Harvard Law Review Harvard Law Review Harvard Law Review

First Amendment: Speech

Elonis v. United States

Federal law has criminalized communicating threats in interstate commerce since 1939.1×1. See 18 U.S.C. § 875(c) (2012); United States v. Jeffries, 692 F.3d 473, 484 (6th Cir. 2012) (Sutton, J., dubitante) (reviewing history of federal threats statute). Even though this prohibition is in tension with principles of free speech,2×2. See G. Robert Blakey & Brian J. Murray, Threats, Free Speech, and the Jurisprudence of the Federal Criminal Law, 2002 BYU L. Rev. 829, 874–921 (describing “true threats” doctrine as “a potentially devastating legal sword to draw and wield for those who seek to silence and delegitimize speech with which they disagree,” id. at 875–76). the Supreme Court has held that “true threats” are not protected by the First Amendment.3×3. See R.A.V. v. City of St. Paul, 505 U.S. 377, 388 (1992) (citing Watts v. United States, 394 U.S. 705, 707 (1969)). In Virginia v. Black, 538 U.S. 343 (2003), the Supreme Court identified “true threats” as “those statements where the speaker means to communicate a serious expression of an intent to commit an act of unlawful violence to a particular individual or group of individuals.” Id. at 359. Yet the Court has not fully defined this category of unprotected speech.4×4. For an overview of the Court’s sparse “true threats” jurisprudence, see Paul T. Crane, Note, “True Threats” and the Issue of Intent, 92 Va. L. Rev. 1225, 1229–34 (2006); Steven G. Gey, A Few Questions About Cross Burning, Intimidation, and Free Speech, 80 Notre Dame L. Rev. 1287, 1326–31 (2005). Two central and unresolved questions are first whether a defendant can be convicted under the federal threats statute absent proof that he subjectively intended to threaten anyone, and second, if the statute itself does not require this evidence, whether the First Amendment does. In recent decades, a majority of lower courts have coalesced around a rule permitting criminal liability where a “reasonable person” would understand the defendant’s words as a threat.5×5. For a comprehensive discussion of the lower courts’ approaches to “true threats,” including a description of the subtle differences in how the “reasonableness” rules have been applied, see Crane, supra note 4, at 1243–52 (writing post-Black); and Blakey & Murray, supra note 2, at 937–1010 (writing pre-Black). Almost no circuit altered its rule in light of Black’s definition of “true threat.” See, e.g., United States v. White, 670 F.3d 498, 508 (4th Cir. 2012) (rejecting specific intent requirement post-Black and noting circuit agreement); United States v. Mabie, 663 F.3d 322, 332–33 (8th Cir. 2011) (collecting cases). But see United States v. Bagdasarian, 652 F.3d 1113, 1117 (9th Cir. 2011) (concluding Black implies subjective intent standard). Last Term, in Elonis v. United States,6×6. 135 S. Ct. 2001 (2015). the Supreme Court disagreed with this standard and held that a conviction under 18 U.S.C. § 875(c) may not be based solely on a reasonable person’s interpretation of the defendant’s words.7×7. Id. at 2012–13. The Court reversed the defendant’s conviction on narrow statutory grounds rooted in principles of mens rea and did not reference the constitutional “true threats” doctrine. The majority, however, left undecided the minimum mental state required for criminal liability. As a result, lower courts are left to answer both questions originally presented in Elonis. Certain parallels between obscenity and threats, hinted at in Elonis, suggest one approach to this task.

