Raphael GOLB, Petitioner–Appellant–Cross Appellee, v. The ATTORNEY GENERAL OF THE STATE OF NEW YORK, Respondent–Appellee–Cross Appellant,
In the academic debate about who wrote the Dead Sea Scrolls, petitioner–appellant Raphael Golb (“Golb”) was deeply committed to the side championed by his father Norman Golb. To further his father's position, Golb wrote several emails impersonating other scholars in an apparent effort to at least embarrass proponents of the rival view and, in some instances, harm their reputations. A Manhattan grand jury charged him with, inter alia, multiple counts of criminal impersonation in the second degree and forgery in the third degree. He was convicted of most of those charges and pursued appeals up through the New York Court of Appeals, which vacated several convictions and narrowed the scope of the criminal impersonation statute, but left most of the convictions intact.
In this federal habeas proceeding, the District Court for the Southern District of New York (Failla, J.) granted relief from two of Golb's convictions, and denied it as to the other 17.
Golb makes three arguments as to why his surviving convictions must be vacated. In opposition, the state prosecutors argue on the merits and rely on the deference owed to state courts under the Antiterrorism and Effective Death Penalty Act of 1996 (“AEDPA”), Pub. L. No. 104–132, 110 Stat. 1214 (1996).
• First, Golb invokes Shuttlesworth v. City of Birmingham, 382 U.S. 87 (1965), arguing that his impersonation convictions must be vacated if the jury might have relied on the impermissibly overbroad literal terms of the statute that the Court of Appeals subsequently narrowed. We owe no AEDPA deference on this question because the New York Court of Appeals did not answer it “on the merits.” 28 U.S.C. § 2254(d). We conclude that four of Golb's criminal impersonation convictions must be vacated under Shuttlesworth, but that five of them are so reliably supported by the evidence that they survive.
• Second, Golb argues that the criminal impersonation statute is facially unconstitutional. We do owe AEDPA deference as to this challenge, and we conclude that the New York courts' resolution of this issue was not so unreasonable as to require habeas relief.
• Third, Golb argues that the criminal forgery statute is unconstitutionally overbroad. We conclude that the statute, as interpreted by the trial court and the Court of Appeals, is so clearly overbroad as to be facially unconstitutional notwithstanding AEDPA deference. We narrow the statute to save it, and grant the habeas petition as to some (but not all) of the forgery convictions.
This case arises from a protracted academic debate about who wrote the Dead Sea Scrolls, ancient documents discovered in the 1940s and 1950s in a group of caves near Jerusalem. Most academics think that they were written by a Jewish sect called the Essenes who reportedly lived nearby (the “Essene Theory”), while the defendant's father Norman Golb, a professor at the University of Chicago, argues that they had many disparate authors and were hidden in caves when Roman armies attacked Jerusalem in 70 A.D. (the “Golb Theory”).
The defendant is not a Scrolls scholar, but he has devoted much time to advocating for his father's theory online. His anonymous and pseudonymous advocacy raises no issue here. But one of Golb's email tactics was to impersonate other Scrolls scholars. The surviving criminal counts in this case are based on ten such emails, in which Golb impersonated, variously, Frank Cross, Lawrence Schiffman, and Jonathan Seidel—all scholars of the Scrolls. We review the three impersonations in turn.
In mid–2008, Bart Ehrman, a proponent of the Essene Theory, was invited to lecture at a museum exhibit about the Scrolls. Golb wrote an anonymous blog post arguing that Ehrman should not have been invited and criticizing the Essene Theory. Golb also created an email address ([email protected]) which he used to impersonate Frank Cross, a well-known Scrolls scholar who has taught at Harvard and Wellesley. Using that email address, Golb sent emails to four scholars at the University of North Carolina—the host of the exhibit—which contained a link to his blog post and stated: “It looks like Bart [Ehrman] has gone and put his foot in his mouth again ․ I'm seeing this crop up everywhere on the web.” Joint App'x at 1066. The email was signed “Frank Cross.”
