly be read to actually accuse plaintiff of being a criminal. I agree with both Judge Niemeyer and with the district court that the statements and columns at issue here make no such accusation.
The columns do present a list of circumstantial evidence against plaintiff, and plaintiff claims that several pieces of this evidence are false. But none of the supposedly false statements identified by plaintiff is sufficient to suggest that plaintiff is the anthrax murderer. See 416 F.3d at 338 (Niemeyer, J., dissenting) (“Reporting suspicion of criminal conduct—even elaborately and sometimes inaccurately— does not amount to an accusation of criminal conduct as necessary to support Dr. Hatfill’s claim.”). Stating that plaintiff had access to an isolated residence where visitors received anthrax vaccinations, or reporting that his own vaccinations were up to date, is a far cry from suggesting that he killed innocents by sending the anthrax virus through the mail. And where it is directly preceded by reminders that plaintiff should be presumed innocent and that “[t]here is not a shred of traditional physical evidence linking him to the attacks,” the claim that he failed polygraph tests of an unspecified nature should not cause a reasonable reader to infer that he is in fact guilty. Finally, placing plaintiff among a “handful of individuals who had the ability, access and motive to send the anthrax” suggests only that he ought to be one of multiple suspects; it hardly points the finger at plaintiff alone.
Viewed as a whole, the columns do not pin guilt on plaintiff, but instead urge the investigation of an undeniable public threat. The columns involve pointed criticism of the Executive branch, primarily targeting the FBI for its purported “lackadaisical ineptitude in pursuing the anthrax killer.” The first column opines that “[o]ne of the first steps we can take to reduce our vulnerability [to a biological attack] is to light a fire under the F.B.I. in its investigation,” and the remaining columns seek to light that fire. They recount alleged facts suggesting why plaintiff is a suspect; they do so, however, not to cast him as the murderer, but to illustrate the unfortunate effects of government inaction *257not only upon an anxious nation but also upon a man who finds himself at the center of a dawdling investigation.
The columns expressly avoid premature accusation by repeatedly reminding readers that the burden of the government’s failure falls not only upon the public, who seek the safety of identifying the true culprit, but also upon plaintiff, who deserves an end to the “unseemly limbo” of being a suspect. Kristof makes plain his belief that “the presumption of innocence has already been maimed since 9/11 for foreign Muslims, and it should not be similarly cheapened with respect to [plaintiff].” Thus, “[i]t must be a genuine assumption that [plaintiff] is an innocent man caught in a nightmare,” particularly considering the complete lack of physical evidence against him. The columns report further that plaintiff “denies any wrongdoing, and his friends are heartsick at suspicions directed against a man they regard as a patriot.” While one reason to pursue the investigation vigorously is to arrest plaintiff if further evidence shows him to be the culprit, another major reason is to “exculpate him and remove [the] cloud of suspicion” if it does not. In view of their overall purpose and in light of their repeated disclaimers, these columns cannot “naturally and presumably be understood ... as charging a crime.” Food Lion, 458 S.E.2d at 584.
From a legal standpoint, the panel majority extends liability for defamation per se beyond the two Virginia cases on which it relies. The statements that the Virginia Supreme Court found to be defamatory in Carwile v. Richmond Newspapers, Inc., 196 Va. 1, 82 S.E.2d 588 (1954), and Schnupp v. Smith, 249 Va. 353, 457 S.E.2d 42 (1995), were far more accusatory than those at issue here.
The defendant in Carwile published an article stating that public officials had the discretion to seek the plaintiffs disbarment, necessarily implying that the plaintiff “could and should be subjected to disbarment proceedings.” 82 S.E.2d at 592; see also id. at 589-90. In contrast, the columns here did not suggest that the FBI “could and should” indict plaintiff. Indeed, the problem they identify is the lack of sufficient information with which to either charge or exonerate him.
The defamation in Schnupp involved a description of the plaintiff committing the alleged criminal offense itself, stating that he acted as the driver for someone engaged in what could reasonably be inferred as a drug purchase. 457 S.E.2d at 45; see also id. at 46. The columns here, however, do not describe plaintiff taking direct actions—e.g., placing mysterious powder into envelopes—that might reasonably be inferred as criminal in and of themselves. In fact, their central thrust is the government’s failure to uncover any such evidence.
Moreover, in neither Carwile nor Schupp was the allegedly defamatory language tempered by express disclaimers of prejudgment like those that pervade the columns at issue in this case. Taken together, these differences suggest that the columns here fall outside the previously defined scope of defamation per se in Virginia. At a minimum, these precedents do not provide the type of unambiguous support necessary to impose liability under constitutionally tense circumstances.