In May 2010, Anthony Douglas Elonis’s wife Tara left him.8×8. United States v. Elonis, 730 F.3d 321, 324 (3d Cir. 2013). In the months that followed, Elonis used his Facebook account to publicize violent rap lyrics that he wrote under the rap pseudonym “Tone Dougie.”9×9. Elonis, 135 S. Ct. at 2005. Some of Elonis’s posts contained alarming language directed at his newly estranged wife.10×10. Elonis, 730 F.3d at 324. For example: “There’s one way to love you but a thousand ways to kill you. I’m not going to rest until your body is a mess, soaked in blood and dying from all the little cuts.”11×11. Id. Elonis included disclaimers in his posts such as “I ain’t a legitimate threat” and links to the Wikipedia article explaining “Freedom of Speech.” Brief for the Petitioner at 7–8, Elonis, 135 S. Ct. 2001 (No. 13-983). Elonis did not send messages directly to Tara, but he made sure that she would see them. See Brief for the United States at 3, Elonis, 135 S. Ct. 2001 (No. 13-983). Things escalated in the months that followed; by November, Tara was frightened enough by Elonis that she obtained a Protection From Abuse order.12×12. Brief for the United States, supra note 11, at 4. But the messages did not stop. In fact, Elonis responded: “Fold up your PFA and put it in your pocket[.] Is it thick enough to stop a bullet?”13×13. Elonis, 730 F.3d at 325. Elonis eventually caught the attention of the FBI when he described online his plan to “initiate the most heinous school shooting ever imagined.”14×14. Id. at 326. After two agents visited his home to investigate, Elonis wrote publicly about the “Little Agent Lady” that it “[t]ook all the strength [he] had not to turn the bitch ghost” by “slit[ting] her throat.”15×15. Id.

In December, federal prosecutors intervened.16×16. Id. A grand jury in the Eastern District of Pennsylvania indicted Elonis for communicating a threat to injure the person of another in violation of 18 U.S.C. § 875(c).17×17. Id.; see also 18 U.S.C. § 875(c) (2012). The five-count indictment charged that Elonis threatened five targets: (1) patrons and employees of the park where he worked, (2) Tara Elonis, (3) state law enforcement, (4) a kindergarten class, and (5) an FBI agent. Elonis, 730 F.3d at 326. Elonis moved to dismiss the indictment and argued that the Supreme Court’s definition of “true threat” in Virginia v. Black18×18. 538 U.S. 343 (2003). barred his prosecution absent proof that he subjectively intended to threaten the subjects of his posts.19×19. Elonis, 730 F.3d at 327; see also United States v. Elonis, No. 11-13, 2011 WL 5024284, at *1 (E.D. Pa. Oct. 20, 2011). The district court denied his motion.20×20. See Elonis, 2011 WL 5024284, at *4. At trial, the district court instructed the jury that Elonis could be convicted if a reasonable speaker would foresee that the relevant posts would be interpreted as threats.21×21. The jury instruction read: “A statement is a true threat when a defendant intentionally makes a statement in a context or under such circumstances wherein a reasonable person would foresee that the statement would be interpreted by those to whom the maker communicates the statement as a serious expression of an intention to inflict bodily injury or take the life of an individual.” Elonis, 730 F.3d at 327. The jury convicted Elonis, and the court sentenced him to forty-four months of incarceration followed by three years of supervised release.22×22. Id. The jury convicted Elonis on Counts 2 through 5, and the district court held that Counts 3 and 5 were “true threats.” Id.

The Third Circuit affirmed.23×23. Id. at 332. Writing for the panel, Judge Scirica held that “Black does not clearly overturn the objective test the majority of circuits [have] applied to § 875(c).”24×24. Id. Instead, the “objective intent standard protects non-threatening speech while addressing the harm caused by true threats.”25×25. Id.

The Supreme Court reversed and remanded.26×26. Elonis, 135 S. Ct. at 2013. Writing for the Court, Chief Justice Roberts27×27. The Chief Justice was joined by Justices Scalia, Kennedy, Ginsburg, Breyer, Sotomayor, and Kagan. held that the district court’s jury instruction erroneously employed a “reasonable person” standard “inconsistent with ‘the conventional requirement for criminal conduct — awareness of some wrongdoing.’”28×28. Elonis, 135 S. Ct. at 2011 (quoting Staples v. United States, 511 U.S. 600, 606–07 (1994)). The Chief Justice characterized the Third Circuit’s rule as a “negligence” standard and found that it was “not sufficient to support a conviction under Section 875(c).” Id. at 2013. The Chief Justice began with the text of § 875(c): a communication must be transmitted and that communication must contain a threat.29×29. Id. at 2008. But he concluded that the statute itself does not indicate a mental state: dictionary definitions of “threat” imply “the notion of an intent to inflict harm,” but these definitions “speak to what the statement conveys — not to the mental state of the author.”30×30. Id. The competing textual interpretations advocated for by Elonis and the Government did not fill the statutory void. See id.