In the fall of 2008, the Jewish Museum in New York City invited Professor Lawrence Schiffman of New York University, also a proponent of the Essene Theory, to lecture at its exhibit on the Scrolls. Golb published an article using the pseudonym “Peter Kaufman” which accused Schiffman of plagiarizing some of Norman Golb's work. The same day, Golb created the email address “[email protected]” and sent the following message to four of Schiffman's graduate students, including a link to the “Peter Kaufman” article accusing Schiffman of plagiarism:
Miryam, Sara, Cory, Ariel,
Apparently, someone is intent on exposing a minor failing of mine that dates back almost fifteen years ago.
You are not to mention the name of the scholar in question to any of our students, and every effort must be made to prevent this article from coming to their attention. This is my career at stake. I hope you will all understand.
http://www.nowpublic.com/culture/plagiarism-and-de ad-sea-scrolls-did-nyu-professor-snitch-chicago-historians-work 1
Id. at 1130.
The next day, Golb sent another message from the [email protected] email address, this time to every member of Schiffman's department at NYU, again attaching a link to the “Peter Kaufman” accusatory article:
Apparently, someone is intent on exposing a minor failing of mine that dates back almost fifteen years ago.
Every effort must be made to prevent this article from coming to students' attention. This is my career at stake. I hope you will all understand.
Id. at 1139.
More emails from the Schiffman address went to the Dean and Provost of NYU, asking what the putative writer could do to “counter charges of plagiarism that have been raised against me,” and conceding “[i]t is true that I should have cited Dr. Golb's articles when using his arguments, and it is true that I misrepresented his ideas.” Id. at 1137. Golb's email to the Dean was then forwarded by Golb to NYU's student newspaper, with the added exhortation: “I must ask you not to publish a word about this.” Id. at 1136.
A few months later, a Scrolls exhibit opened at the Royal Ontario Museum. Golb created the email address “[email protected]” and used it to ask the Museum's Board of Trustees whether Norman Golb would be invited to lecture, while also calling a previous public statement by the Museum's curator “shockingly obscurantist.” Id. at 1039–41. The prosecutors contend that Golb's use of “seidel.jonathan” was an attempt to impersonate a Professor of Judaic Studies at the University of Oregon. Golb–as–Seidel then emailed the exhibit's curator to ask whether she would respond to Golb's father's critique of the exhibit. Similar emails from the same address were dispatched to dozens of Scrolls scholars; those, however, added an unfavorable critique of Norman Golb's work.
A Manhattan grand jury charged Golb in a 51–count indictment. The only convictions now before us, and the only counts we discuss, are nine for criminal impersonation in the second degree and ten for forgery in the third degree. (An explanation of which emails correspond to each count is in the margin.)2 The criminal impersonation statute provides:
A person is guilty of criminal impersonation in the second degree when he ․ [i]mpersonates another and does an act in such assumed character with intent to obtain a benefit or to injure or defraud another[.]
N.Y. Penal Law § 190.25(1). The forgery statute provides:
A person is guilty of forgery in the third degree when, with intent to defraud, deceive or injure another, he falsely makes, completes or alters a written instrument.
N.Y. Penal Law § 170.05.
At trial, Golb sought jury instructions defining several statutory terms, including “injure,” but the jury charge offered no definitions of “injure” or of “deceive.” The jury convicted Golb on 14 counts of criminal impersonation and ten counts of forgery, and the First Department affirmed all of those convictions. 960 N.Y.S.2d 66 (1st Dep't 2013) (“Golb I”).
The New York Court of Appeals entered an order accepting the appeal. 20 N.Y.3d 1099 (2013) (“Golb II”). Golb argued to that court that the absence of limiting jury instructions made the criminal impersonation and forgery statutes unconstitutionally overbroad. The majority agreed with Golb “that the statutory terms ‘injure’ and ‘benefit’ cannot be construed to apply to any injury or benefit, no matter how slight,” but they “conclude[d] that injury to reputation is within the ‘injury’ contemplated by Penal Law § 190.25.” 23 N.Y.3d 455, 465 (N.Y. 2014)(“Golb III”) (emphasis in original). Given that ruling, the majority found sufficient evidence that the emails impersonating Schiffman, Seidel, and Cross were sent “with intent to do real harm.” Id. at 466. Nine of the 14 surviving criminal impersonation convictions were therefore affirmed; the other five, based on the creation of the fake email accounts, or on other insubstantial email conduct, were vacated. Id. The ten forgery convictions were affirmed in a single paragraph that did not purport to narrow the statute. Id. at 468. Then–Chief Judge Lippman partially dissented on the ground that the First Amendment required vacatur of all the criminal impersonation and forgery convictions. Id. at 469–71 (Lippman, C.J., dissenting).