B.
The third count of plaintiffs complaint states a cause of action for intentional infliction of emotional distress. He alleges that the “intentional public identification of [plaintiff] with the anthrax murders ... was unconscionable, malicious, intentional, *258and calculated to inflict grievous emotional distress.” Compl. ¶ 44.
In permitting this claim to proceed, the panel again commits errors of both fact and law. The panel errs factually when it concludes that “[accepting [plaintiffs] allegations as true, [defendant] intentionally published false charges accusing him of being responsible for anthrax mailings that resulted in five deaths.” 416 F.3d at 336. As the above analysis of the columns demonstrates, defendant did no such thing. The columns neither “accuse” plaintiff of being the anthrax killer nor level “charges” of murder against him.
The panel errs legally when it concludes that defendant’s actions meet the narrow conditions under which an action for intentional infliction of emotional distress may lie. That tort is “not favored” in Virginia law. See Ruth v. Fletcher, 237 Va. 366, 377 S.E.2d 412, 415 (1989) (internal quotation marks omitted). It applies only in the most extreme of circumstances: where a defendant’s conduct is “so outrageous in character, and so extreme in degree, as to go beyond all possible bounds of decency, and to be regarded as atrocious, and utterly intolerable in a civilized community.” Russo v. White, 241 Va. 23, 400 S.E.2d 160, 162 (1991) (internal quotation marks omitted).
I am quite at a loss to see how publication of these columns “go[es] beyond all possible bounds of decency.” They report on a matter of unquestioned public interest with urgent national security implications. The First Amendment expressly specifies that the “civilized community” in which we live is one that encourages public commentary of this type. Even assuming the columns contain the asserted factual errors, their publication is neither “intolerable” nor “atrocious.” The panel offers no decision from Virginia or any other state that holds a news report on a subject of unquestioned public interest to be an intentional infliction of emotional distress. Its action is unprecedented.
III.
The consequences of this decision for the First Amendment run deep. If one purpose of public commentary is to assess the functioning of government, these columns were surely in that vein. In fact, the anthrax mailings and the government’s response to them lie at the heart of legitimate public inquiry. It is true that Kris-tofs columns conveyed an unmistakable sense of urgency, but it is often the job of a commentator to prod officials to take action. It is also true that Kristof repeated his calls for an investigation, but multiple iterations are sometimes necessary to get the attention of anyone in charge.
Surely it is open to a columnist to comment—and comment vigorously—on whether law enforcement is properly carrying out those responsibilities with which it is charged. Bureaucracies have been known to be both sluggish and inept, and it is not only the prerogative, but the duty, of news organizations to insist that those agencies get off the dime. The costs of inaction can be, as here, to leave someone in the prolonged limbo of suspicion and to jeopardize public safety by leaving the wrongdoer—whomever he may be—at large.
The perils of inaction and of overzealous action on the part of law enforcement are alike proper subjects for a free press. In fact, if there is any area that merits public scrutiny, it would seem to be the workings of our criminal justice system. Both in its investigative and judicial aspects, that system is capable of serious malfunction.
Our criminal justice system does not suddenly reveal itself the day a verdict is *259announced. Rather, it undergoes public scrutiny as it unfolds. It is often difficult, if not impossible, to cover the long continuum of justice in John Doe fashion without the use of a suspect’s identity or name, as daily media reports on criminal activity make clear. It is the obligation of news media not to fasten guilt upon those who have not been tried, but it is equally their obligation not to deprive the public of a meaningful report. These columns were hard-hitting, to be sure, but they did not forsake the essential balance that our law requires.
In short, I believe that defendant was simply doing its job. It is a job that the Constitution protects, and I would not construe gray areas of Virginia law to punish it and deter others from performing it. It is tempting, I recognize, to view the press’s assertions of its freedoms as something of a self-interested wail. But before succumbing too fully to this impulse, we might ask who else will do the job of calling bureaucratic judgments to account. The public’s right to know in this case was not a matter of voyeurism, titillation, or idle curiosity. The bioterrorism presaged by these anthrax mailings was no small matter, and it may one day pose a threat on a very large scale. Let us hope that on that day, reluctance to take issue with authority has not become our norm. “[T]he pall of fear and timidity imposed upon those who would give voice to public criticism is an atmosphere in which the First Amendment freedoms cannot survive.” Sullivan, 376 U.S. at 278, 84 S.Ct. 710.
Judge MICHAEL and Judge KING join me in this dissent.