Faced with statutory silence on the required mens rea, the Chief Justice turned to a familiar maxim: “[W]rongdoing must be conscious to be criminal.”31×31. Id. at 2009 (quoting Morissette v. United States, 342 U.S. 246, 252 (1952)). This background principle of criminal law means the Court “interpret[s] criminal statutes to include broadly applicable scienter requirements, even where the statute by its terms does not contain them.”32×32. Id. (quoting United States v. X-Citement Video, Inc., 513 U.S. 64, 70 (1994)). But the presumption in favor of scienter is not without limit: the Court may graft onto a criminal statute only the mental state “necessary to separate wrongful conduct from ‘otherwise innocent conduct.’”33×33. Id. at 2010 (quoting Carter v. United States, 530 U.S. 255, 269 (2000)). A “general requirement that a defendant act knowingly,” id., can sometimes suffice to protect the “innocent actor” from prosecution, id. (quoting Carter, 530 U.S. at 269), but other statutes require a heightened mental state lest they risk criminalizing blameless conduct, id.

The Chief Justice then returned to the statute. Because the only element of § 875(c) that separates “wrongful” from “innocent” conduct is “the threatening nature of the communication,” a defendant must possess some mental state regarding “the fact that the communication contains a threat.”34×34. Id. at 2011. As to this “crucial element,” the Third Circuit’s standard was akin to civil liability in tort law: Elonis’s conviction was premised on the jury’s conclusion that a “reasonable person” would interpret his communications as threats.35×35. Id. The Chief Justice rejected this standard because it “reduces culpability on the all-important element of the crime to negligence.”36×36. Id. (quoting United States v. Jeffries, 692 F.3d 473, 484 (6th Cir. 2012) (Sutton, J., dubitante)). The Court has “long been reluctant to infer that a negligence standard was intended in criminal statutes.”37×37. Id. (quoting Rogers v. United States, 422 U.S. 35, 47 (1975) (Marshall, J., concurring)). The Chief Justice concluded by confirming that the mental states of purpose and knowledge support conviction under § 875(c), but did not decide whether a reckless mens rea is sufficient because this question was not briefed by the parties, did not yet divide the circuits, and was thus not ripe for determination. Id. at 2013.

Justice Alito concurred in part and dissented in part.38×38. Id. at 2013 (Alito, J., concurring in part and dissenting in part). After reproving the majority for unjustifiably failing to clarify what mental state § 875(c) requires, Justice Alito argued that “recklessness [is] enough.”39×39. Id. at 2014. A defendant acts “wrongfully” if he “consciously disregards the risk” that his words “will be interpreted as a true threat.”40×40. Id. at 2016; see also id. at 2015–16 (citing, e.g., Model Penal Code § 2.02(2)(c) (Am. Law Inst. 1985)). Justice Alito agreed that the presumption in favor of scienter foreclosed a negligence standard, but contended that “no further presumptions [were] defensible.”41×41. Id. at 2015. Instead, “when Congress does not specify a mens rea in a criminal statute,” judges may not infer that any mental state higher than recklessness is required “without stepping over the line that separates interpretation from amendment.” Id. Moreover, the Court has “described reckless conduct as morally culpable” in other contexts. Id. (citing Farmer v. Brennan, 511 U.S. 825, 835–36 (1994); Tison v. Arizona, 481 U.S. 137, 157 (1987); Garrison v. Louisiana, 379 U.S. 64, 75 (1964); New York Times Co. v. Sullivan, 376 U.S. 254, 279–80 (1964)). Nor would a recklessness standard run afoul of the First Amendment under Justice Alito’s reasoning: The Constitution does not protect “true threats” because they “may cause serious emotional stress for the person threatened” and “may lead to a violent confrontation.”42×42. Id. at 2016. And although Elonis claimed that his words were “constitutionally protected works of art,” context revealed them to be unprotected “real threat[s] in the guise of rap lyrics.”43×43. Id. Justice Alito also dismissed concerns that the statute is overbroad. See id. at 2017.