Golb's subsequent motion for reargument in the Court of Appeals contended that, under Shuttlesworth, 382 U.S. 87, he was entitled to a new trial on the criminal impersonation counts because he was tried using an overbroad version of the statute that the Court of Appeals had subsequently narrowed. The Court of Appeals denied the motion for reargument saying nothing more than “Motion for reargument denied.” 24 N.Y.3d 932 (N.Y. 2014) (“Golb IV”). On remand, the trial court rejected the Shuttlesworth argument on procedural grounds and resentenced Golb to two months in prison on the surviving convictions. The First Department also rejected the Shuttlesworth argument on procedural grounds, 5 N.Y.S.3d 46 (1st Dep't 2015) (“Golb V”), and the Court of Appeals denied leave to appeal, 26 N.Y.3d 929 (N.Y. 2015) (“Golb VI”).
Golb then filed this federal habeas petition. The district court granted relief as to two criminal impersonation counts, denied relief as to the seven other criminal impersonation counts, and denied relief as to the ten forgery counts. No. 15–cv–1709, 2016 WL 297726 (S.D.N.Y. Jan. 21, 2016) (“Golb VII”). The parties cross-appeal.
AEDPA instructs federal courts in habeas cases to give great deference to state court decisions when those decisions are “on the merits.” 28 U.S.C. § 2254(d). In such cases, we may grant a habeas petition only if the state court decision:
(1) resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States; or
(2) resulted in a decision that was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.
Id. The standard is deliberately “difficult to meet and highly deferential,” Contreras v. Artus, 778 F.3d 97, 106 (2d Cir. 2015) (internal punctuation omitted), and permits the grant of habeas “in cases where there is no possibility fairminded jurists could disagree that the state court's decision conflicts with [the Supreme] Court's precedents.” Harrington v. Richter, 562 U.S. 86, 102 (2011).
However, this deference arises only as to “any claim that was adjudicated on the merits in State court proceedings.” 28 U.S.C. § 2254(d). The question whether Golb's various claims were “adjudicated on the merits” in the New York courts requires considerable analysis. Two Supreme Court cases provide most of the governing law on whether state court decisions were decided on the merits. In Harrington, 562 U.S. 86, the Supreme Court said that a state court had adjudicated a claim “on the merits” notwithstanding that the appeal was rejected in a single-sentence summary order. As the Court explained, a state court need not give reasons for rejecting a claim:
When a federal claim has been presented to a state court and the state court has denied relief, it may be presumed that the state court adjudicated the claim on the merits in the absence of any indication or state-law procedural principles to the contrary. The presumption may be overcome when there is reason to think some other explanation for the state court's decision is more likely.
Id. at 99–100 (internal citations omitted). The presumption was not overcome in that case, and the Court did not specify what it might take to do so.
Two years later, in Johnson v. Williams, 568 U.S. 289 (2013), the Court considered the petition of a defendant who had argued parallel state and federal claims to the state court. Although the state court explicitly rejected only the state claim, the Supreme Court held that AEDPA deference was nevertheless owed on the federal claim because there were any number of reasons the state court might not have felt the need to explicitly mention a federal claim when rejecting a collateral attack or an appeal. Id. At 293. However, the Court made clear that the presumption that claims were adjudicated on the merits is “strong” but “not irrebuttable”: a footnote cited Baldwin v. Reese, 541 U.S. 27, 29 (2004) for the proposition that, “when a defendant does so little to raise his claim that he fails to fairly present it in each appropriate state court, the [ ] presumption is fully rebutted.” Id. at 302 n.3 (internal citation and quotation marks omitted).
Accordingly, when a state court denies a claim that was squarely presented, there is a strong presumption that the denial is “on the merits.” On the other hand, there is no such automatic presumption if the petitioner failed to squarely present the claim in state court. We might still conclude that a state court rejected an argument on the merits even if a petitioner did not raise it—a court might sua sponte raise and reject a claim, for example—but no presumption arises.
Ordinarily, such a rule does a petitioner no good. Exhaustion requirements and procedural default rules prevent a petitioner from securing de novo habeas review of federal claims not properly raised before the state court. In an unusual case, however, a petitioner may pass between obstacles—as in Golb's Shuttlesworth claim. Golb's direct appeal briefs failed to squarely present the Shuttlesworth claim to the Court of Appeals because his Shuttlesworth claim first came into being when that court issued its opinion narrowing the criminal impersonation statute in Golb III. By its nature, a Shuttlesworth claim will usually arise only “where a State Supreme Court narrows an unconstitutionally overbroad statute.” Osborne v. Ohio, 495 U.S. 103, 118 (1990). This is therefore the rare case in which there is both no failure of exhaustion and no presumption that the Court of Appeals rejected the federal claim on the merits.
The prosecutors advance two reasons why we nevertheless owe AEDPA deference on the Shuttlesworth claim. First, they argue that the Court of Appeals implicitly raised and rejected any Shuttlesworth claim on its own in Golb III by upholding certain criminal impersonation counts of conviction under its narrowed construction. We are not persuaded.
If Golb III rejected a Shuttlesworth challenge sua sponte, we would expect it to reference Shuttlesworth directly or indirectly. It does not. In limiting the scope of the statute, the Court of Appeals found “sufficient evidence to support the jury's finding that defendant's emails impersonating Schiffman, Seidel and Cross were more than a prank intended to cause temporary embarrassment or discomfiture, and that [Golb] acted with intent to do real harm.” Golb III, 23 N.Y.3d at 466. That wording does not respond to the Shuttlesworth question. Sufficiency under a curtailed version of the statute does not bear upon whether we are sure “that defendants are convicted under the statute as it is subsequently construed and not as it was originally written,” as Shuttlesworth requires. Osborne, 495 U.S. at 118. It is the difference between whether a reasonable jury could have convicted given the right instruction, and whether there is assurance beyond a reasonable doubt that a jury would have done so. See Chapman v. California, 386 U.S. 18, 24 (1967). We cannot conclude that Golb III rejected the Shuttlesworth claim “on the merits.”
The prosecutors also point out that Golb explicitly raised a Shuttlesworth claim in an unsuccessful motion for reargument based on the Court of Appeals' narrowed interpretation. According to the prosecutors, Golb IV's rejection of that motion in four words—“Motion for reargument denied”—is a decision “on the merits.” However, in the New York Court of Appeals (a court of discretionary jurisdiction), the “denial of a motion for leave to appeal is not equivalent to an affirmance and has no precedential value.” Calandra v. Rothwax, 65 N.Y.2d 897, 897 (N.Y. 1985). The rule is the same in the United States Supreme Court: “the denial of a writ of certiorari imports no expression of opinion upon the merits of the case.” Teague v. Lane, 489 U.S. 288, 296 (1989) (internal quotation marks and citation omitted). By analogy, the Court of Appeals' discretionary denial of rehearing was not a decision “on the merits,” and we accordingly owe no deference on the Shuttlesworth claim.
We reach a different conclusion as to Golb's facial challenges to the criminal impersonation and forgery statutes. Golb squarely presented his facial challenges to the Court of Appeals in his first set of briefs. Although that court's opinion in Golb III did not explicitly say why it rejected those challenges, “[w]hen a state court rejects a federal claim without expressly addressing that claim, a federal habeas court must presume that the federal claim was adjudicated on the merits[.]” Johnson, 568 U.S. at 301. That presumption can sometimes be rebutted, id., but Golb has not done nearly enough to rebut it here. When a state court rejects a constitutional claim without explanation, as here, our job is to “determine what arguments or theories supported or ․ could have supported [ ] the state court's decision,” and then “ask whether it is possible fairminded jurists could disagree that those arguments or theories are inconsistent with the holding in a prior decision of [the Supreme] Court.” Harrington, 562 U.S. at 102 (emphasis added).