Justice Thomas dissented.44×44. Id. at 2018 (Thomas, J., dissenting). After observing that the majority failed to provide guidance to “everyone from appellate judges to ev-eryday Facebook users,”45×45. Id. Justice Thomas applied “ordinary rules of statutory construction” and agreed with nine circuit courts that § 875(c) requires only “proof of general intent.”46×46. Id. at 2021. And, contrary to the majority’s conclusion, general intent — “knowledge with respect to the actus reus of the crime”47×47. Id. at 2019 (quoting Carter v. United States, 530 U.S. 255, 268 (2000)). — is not negligence. A defendant must know the “ordinary meaning in context” of the words he communicates48×48. See id. at 2019–21 (citing, e.g., Hamling v. United States, 418 U.S. 87 (1974); Rosen v. United States, 161 U.S. 29 (1896)). but not that a “jury would conclude that his communication constituted a ‘threat’ as a matter of law.”49×49. Id. at 2021. Justice Thomas also rejected Elonis’s constitutional claim, offering historical evidence that “true threats” have never been afforded First Amendment protection.50×50. See id. at 2024–28. Justice Thomas directly dismissed Elonis’s reliance on Watts v. United States, 394 U.S. 705 (1969), and Virginia v. Black, 538 U.S. 343 (2003), pointing out that neither case directly addressed the question at hand. See Elonis, 135 S. Ct. at 2026–27.

By adopting a definition of “threat” within § 875(c) that is not tethered to any mental state,51×51. See id. at 2008 (majority opinion). The Court had not previously construed “threat” in the context of § 875(c). Id. at 2014 (Alito, J., concurring in part and dissenting in part). the Elonis majority left the “recklessness door” conspicuously open, and yet provided almost no guidance to lower courts on whether to walk through it or how to confront the constitutional challenges to the statute on the other side. Drawing parallels between obscenity, another category of unprotected speech, and true threats suggests answers to these unresolved questions. Specifically, the analogy both hints that recklessness might be sufficient mens rea for conviction under § 875(c) and indicates ways in which courts might construe the statute in light of First Amendment interests.

In Elonis, the Court parted ways with lower courts that have almost uniformly interpreted § 875(c)’s use of “threat” in the shadow of the First Amendment’s “true threats” doctrine.52×52. See Blakey & Murray, supra note 2, at 937–38 (“[C]ourts treat the First Amendment as intermingled with the statutory construction. In doing so, the courts find that the statute itself requires a showing of a constitutional ‘true threat,’ they define ‘true threat’ for both constitutional and statutory purposes, and they conclude that any speech that falls within the statute, as so construed, is unprotected.” Id. at 937 (footnote omitted)). Without reference to the First Amendment, the Elonis majority endorsed an objective statutory definition of “threat.”53×53. Elonis, 135 S. Ct. at 2008. In fact, Chief Justice Roberts did not cite any authority in support of the statutory definition of “threat” adopted in Elonis. Id. Though this approach is striking in light of Watts v. United States’s54×54. 394 U.S. 705 (1969). insistence that a statute that criminalizes “a form of pure speech[] must be interpreted with the commands of the First Amendment clearly in mind,”55×55. Id. at 707. it is defensible in light of the principle that, whenever possible, cases should be decided on statutory rather than constitutional grounds. Still, the Court’s decision to reject a subjective definition is the reason a recklessness standard remains viable.56×56. A person cannot be reckless as to the possibility that his words will be interpreted as a “threat” within the meaning of § 875(c) if the statute itself defines “threat” subjectively.

Because Elonis was decided on statutory grounds, “true threats” remain a doctrinal puzzle for lower courts. Given the scarcity of relevant Supreme Court precedent, some scholars and judges have drawn parallels between threats and incitement, a category of unprotected speech that requires subjective intent.57×57. See, e.g., Steven G. Gey, The Nuremberg Files and the First Amendment Value of Threats, 78 Tex. L. Rev. 541, 590–92 (2000). These arguments are persuasive because the line between threats and incitement is muddled: the same language may be viewed in context to be both a “true threat” and incitement to violence.58×58. See Clay Calvert et al., Rap Music and the True Threats Quagmire: When Does One Man’s Lyric Become Another’s Crime?, 38 Colum. J.L. & Arts 1, 6 (2014) (describing “incitement” as a “close legal cousin” to “true threats”); see also Gey, supra note 4, at 1327–31 (discussing NAACP v. Claiborne Hardware Co., 458 U.S. 886 (1982), and the case’s doctrinal intermingling of inciting and threatening speech). Nonetheless, incitement is an imperfect analogue. Justifications for criminalizing language that incites violence focus on the risk that a third party will commit violence against the subject, not the harm the speech itself causes to the recipient.59×59. See Planned Parenthood of the Columbia/Willamette, Inc. v. Am. Coalition of Life Activists, 290 F.3d 1058, 1106 (9th Cir. 2002) (en banc) (Berzon, J., dissenting) (noting the difference “between a statement that one oneself intends to do something and a statement encouraging or advocating that someone else do it”); Jennifer Elrod, Expressive Activity, True Threats, and the First Amendment, 36 Conn. L. Rev. 541, 564–75 (2004) (distinguishing threats and incitement along several axes, including the objective of the speaker). Moreover, unlike threats, which are often made privately, incitement frequently occurs in the public sphere.60×60. See Planned Parenthood, 290 F.3d at 1099 (Kozinski, J., dissenting) (discussing the import of the public/private distinction in the context of threats). Within the categories of unprotected speech, obscenity provides an alternative comparison. Though still imperfect, an analogy to obscenity is superior to an analogy to incitement in solving the two questions left unanswered by Elonis.

First, obscenity and threats share characteristics that are relevant to the mens rea question at the heart of Elonis. Courts must now ascertain whether a reckless mental state adequately separates “wrongful” from “innocent” conduct in the context of threatening speech.61×61. See Elonis, 135 S. Ct. at 2010. “Wrongfulness” can be viewed in light of the harm that the statute seeks to prevent.62×62. For two examples of this type of statutory “harm-centric” reasoning in the threats context, see Roy v. United States, 416 F.2d 874, 877 (9th Cir. 1969); and United States v. Jeffries, 692 F.3d 473, 480 (6th Cir. 2012). Under this reasoning, it makes sense for threats to require the same mens rea as obscenity63×63. Cf. Brief Amicus Curiae of the Criminal Justice Legal Foundation in Support of Respondent at 9, Elonis, 135 S. Ct. 2001 (No. 13-983) (“The closest analogy . . . can be found in the pornography cases.”). because these statutes target comparable harms, and the causal mechanism for these harms is the same. Both types of prohibitions mitigate a perceived individual and social harm that arises immediately upon mere exposure to the relevant speech64×64. Compare Roth v. United States, 354 U.S. 476, 494 (1957) (Warren, C.J., concurring), and Paris Adult Theatre I v. Slaton, 413 U.S. 49, 57–62 (1973) (identifying potential harms of obscenity, such as to a community’s quality of life or public safety), with R.A.V. v. City of St. Paul, 505 U.S. 377, 388 (1992) (noting that mere receipt of a threat produces fear and disrupts daily life for the audience), and Brief of Amicus Curiae National Center for Victims of Crime in Support of Respondent United States at 7–12, Elonis, 135 S. Ct. 2001 (No. 13-983) (discussing how threats can produce fear, anxiety, and additional stresses for recipients). by shielding audiences from expressions that produce a noncognitive — almost physical — response in the recipient.65×65. See Kenneth L. Karst, Threats and Meanings: How the Facts Govern First Amendment Doctrine, 58 Stan. L. Rev. 1337, 1339–46 (2006) (stressing physical impact that “fear” has on the recipient of a threat); Gey, supra note 57, at 593–94 (observing that the “concept of true threats . . . rests on the assumption that a true threat is outside the scope of First Amendment protection because it operates more like a physical action than a verbal or symbolic communication of ideas,” and arguing that this “speech is unprotected to the extent that it causes an immediate, visceral reaction that by its nature is unmediated by the intervention of time for reflection and the possibility of a rational response”).

Second, similarities between the constitutional challenges in obscenity and threats cases may shed light on the lurking First Amendment issue after Elonis. Current precedent says “[a]ll that the First Amendment requires in the context of a § 875(c) prosecution is that the threat be real — a ‘true threat.’”66×66. Jeffries, 692 F.3d at 478 (citing Watts v. United States, 394 U.S. 705, 708 (1969) (per curiam)). But such a threat “must be distinguished from . . . constitutionally protected speech”67×67. Watts, 394 U.S. at 707. because not all statements “intending to create a fear of violence” are unprotected.68×68. Planned Parenthood of the Columbia/Willamette, Inc. v. Am. Coalition of Life Activists, 290 F.3d 1058, 1094 (9th Cir. 2002) (en banc) (Kozinski, J., dissenting) (quoting NAACP v. Claiborne Hardware Co., 458 U.S. 886, 927 (1982)). Only speech that “communicate[s] a serious expression of an intent to commit an act of unlawful violence to a particular individual or group of individuals” may be criminalized.69×69. Virginia v. Black, 538 U.S. 343, 359 (2003). This formulation mirrors Justice Brennan’s admonitions in Roth v. United States70×70. 354 U.S. 476 (1957). that “sex and obscenity are not synonymous”71×71. Id. at 487. and that “standards for judging obscenity” must protect speech that treats sex in a nonobscene manner.72×72. Id. at 488. And just as “[t]here is no external measuring rod for obscenity,”73×73. Smith v. California, 361 U.S. 147, 165 (1959) (Frankfurter, J., concurring). so too does the term “true threat” lack an ex ante discernable definition.

Building on this analogy can guide lower courts as they consider both the threshold statutory question of mens rea under § 875(c) and future constitutional challenges to the statute.

To begin, although the Elonis Court does not make these parallels explicit, it implies that obscenity cases might hold lessons for the mens rea question in the threats context. The Court centers its most in-depth analysis of mens rea on obscenity cases.74×74. See, e.g., Elonis, 135 S. Ct. at 2011–12. The disagreement between the majority and Justice Thomas over the reading of these cases confirms that the issue of mens rea for obscenity is itself fraught.75×75. An early obscenity case provided an analogously incomplete answer to the mens rea question. See Smith, 361 U.S. at 153–54 (“We need not and most definitely do not pass today on what sort of mental element is requisite to a constitutionally permissible prosecution of a bookseller for carrying an obscene book in stock . . . .” Id. at 154.). But the Elonis majority characterized a key obscenity precedent in a light that resembles recklessness. It described Hamling v. United States76×76. 418 U.S. 87 (1974). as requiring that a defendant know the “character” of obscene material he distributes, “not simply its contents and context,”77×77. Elonis, 135 S. Ct. at 2012. and emphasized that such a requirement ensured “not innocent but calculated purveyance of filth” was outlawed.78×78. Id. (quoting Hamling, 418 U.S. at 122). “Calculated purveyance” is not a precise term, but the majority would require only that Elonis “know the threatening nature of his communication.”79×79. Id. Under this formulation, Hamling sounds like a recklessness standard. The defendant must know the “character” of the material — and thus the substantial and unjustifiable risk that it is obscene — yet distribute it nonetheless.80×80. Such a reading is consistent with an earlier deployment of Hamling. See Posters ‘N’ Things, Ltd. v. United States, 511 U.S. 513, 524–25 (1994) (reasoning that under Hamling, a defendant need not know the items he sold were “drug paraphernalia” but only that they “were likely to be used with illegal drugs,” id. at 524); cf. Arnold H. Loewy, Essay, Obscenity, Pornography, and First Amendment Theory, 2 Wm. & Mary Bill Rts. J. 471, 482 n.70 (1993).

By juxtaposing Hamling’s “calculated purveyance” standard with the facts of Elonis, the majority invites lower courts to adopt a mens rea of recklessness for § 875(c). Notably, the Court in Elonis flatly rejected the strongest argument for distinguishing “threat” from “obscenity” and imposing a mens rea requirement of purpose — the textual argument that the word “threat” itself contains an intentionality component.81×81. See Elonis, 135 S. Ct. at 2008. Without this textual hook, mandating different levels of intentionality for obscenity and threats would create a discontinuity between two otherwise analogous speech prohibitions.82×82. Without a justification that applies equally to obscenity, threats would become “one of the most protected categories of unprotected speech.” Id. at 2027 (Thomas, J., dissenting).

Similarly, the obscenity cases provide lessons for addressing the constitutional question left unresolved by Elonis. The majority’s example of the actus reus of § 875(c) — “express[ing] an intention to inflict loss or harm”83×83. Id. at 2008 (majority opinion) (quoting Webster’s New International Dictio-nary 2633 (2d ed. 1954)). — is potentially sweeping.84×84. Even as the majority distanced itself from Black’s definition of “true threat,” the Justices did not provide a statutory substitute. Instead, the majority provided only this example: “[A]n anonymous letter that says ‘I’m going to kill you’ is [a threat] regardless of the author’s intent. A victim who receives that letter in the mail has received a threat, even if the author believes (wrongly) that his message will be taken as a joke.” Id. In fact, the reasoning in many “true threats” cases still looks a lot like this: “I know it when I see it.”85×85. Jacobellis v. Ohio, 378 U.S. 184, 197 (1964) (Stewart, J., concurring). Even in Elonis, when rejecting Elonis’s reliance on the First Amendment, Justice Alito proceeded by invoking the “true threats” exception, recounting the language of Elonis’s threats, and concluding that they qualified as “true threats.” See Elonis, 135 S. Ct. at 2016–17 (Alito, J., concurring in part and dissenting in part). But if criminal liability turns solely on whether the defendant subjectively knew of and disregarded the risk that his message was objectively a threat, then clearly delineating ex ante what constitutes a constitutionally unprotected threat is paramount. The Court’s obscenity cases are again apposite: criminal liability turns on how a jury categorizes the relevant material, making an adequate definition for “obscenity” essential to the constitutionality of such statutes.

The Court’s current definition of a “true threat,” however, is categorical and thus resembles early treatment of “obscenity.”86×86. See Roth v. United States, 354 U.S. 476, 497 (1957) (Harlan, J., concurring in part and dissenting in part). “[T]rue threats” are cast as “a peculiar genus . . . which is as distinct, recognizable, and classifiable as poison ivy is among other plants.”87×87. Id. (emphasis omitted). Whether a particular utterance can be prosecuted under § 875(c) “becomes a mere matter of [ex post] classification” left to the jury.88×88. Id. Yet as countless challenges to obscenity statutes reveal,89×89. In the years following Roth and in light of the alleged constitutional infirmities in the definition of “obscenity,” the Court adopted the much-maligned practice of simply summarily reversing convictions that a majority of Justices felt were unconstitutional. See Miller v. California, 413 U.S. 15, 22 n.3 (1973). such a standard is in tension with the principle that speech restrictions not be overbroad or vague.

Consequently, to address alleged constitutional infirmities in § 875(c), the Court’s doctrinal response to challenges to obscenity prosecutions is an apt starting point. Specifically, “threat,” like “obscenity,” is susceptible to judicial narrowing.90×90. The landmark obscenity case, Miller v. California, 413 U.S. 15, addressed constitutional concerns by construing “obscenity” narrowly. See id. at 24; accord Osborne v. Ohio, 495 U.S. 103, 119–21 (1990) (discussing judicial narrowing of definition of “obscenity”). Courts have already attempted to devise tests that separate out threatening language that may not be suppressed consistent with the First Amendment.91×91. For example, one circuit added an “immediacy” requirement to the definition of “true threat.” See United States v. Kelner, 534 F.2d 1020, 1028 (2d Cir. 1976). Another judge proposed that the “difference between a true threat and protected expression” is that a “true threat warns of violence or other harm that the speaker controls.” Planned Parenthood of the Columbia/Willamette, Inc. v. Am. Coalition of Life Activists, 290 F.3d 1058, 1089 (9th Cir. 2002) (en banc) (Kozinski, J., dissenting). Scholars have similarly worked to develop a more precise definition of a “true threat.” See, e.g., Jennifer E. Rothman, Freedom of Speech and True Threats, 25 Harv. J.L. & Pub. Pol’y 283, 333–36 (2001); Gey, supra note 57, at 590; Blakey & Murray, supra note 2, at 1076–77. The purpose of analogizing threats to obscenity is to highlight that such an approach — narrowing the actus reus component of a constitutionally suspect criminal speech prohibition rather than imposing a heightened specific intent mens rea — is an alternative to Brandenburg v. Ohio’s92×92. 395 U.S. 444 (1969). constitutional requirement of a specific intent to incite violence before a defendant may be convicted on the basis of his language.93×93. See id. at 447 (per curiam).

Whenever the state undertakes to suppress pure speech, difficult constitutional concerns abound. The Court’s tortured obscenity line of cases is illustrative. Yet because courts have been grasping toward a constitutionally adequate definition of “threat,” but have also defaulted to the kind of “I know it when I see it” reasoning employed in early obscenity cases, they might acknowledge this parallel and model a narrowed definition of “threat” off of the rule for “obscenity” that was finally settled in Miller v. California. Though “[t]his may not be an easy road, free from difficulty,”94×94. Miller, 413 U.S. at 29. such a construction of “threat” is one alternative to reading a mental state requirement of knowledge or purpose into § 875(c).