Golb raises three challenges to his convictions. The first, based on Shuttlesworth, is not subject to AEDPA deference. It succeeds as to some of his criminal impersonation counts and fails as to others. The second, a facial challenge to the criminal impersonation statute, must overcome AEDPA deference. That challenge fails entirely. The third, a facial challenge to the forgery statute, must also overcome AEDPA deference. But because the statute as interpreted by the Court of Appeals is unconstitutionally overbroad, we narrow it and grant the habeas petition as to five of the ten forgery convictions.
A. The Shuttlesworth Claim
In Shuttlesworth, an African-American civil rights leader was convicted after a bench trial of violating a broad anti-loitering law. Shuttlesworth, 382 U.S. at 89. Two years later in a separate case, the Court of Appeals of Alabama construed the statute narrowly, thereby avoiding constitutional overbreadth. See Middlebrooks v. City of Birmingham, 170 So.2d 424 (Ala. Ct. App. 1964). Meanwhile, Shuttlesworth's ensuing appeal of his conviction worked its way to the United States Supreme Court. The Court held that the anti-loitering statute, as narrowed by the Alabama Court, was constitutional. But it also ruled that Shuttlesworth's own conviction must be vacated: “Because we are unable to say that the Alabama courts in this case did not judge the petitioner by an unconstitutional construction of the ordinance, the petitioner's conviction under [the loitering statute] cannot stand.” Shuttlesworth, 382 U.S. at 92. The Supreme Court has since clarified that holding: “Shuttlesworth, then, stands for the proposition that where a State Supreme Court narrows an unconstitutionally overbroad statute, the State must ensure that defendants are convicted under the statute as it is subsequently construed and not as it was originally written.” Osborne, 495 U.S. at 118. But how sure must courts be that a petitioner was “convicted under the statute as it is subsequently construed and not as it was originally written”? Id. In a habeas case, the standard of review is whether “the constitutional violation had a substantial and injurious effect or influence in determining the jury's verdict.” Jackson v. Conway, 763 F.3d 115, 140 (2d Cir. 2014) (internal quotation marks and citation omitted).
The prosecutors here argue that Shuttlesworth applies only when a State Supreme Court narrows a statute on constitutional grounds, whereas the New York Court of Appeals narrowed the statute on state law grounds (not because of the First Amendment). They misread both the Court of Appeals' opinion and Shuttlesworth itself. While the majority opinion did not explicitly cite the First Amendment in narrowing the criminal impersonation statute, Golb's arguments about the statute were based on the First Amendment, the dissent was about the First Amendment, and the majority did not explain its result on a ground of state law. The most reasonable way to read the opinion is as a narrowing construction to save the statute from violating the First Amendment. In any event, the Bill of Rights has more than one Amendment. Even if the Court of Appeals had narrowed the statute on a state law ground, a conviction under a statute that never criminalized the conduct of conviction would violate the Fifth Amendment right to Due Process.
The criminal impersonation statute criminalizes “[i]mpersonat[ing] another ․ with intent to obtain a benefit or to injure or defraud another.” N.Y. Penal Law § 190.25(1). The trial court, over objection, declined to define “injure” for the jury, allowing the jurors to convict for any conceivable interpretation of what it means to intend to “injure.” The Court of Appeals found that the term “cannot be construed to apply to any injury;” it is limited to intent to cause 1) “a tangible, pecuniary injury,” 2) “interfere [nce] with governmental operations,” or 3) harm to “reputation.” Golb III, 23 N.Y.3d at 465–66 (emphasis in original). The Shuttlesworth inquiry in this case is whether the difference in interpretation between the Court of Appeals and the trial court might reasonably have led to Golb's wrongful conviction on any of the criminal impersonation counts. Those counts can be grouped into four categories.
1. Five counts are based on emails sent under the name Lawrence Schiffman. In these emails, Golb-as-Schiffman confessed to plagiarism and misrepresentation, and asked recipients, including a student newspaper, to supress the facts of his misdeeds in order to protect him. These emails were clearly sent with the intent to damage Schiffman's reputation. Indeed, Golb sent the Schiffman emails at the same time that he emailed NYU administrators from a separate pseudonymous account requesting that they investigate the possible plagiarism—with the result that those administrators pursued the allegations of misconduct. As Golb acknowledged during his trial, plagiarism “is one of the more serious forms of unethical conduct anybody can engage in in the academic world.” Joint App'x at 643. The only alternate theory Golb puts forward is that the jury might have convicted on the theory that he simply intended to embarrass Schiffman; but that is too implausible to require a grant of habeas relief. See Jackson, 763 F.3d at 140. The district court denied habeas as to these convictions, and we affirm.
2. Two counts are based on emails sent under the name Jonathan Seidel, disparaging Golb's father. (The emails refer to Norman Golb and his views as “filth.” Joint App'x at 1045, 1064.) The defendant's intent in sending these emails is obscure: Golb–as–Seidel sent other emails seemingly relying on Seidel's good repute. We therefore lack sufficient confidence that the jury convicted based on the narrower meaning of “injure,” as opposed to a mere desire to antagonize. The district court granted the habeas petition as to these convictions, and we affirm its judgment. See Jackson, 763 F.3d at 142(stating that habeas relief is proper where the court has “grave doubt about whether a trial error ․ had substantial and injurious effect or influence in determining the jury's verdict ․” (internal quotation marks omitted)).
3. A single count is based on an email that Golb sent under the name Jonathan Seidel to the Royal Ontario Museum's Board of Trustees. The email asked whether the Museum intended to have Golb's father lecture at an exhibit about the Scrolls. The district court upheld this conviction on the theory that Golb intended to get a benefit—a speaking role for his father—rather than that Golb intended to injure. That is possible, but the email also attacks various individuals in a way likely to cause consternation, rather than any tangible, pecuniary, or reputational impact. The jury may well have impermissibly convicted Golb on this count under the theory that he intended to cause annoyance rather than a culpable injury. We therefore reverse the judgment of the district court and grant the habeas petition as to this conviction.
4. A single count is based on an email Golb sent under the name “Frank Cross” which said that “Bart has gone and put his foot in his mouth again.” Joint App'x at 1066. The district court upheld this conviction on the theory that it was intended to hurt Professor Bart Ehrman's reputation. But the email is so mild and puerile that it might have been intended to embarrass Ehrman without actual injury to his reputation, and at no point in the trial did the prosecutors argue that it was intended to hurt Ehrman. In light of our substantial concern that the jury impermissibly convicted Golb on this count based on the overbroad interpretation of injure, we therefore reverse the judgment of the district court and grant the habeas petition as to this conviction.
B. Golb's Facial Challenge to the Criminal Impersonation Statute
We owe AEDPA deference on Golb's facial challenge to the constitutionality of the criminal impersonation statute, see supra at 18–19, an issue we reach because five criminal impersonation convictions survive Shuttlesworth review. Golb argues that the statute is unconstitutional both because it is overbroad and because the narrowing of it—by including damage to reputation as an “injury”—makes it unconstitutionally vague.
We reject the overbreadth challenge. A statute is facially overbroad
if it prohibits a substantial amount of protected speech. The doctrine seeks to strike a balance between competing social costs. On the one hand, the threat of enforcement of an overbroad law deters people from engaging in constitutionally protected speech, inhibiting the free exchange of ideas. On the other hand, invalidating a law that in some of its applications is perfectly constitutional—particularly a law directed at conduct so antisocial that it has been made criminal—has obvious harmful effects. In order to maintain an appropriate balance, we have vigorously enforced the requirement that a statute's overbreadth be substantial, not only in an absolute sense, but also relative to the statute's plainly legitimate sweep. Invalidation for overbreadth is strong medicine that is not to be casually employed.
United States v. Williams, 553 U.S. 285, 292–93 (2008) (internal citations and quotation marks omitted; emphasis in original).
Golb implicitly concedes that the criminal impersonation statute constitutionally criminalizes some core conduct; there is no doubt that the state may properly forbid (for example) impersonating a pawnshop customer in order to redeem an object of value. But he argues that there are categories that the statute improperly criminalizes: in particular, some types of satire and parody. Specifically, he argues that the statute is facially overbroad because it would criminalize hypothetical parodies that the Constitution protects.
The First Amendment protects parody, see Hustler Magazine, Inc. v. Falwell, 485 U.S. 46 (1988), but Golb misunderstands the genre. While it is true that a parody enjoys First Amendment protection notwithstanding that not everybody will get the joke, it is also true that parody depends on somebody getting the joke; parody succeeds only by its recognition as parody. An author who intends to fool everyone may be pulling a prank or perpetrating a hoax, but the result is not a parody. Parody thus differs from “impersonat[ion]” as the term is used in the criminal impersonation statute.
And even if we did accept Golb's understanding of parody, an overbreadth challenge succeeds only if a statute “prohibits a substantial amount of protected speech.” Williams, 553 U.S. at 292 (emphasis added). The overbreadth must “be substantial, not only in an absolute sense, but also relative to the statute's plainly legitimate sweep.” Id. (emphasis in original). The criminal impersonation statute has substantial legitimate sweep, and Golb's argument is limited to parody with the intent to injure. That overbreadth would be insufficient to invalidate the statute; in any event, a parodist caught up in prosecution could bring an as-applied challenge to conviction. Finally, we owe AEDPA deference on the overbreadth challenge, and it is certainly not the case that the statute is clearly unconstitutionally overbroad.3
Nor is it unconstitutionally vague. “[T]he void-for-vagueness doctrine requires that a penal statute define the criminal offense with sufficient definiteness that ordinary people can understand what conduct is prohibited and in a manner that does not encourage arbitrary and discriminatory enforcement.” Kolender v. Lawson, 461 U.S. 352, 357 (1983). A fairminded jurist could find that the Court of Appeals' interpretation is sufficiently well-defined to meet both elements, and that is enough to compel AEDPA deference. See Harrington, 562 U.S. at 102.
The statute criminalizes impersonation with the intent to harm another's reputation. Reputation is not so elusive a concept that the ordinary person would fail to understand it, particularly in a professional context (such as this one) where the potential injury is evident. Nor is the statute so ill-defined that every fairminded jurist would find that it allows prosecutors to arbitrarily determine whether or not individuals have committed criminal impersonation. The criminal impersonation statute criminalizes a large swath of conduct, but it is fairly clear what conduct falls within. A fairminded jurist could easily conclude that the prosecutorial power it affords is broad but not arbitrary.4
Because the criminal impersonation statute is not unconstitutionally vague or overbroad, the five convictions which survived our Shuttlesworth review stand.
C. Golb's Facial Challenge to the Forgery Statute
We owe AEDPA deference on Golb's facial challenge to the constitutionality of the forgery statute, which criminalizes “falsely mak[ing], complet[ing] or alter[ing] a written instrument” when done “with intent to defraud, deceive or injure another.” N.Y. Penal Law § 170.05. The Court of Appeals affirmed those counts on the ground that Golb “deceived people,” Golb III, 23 N.Y.3d at 468, and the holding treats the phrase “intent to [ ] deceive” to mean an intent to make another believe something that is false, regardless of whether there was intent to cause any possible harm. Id. Such a boundless interpretation renders the statute so overbroad that any fairminded jurist would find it unconstitutional. Pseudonymous product reviews would be criminalized, as would the use of false names by corporate or governmental whistleblowers to avoid detection and retaliation. Even the use of a pen-name is sometimes a benign deception. George Sand, George Eliot, and Currer Bell were noms de plume used to deceive readers into believing that they were men; under the Court of Appeals' interpretation, the authors could all be prosecuted.
Clearly established federal law is to the contrary: “[g]reat works of literature have frequently been produced by authors writing under assumed names. ․ [A]n author's decision to remain anonymous, like other decisions concerning omissions or additions to the content of a publication, is an aspect of the freedom of speech protected by the First Amendment.” McIntyre v. Ohio Elections Comm'n, 514 U.S. 334, 341–42 (1995). Because the Court of Appeals' interpretation of the statute violated “clearly established Federal law, as determined by the Supreme Court of the United States,”—i.e., in McIntyre—that interpretation is unconstitutional despite AEDPA deference. 28 U.S.C. § 2254(d)(1).
A statute, however, is not unconstitutional on its face because a single possible interpretation of it is unconstitutional. Another reading of this statute is more plausible, and it has the virtue of being constitutional. New York's forgery law criminalizes acting “with intent to defraud, deceive or injure.” The verb “deceive” has several definitions. Its broadest, most common meaning, which the Court of Appeals adopted, is to “cause to believe what is false.” “Deceive,” The New Shorter Oxford English Dictionary (1993). Another definition (albeit more narrow and more rare) is to “cheat, defraud, [or] deprive of by deception.” Id. Three rules of statutory interpretation counsel adoption of that narrower, less common definition. First, “if an otherwise acceptable construction of a statute would raise serious constitutional problems, and [ ] an alternative interpretation of the statute is fairly possible, we are obligated to construe the statute to avoid such problems.” I.N.S. v. St. Cyr, 533 U.S. 289, 299–300 (2001) (internal citations and quotation marks omitted). A broader reading of “deceive” makes the statute unconstitutional, while the narrower reading saves it. Second, in the statute, “deceive” appears in series with “defraud” and “injure.” Under the interpretative canon noscitur a sociis, the word “deceive” is deemed to share relevant characteristics with “defraud” and “injure,” both of which entail harm. Third, “ambiguity concerning the ambit of criminal statutes should be resolved in favor of lenity.” Rewis v. United States, 401 U.S. 808, 812 (1971).
The Court of Appeals affirmed Golb's forgery convictions on the basis of deception, Golb III, 23 N.Y.3d at 468, but the jury instructions on forgery failed to define the term. An uninstructed jury would likely understand “deceive” in its commonest definition: to “cause to believe what is false,” without the stipulation of harm. That unconstitutionality requires scrutiny of the forgery convictions. Five of the forgery convictions were based on emails falsely attributed to Schiffman, and those convictions survive because the evidence so clearly supports Golb's intent to deceive and cause injury. As to the other five forgery convictions, however, it is so likely that the jury applied the law in an unconstitutional way to convict Golb that habeas relief is required.
Accordingly, we affirm the district court as to counts 8, 11, 14, 17, and 20; and we reverse (and grant the habeas petition) as to counts 27, 31, 35, 39, and 46.
For the foregoing reasons, the order of the district court is AFFIRMED IN PART and REVERSED IN PART.
1. This link leads to the article Golb wrote under the pseudonym “Peter Kaufman.”
2. Count 7 (criminal impersonation) and count 8 (forgery) are based on an August 4, 2008 email impersonating Schiffman which can be found at Joint App'x 1130. Count 10 (impersonation) and count 11 (forgery) are based on an August 5, 2008 email impersonating Schiffman which can be found at Joint App'x 1139. Count 13 (impersonation) and count 14 (forgery) are based on a different August 5, 2008 email impersonating Schiffman which can be found at Joint App'x 1137. Count 16 (impersonation) and count 17 (forgery) are based on yet another August 5, 2008 email impersonating Schiffman which can be found at Joint App'x 1138. Count 19 (impersonation) and count 20 (forgery) are based on an August 6, 2008 email impersonating Schiffman which can be found at Joint App'x 1142. Count 25 (impersonation) and count 27 (forgery) are based on a November 22, 2008 email impersonating Seidel which can be found at Joint App'x 1039–41. Count 31 (forgery) is based on a November 24, 2008 email impersonating Seidel which can be found at Joint App'x 1044. Count 33 (impersonation) and count 35 (forgery) are based on another November 24, 2008 email impersonating Seidel which can be found at Joint App'x 1045. Count 37 (impersonation) and count 39 (forgery) are based on a December 6, 2008 email impersonating Seidel which can be found at Joint App'x 1064. Count 46 (impersonation) and count 47 (forgery) are based on a July 20, 2008 email impersonating Cross which can be found at Joint App'x 1066.
3. Insofar as Golb argues that the statute was overbroad as applied to him, we agree with the Court of Appeals that the statute was not unconstitutionally applied with regard to the remaining criminal impersonation convictions related to the Schiffman emails. The evidence of Golb's intent to harm Schiffman's reputation was overwhelming.
4. Golb also cites Skilling v. United States, 561 U.S. 358 (2010), for the proposition that any statute criminalizing an intangible harm is automatically vague. He misreads the case. Skilling actually relied on the Supreme Court's assessment that the “honest-services decisions preceding McNally [v. United States, 483 U.S. 350 (1987)],” were “not models of clarity or consistency.” Skilling, 561 U.S. at 405 (emphasis added). In Skilling, the Court hewed closely to those cases' core interpretations of the statute at issue so as to avoid constitutional questions.
DENNIS JACOBS, Circuit Judge: