Technology and the role of intent in constitutionally protected expression.
Intent and context matter enormously in communication. While that has always been true, social media services, smartphones, the cloud, and other recent technological innovations have fundamentally changed the landscape of expression. And they have done so in ways that have profoundly important consequences with regard to understanding the limits of First Amendment protections.
Today, the quantity of expression, measured in distinct, easily archivable communications transmitted from one person to one or more recipients, is dramatically higher than in the past. In less time than it would have taken to produce a single handwritten letter in the pre-digital era, a person can now issue dozens of texts, tweets, or other electronic communications. The potential to send large numbers of messages is not merely theoretical: A 2013 survey found that U.S. smartphone users in the eighteen to twenty-four age group were sending an average of sixty-seven text messages per day. (1) Most adults in higher age brackets also sent dozens of daily text messages, though fewer than those in the eighteen to twenty-four group. (2) To further complicate matters, people are not the only ones speaking: Many tweets, for example, are now issued by computers programmed to mimic human behavior. (3)
In addition, the number of potential pairings of speakers (4) with listeners has skyrocketed. In the United States there are well over sixty million active (5) Twitter users, (6) each of them broadcasting messages that can be read by anyone in the world with an Internet connection. (7) There are also about one hundred million active U.S. users of Instagram (8) and several hundred million active U.S. users of Facebook, (9) many of whom are regularly engaged in publishing content. In the pre-Internet era, only a small percentage of people were broadcasters. Today, most people are.
In combination, these changes mean that even if only a tiny fraction of expression raises questions of First Amendment limits, in absolute terms the number of communications in that fractional category is now immense. A further complication is that technology has made it easy to copy, forward, excerpt, comment on, archive, and remix content. Messages written with one audience in mind are commonly retransmitted to other audiences in other contexts, often in a manner outside the control of the original speaker. Both very short time spans and long time spans can strip context: In minutes, a message in cyberspace can outrun the person who created it, repackaged and repurposed by others in ways that may mask or appear to alter a speaker's original intent. Long time spans can also alter apparent meaning. A digital message can be dredged up and re-circulated months or years after its original composition and then interpreted in light of intervening events that the message's author could not have foreseen.
The potential impact of impulsive and spontaneous expression has also changed. As has always been the case, statements associated with emotions such as surprise, fear, anger, joy, and disappointment are often made in the heat of the moment without significant forethought regarding their consequences. In the pre-digital era such statements were often made verbally in person and would linger in the air for only a few seconds. Today, spontaneous expression is often conveyed electronically and therefore automatically archived, exposing it to scrutiny weeks, months, or years into the future. (10)
In short, because of today's rapidly changing communications technology landscape, the role of speaker intent is more complex than ever before. And questions regarding intent are particularly acute when expression allegedly falls outside the bounds of First Amendment protection and, relatedly, within the scope of criminal statutes addressing communications. This issue was central to Elonis v. United States, (11) which considered a conviction arising from a series of threatening Facebook posts. But while the Supreme Court's June 2015 ruling in Elonis confirmed the importance of intent with respect to the federal criminal statute in question, (12) it did not clarify the specific level of intent required for conviction under that statute. (13) More fundamentally, the Court did not reach the broader constitutional question regarding the role of intent with respect to the scope of First Amendment protections.
This Article argues that as communications technologies continue to evolve, they will increasingly enable information dissemination in ways that can reduce, alter, or remove context from a speaker's original expression. In light of these changes, maintaining a traditional view of the role of mens rea will provide an increasingly important bulwark against inadvertently criminalizing expression that in fact deserves First Amendment protection. (14) This Article also takes the position--perhaps unpopular in some circles in light of alarming but increasingly common views suggesting that the First Amendment is "outdated" (15)--that failure to properly recognize speaker intent would have consequences far beyond criminal law, leading to an erosion of civil liberties regarding expression (including technology-facilitated expression) in a much broader set of non-criminal contexts.
The rest of this Article is organized as follows: Section I discusses how technological changes including the Internet, social networking, and smartphones are complicating the relationship between intent and communication. Section II provides historical context regarding mens rea and the growth in statutes and regulations that are silent on intent, and also briefly discusses recent attempts at federal mens rea reform. Section III reviews the 2015 Supreme Court ruling in Elonis v. United States and considers the post-Elonis landscape regarding intent. Section IV connects the technological and legal landscapes together, using a series of hypothetical scenarios to illustrate how an examination of intent is vital in light of the many ways that information can be conveyed using existing and emerging communications technologies. Section V presents conclusions. While the discussion herein focuses primarily on intent in the context of expression that could expose a speaker to criminal liability, some (though certainly not all) of the examples and analysis are also relevant in the context of civil liability. (16)
I. TECHNOLOGY AND INTENT IN COMMUNICATIONS
For most of American history, communications and intent have been more consistently and reliably intertwined than they are today. Sending a letter in the late eighteenth century, a telegram in the nineteenth century, or making a phone call in the twentieth century all involved an affirmative choice to convey information to--and typically to also receive information from--another person. In addition, it has long been possible to convey a single message to many people, often with the aid of technology. In the Founding Era, the printing press was used to enable wide distribution of written expression. In later centuries, audio amplification and recording and then radio and television broadcasts facilitated one-to-many communication. In all of these cases, while interpretations of the content of a given message could vary, there was little uncertainty regarding either the intent of the speaker to disseminate it or the audience he or she was aiming to reach. A person giving a television speech in 1980 was separated by two centuries from someone publishing a pamphlet in 1780, but both speakers had in common the goal of using the most effective technology of the day to ensure distribution of a message to large numbers of people.
A. Current Communications Technology
Technology advances since the 1990s (17) have upended this landscape in ways that can decouple the connection between the transmission of content and the intent to deliver it to a recipient. Of course, intentionally directed communications remain common: An e-mail message sent to a single recipient is in some respects the modern equivalent of a letter, created and conveyed with intent to the person who then receives and reads it. But there are many forms of modern expression that have no comparable pre-digital era analog. Today it is possible, and in fact common, for a person to create a document and cause it to be transmitted across state and perhaps international boundaries, even when he or she never intends for any other person to see it. This is exactly what occurs when someone creates a private document using a cloud-based service such as Google Docs. Another change from the past is that recipients of messages often play a far more active role in their transmission. In contrast with a nineteenth-century recipient of a letter or a telegram, who was generally uninvolved in its conveyance, users of digital services often affirmatively request transmission of information from a remote location to their own devices. Communications are often delivered to a user only after the user takes an action such as clicking a link, entering a web page address, "following" someone on Twitter, or launching an app that specifically requests content delivery. (18)
There are also many more shades of gray than in the past with respect to how a speaker might select and interact with an intended audience. A person can engage in what amounts to a broadcast, sending a communication using a mechanism (such as a public tweet) that makes it accessible to anyone in the world with an Internet connection. At the other end of the spectrum, as noted above, creating an unshared, private document in the cloud means transmitting its contents--perhaps over thousands of miles--to no one but its creator. There are many points between these extremes. A speaker can direct a communication to a single recipient by using an e-mail message sent to a single addressee, a "direct message" on Twitter, a "private message" on Facebook, or through "Instagram Direct" on Instagram. (19) Alternatively, a sender can issue a message intended for a group of recipients such as a collection of people on an e-mail distribution list. Social networking audience groups introduce yet another set of layered complexities: On Facebook, "friends" are established by mutual agreement, and can involve different levels of friendship associated with different levels of content exposure. (20) Most Twitter accounts are public, limiting a Twitter account holder's ability to control the composition of his or her principal audience. (21) However, Twitter also offers an option to set up an account for sending "protected tweets" visible only to followers approved by the sender. (22) A further variation is that tweets can include text intended to attract attention from particular Twitter account holders (23) or from people interested in a particular topic. (24)
Moreover, it has become extremely difficult to know the physical path taken by digital information--an issue that is particularly relevant given the differing legal frameworks that can apply to communications within a single state, across state lines, and across international boundaries. (25) Using a smartphone app means sending and receiving data along a complex path that includes the smartphone itself, a nearby mobile base station or Wi-Fi access point, an associated network operated by a wireless carrier, and one or more servers operated by the company that provided the app. Those servers might reside on the company's premises or at an off-site data center owned by a third party. The details of these various communications systems are typically invisible to the smartphone user, who often has little or no information about the physical routing taken by data he or she sends or receives.
A closely related consideration is that using the Internet or a smartphone means using the cloud--a term that intentionally masks a complex landscape with rapidly evolving legal implications. Google's data centers are located in more than half a dozen U.S. states as well as in many overseas locations. (26) Facebook has data centers in Iowa, Oregon, North Carolina, Texas, and Sweden. (27) Amazon Web Services has data centers in each of dozens of "availability zones" around the world. (28) Anyone who uses services provided by Google, Facebook, or Amazon--that is, nearly everyone--will often be transmitting content to (and receiving content from) one or more of these data centers.
In addition to abstracting away the concept of physical location, the cloud acts to magnify the effective storage and communications capabilities of consumer devices. A smartphone not only has its own formidable computational and storage resources, but also serves as a portal to access the vastly greater resources of data centers that are literally the size of entire city blocks. A few buttons pressed on a smartphone can harness the power of a data center to broadcast a message so that it will be instantly available to billions of people. The process operates in the inverse sense as well: Smartphones, or any other device that runs a web browser, can be used to cause data centers to extract the one message from among billions that a listener might be seeking. And all of this can happen nearly instantaneously, leading to fundamental changes in how information is conveyed, excerpted, quoted, repurposed, archived, and searched.
B. Algorithms, Artificial Intelligence, and the Future of Expression
Future advances will further complicate the links between intent and communications, though predicting the technological future is always a challenging exercise. After all, when Internet browsers became headline news in 1995, (29) few communications "experts" would have been able to accurately foresee the growth in smartphones and social networking that occurred in the 2000s. But current trends provide at least a basis for identifying some ways that communication will likely evolve in the coming decade.
One key future development will be an increased use of algorithms, (30) and more generally, artificial intelligence, (31) in expression. This will lead to more expression retransmitted by--and increasingly composed by--machines. To some extent this is already happening. "Bots" (32) are commonly used to scan and identify tweets containing specific keywords, and then to re-tweet (33) them from a new account. (34) Bots are also frequently used to compose original tweets, for example, by automatically searching through newly published news articles, identifying those that address a topic of interest, and then formulating and transmitting a tweet including a link to the article.
Another trend that is already gaining traction (35) and that will accelerate in the future is automatic generation (called "autogeneration") of entire articles. Consider a December 2015 Yahoo! Finance article discussing quarterly financial results from AeroVironment, a publicly traded California-based company that manufactures drones. (36) The opening sentences of the article stated "AeroVironment Inc. (AVAV) on Tuesday reported fiscal second-quarter net income of $4.4 million, after reporting a loss in the same period a year earlier. On a per-share basis, the Simi Valley, California-based company said it had net income of 19 cents. The results topped Wall Street expectations." (37)
Those specific sentences were written not by a human writer, but instead by an algorithm that automatically inserted the figures "$4.4 million" and "19 cents" into a template, and automatically identified that those results "topped" (as opposed to met, or fell short of) Wall Street expectations. Instead of identifying a non-existent human author, the story contained a notation at the end stating "[t]his story was generated by Automated Insights." (38) In addition to writing stories on their own, machines can collaborate with human authors. A December 2015 Associated Press article regarding an earnings release from FedEx bore the byline "David Koenig." (39) But a note at the bottom of the article also gave a nod to a non-human co-author: "Elements of this story were generated by Automated Insights." (40)
Of course, humans have a hand even in auto-generated text. For every article that is generated by a computer, somewhere upstream are people who wrote the software running on the computer and the text for some or all of the template. As a result, today's auto-generated articles are often formulaic, generated through what amounts to an automated process of filling in the blanks in a previously prepared template. There is an enormous creative distance separating such stories from traditional articles written by skilled, seasoned authors. But history shows that advances in artificial intelligence will erode some of that distance, enabling computers to compose increasingly sophisticated documents that will sometimes be very difficult to identify as machine-generated.
So what does all of this have to do with the First Amendment? The overwhelming majority of expression, machine-generated or not, raises no First Amendment issues. But true threats, (41) incitements to imminent lawless action, (42) and other unprotected expression--like all other content today --is created and disseminated in a landscape that has grown far more complex in only a generation. This can make it much harder to discern the intent and context that went into the original composition and transmission of a message. Criminalizing expression places this issue front and center, even when--in fact, especially when--a criminal statute relating to expression is written without an explicit mens rea requirement.
In short, while state of mind has always mattered in expression, technology can make it harder to infer. This has profound consequences for expression with respect to criminal law and beyond. In addressing the confluence of technology and intent, it is helpful to provide context by considering the historical view of mens rea in criminal law, how that view has evolved, and the current landscape with respect to mens rea and the First Amendment in light of the June 2015 Elonis ruling.
II. MENS REA IN CONTEXT
The concept of mens rea--the Latin phrase for a guilty mind--has deep roots in American criminal law and, in turn, in English common law. (43) A crime has historically been understood to require the combination of both the act itself, or actus reus, and awareness that the act was wrong, or mens rea. Edward Coke's seventeenth century Institutes of the Laws of England stated that "actus non facit reum nisi mens sit rea" (44)--an act itself is not culpable unless it is done with a guilty mind. In a 1798 bankruptcy case in England, Lord Kenyon invoked the same Latin phrase, writing that "it is a principle of natural justice, and of our law, that actus non facit reum nisi mens sit rea. The intent and the Act must both concur to constitute the crime." (45) William Blackstone's eighteenth century Commentaries on the Laws of England observed that "as a vitious will without a vitious act is no civil crime, so on the other hand, an unwarrantable act without a vitious will is no crime at all." (46)
As the Supreme Court explained in (1952) in Morissette v. United States, (47) the understanding of crime as a "compound concept" involving the pairing of "an evil-meaning mind with an evildoing hand" "took deep and early root in American soil": (48)
As the states codified the common law of crimes, even if their enactments were silent on the subject, their courts assumed that the omission did not signify disapproval of the principle but merely recognized that intent was so inherent in the idea of the offense that it required no statutory affirmation. (49)
Using language that strongly echoed Blackstone, the Morissette Court wrote that:
[the] contention that an injury can amount to a crime only when inflicted by intention is no provincial or transient notion. It is as universal and persistent in mature systems of law as belief in freedom of the human will and a consequent ability and duty of the normal individual to choose between good and evil. (50)
The 1952 Morissette decision is one of many in which the Supreme Court has highlighted the central role of mens rea in criminal law. In 1985, in Liparota v. United States, (51) the Court considered allegedly criminal misuse of food stamps, and held that "in a prosecution for violation of [a federal statute addressing food stamp fraud], the Government must prove that the defendant knew that his acquisition or possession of food stamps was in a manner unauthorized by statute or regulations." (52)
(52) In (1994) there were three Supreme Court rulings directly addressing the issue of criminal intent. In Posters 'N' Things, Ltd. v. United States, (53) the Court held that conviction of a defendant for selling drug paraphernalia required the government to "establish that the defendant knew that the items at issue are likely to be used with illegal drugs." (54) In United States v. X-Citement Video, Inc., (55) the Court ruled that a defendant's knowledge of the facts making an act a crime was "the crucial element separating legal innocence from wrongful conduct." (56) And in Staples v. United States, (57) the Court concluded that while the language of the statute at issue did not mention intent, "[s]ilence does not suggest that Congress dispensed with mens rea," (58) and thus "to obtain a conviction, the Government should have been required to prove that petitioner knew of the features" (59) of a firearm that brought it within the scope of the National Firearms Act.
The Court revisited mens rea again in 2000 in Carter v. United States, (60) writing that the "presumption in favor of scienter requires a court to read into a statute only that mens rea which is necessary to separate wrongful conduct from 'otherwise innocent conduct.'" (61) This language was cited again by the Court in its 2015 Elonis v. United States ruling: "When interpreting federal criminal statutes that are silent on the required mental state, we read into the statute 'only that mens rea which is necessary to separate wrongful conduct from otherwise innocent conduct.'" (62)
B. When Intent Does Not Matter: The Growth of Strict Liability Crimes
Citing the Supreme Court's repeated affirmation of the role of mens rea tells only part of the story. While mens rea has long been a central feature of American criminal law, it is far from universally applied. Starting in about the middle of the nineteenth century, an increasing number of crimes in the United States were prosecuted without consideration of intent. (63) Writing in 1933, Harvard Law professor Frances Bowes Sayre observed that:
[W]e are witnessing today a steadily growing stream of offenses punishable without any criminal intent whatsoever. Convictions may be had for the sales of adulterated or impure food, violations of the liquor laws, infractions of anti-narcotic acts, and many other offenses based upon conduct alone without regard to the mind or intent of the actor. (64)
Sayre, who coined the term "public welfare offenses" (65) to describe these statutes, attributed to them to the "increasing social regulation" made necessary by "the growing complexities of twentieth century life." (66)
Sayre recognized that these "complexities" were in significant part driven by technological changes, (67) writing that:
[the] invention and extensive use of high-powered automobiles require new forms of traffic regulation[;]... the development of modern medical science and the congested living accommodations of modern cities require new forms of sanitary and health regulation; the growth of modern factories requires new forms of labor regulation; the development of modern building construction and the growth of skyscrapers require new forms of building regulation. (68)
Thus, while issues of intent in relation to crime and digital technology are new, technology in the broader sense has shaped the conception the role of intent in crime for well over a century.
In part, this change was rooted more in practicality than in philosophy. The growth during the early twentieth century in the number of sectors subject to government oversight, many of which had no direct pre-industrial analog, led to an increase in the number of statutes and regulations involving potential criminal liability. Courts would have been overwhelmed if they were required to examine mens rea for all offenses, no matter how minor. (69) This led to a bifurcation that is still in place today between crimes where intent is evaluated and those where it is not: Minor offenses (and some major ones, including statutory rape and some drug-related crimes) are often treated under a strict liability framework in which guilt or innocence is determined based only on the act itself. A person who parks illegally and then gets a citation will find it futile to contest it by arguing that he or she did not see the "No Parking" sign. But for more serious offenses, intent is often at the center of the inquiry. If a person elbows someone in the face, understanding state of mind is the key to distinguishing an unfortunate accident from battery.
There are now many federal and state crimes on the books that are silent on intent, and that can therefore be prosecuted without regard to the defendant's state of mind. And, as the number of such statutes grows, there is an ever-increasing risk that people will be subjected to criminal prosecution for innocent actions. Conversely, statutes that lack clarity on intent can also be misapplied in ways that let the guilty go free.
Additionally, today there are more statutes addressing serious crimes that lack a mens rea requirement. As a result, defendants can be in the unenviable position of facing prison time for innocuous or inadvertent acts. There are numerous modern examples where this has occurred. Professional auto racer Bobby Unser and a friend got lost while snowmobiling in Colorado in December 1996. After two days in the cold they were able to find their way out, but not before accidentally straying into the San Juan Wilderness. In 1997, Unser was convicted and fined for violating the Wilderness Act of 1964. He had faced the possibility of imprisonment for up to six months. (70) In 2003, an Alaska man was criminally charged after selling sea otters to a person he believed to be a Native American, but who in fact was not. He pled guilty to violating the Marine Mammal Protection Act. (71) In 2011, a Virginia woman was threatened with imprisonment for allegedly violating the Migratory Bird Treaty Act after her eleven-year old daughter rescued a baby woodpecker that was about to be killed by a cat. (72) There are many more cases of people who have become unwittingly entangled in federal criminal prosecutions, often for violations of laws that they did not even know existed. (73)
C. Mens Rea Reform Initiatives
Criminal statutes that are silent on intent are part of a broader, well-recognized over-criminalization problem in the United States. (74) One way to address this is through legislative remedies. However, while there is bipartisan agreement in Congress that over-criminalization is a problem, (75) proposals to address it through mens rea reform--the term for legislation that would require courts to apply a mens rea standard in cases involving statutes drafted without reference to intent--are controversial.
In late 2015 two bills were introduced in the House and Senate that address mens rea reform. Both bills, for federal criminal statutes where the text is silent on state of mind, would require prosecutors to prove a specific level of intent. The House bill, called the "Criminal Code Improvement Act of 2015," (76) if enacted would provide that:
[i]f no state of mind is required by law for a Federal criminal offense--(1) the state of mind the Government must prove is knowing; and (2) if the offense consists of conduct that a reasonable person in the same or similar circumstances would not know, or would not have reason to believe, was unlawful, the Government must prove that the defendant knew, or had reason to believe, the conduct was unlawful. (77)
The Senate bill, titled the "Mens Rea Reform Act of 2015," (78) would require (with some exceptions) (79) the government:
to prove beyond a reasonable doubt that the defendant acted--(1) with the state of mind specified in the text of the covered offense for each element for which the text specifies a state of mind; and (2) willfully, with respect to any element for which the text of the covered offense does not specify a state of mind. (80)
Both bills (81) have generated stiff opposition and neither is likely to become law. (82) Thus, as a practical matter, for federal criminal statutes in which the text is silent on intent, issues of mens rea will continue to be argued and decided in the courts for the foreseeable future. (83)
III. ELONIS V. UNITED STATES
Most statutes that lack a mens rea requirement relate to crimes having little to do with either expression or with the digital communications technology changes in recent years. For example, the prohibitions cited above on activities like transporting migratory birds, selling otters to non-Native Americans, and snowmobiling in federal wilderness areas (84) are implicated by activities that are not uniquely modern and that do not typically involve freedom of expression. (85)
By contrast, when the government criminalizes expression, contemporary technology is particularly relevant for the simple reason that the Internet, smartphones, and social networks have revolutionized how people convey information. (86) Such statutes lie at a critically important confluence of questions regarding mens rea, the scope of the First Amendment, and changing technologies. In fact, these statutes create their own weather with respect to this intersection because technological evolution makes possible a whole new range of behaviors that can potentially fall within the scope of laws that may be decades old. This was precisely what occurred with Elonis, which considered twenty-first century Facebook postings in light of 18 U.S.C. [section] 875(c), a statute little changed since its enactment over half a century ago, in 1948. (87)
A. Subjective vs. Objective Standards
Elonis relates to longstanding questions regarding whether and how the state of mind of a person making an allegedly unlawful threat should factor into its evaluation. Broadly speaking, a threat can be evaluated using a subjective or (88) an objective standard. (89) A subjective standard requires a jury to get inside the mind of a defendant and evaluate intent. (90) By contrast, under an objective standard the speaker's intent is irrelevant. Instead, what matters is whether a reasonable person would understand the statement to convey an intent to inflict bodily harm, regardless of whether or not such an intent is actually present in the mind of the speaker. (91)
The objective-subjective question can be asked with respect to a particular statute if the text and legislative history of that statute are silent on mens rea. In addition, there is a more fundamental constitutional question regarding the First Amendment implications of interpreting a statute criminalizing expression to either require or preclude an analysis of the defendant's state of mind. The constitutional question is intertwined with jurisprudence on "true threats," (92) which the Supreme Court first explicitly identified as unprotected expression in its 1969 Watts v. United States (93) decision.
Since Watts provided little further guidance, lower courts in subsequent decades struggled to define true threats. In 2003, more than thirty years after Watts, the Court provided what remains its most expansive discussion to date of what constitutes a true threat, in Virginia v. Black (94):
True threats encompass 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.... The speaker need not actually intend to carry out the threat. Rather, a prohibition on true threats protects] individuals from the fear of violence and from the disruption that fear engenders, in addition to protecting people from the possibility that the threatened violence will occur. Intimidation in the constitutionally prescribable sense of the word is a type of true threat, where a speaker directs a threat to a person or group of persons with the intent of placing the victim in fear of bodily harm or death. (95)
Unfortunately, this did not resolve the subjective-versus-objective question. As Paul Crane wrote, "in providing a definition, the Court created more confusion than elucidation." (96) Advocates of a subjective standard found their views confirmed in the Black decision's phrases "speaker means to communicate" and "speaker directs a threat... with the intent of." (97) But proponents of an objective standard viewed the Court's use of "encompass" and "a type of true threat," (98) as conveying that true threats were a larger category that included statements made with intent, but also included statements where speaker intent was irrelevant. Unsurprisingly, this divergence of interpretations was reflected in lower court jurisprudence in the years following Black. The Ninth (99) and Tenth (100) Circuits and some state supreme courts (101) employed a subjective standard, while the Third, (102) Fourth, (103) Sixth, (104) Seventh, (105) Eighth, (106) and Eleventh (107) Circuits and some other state courts of last resort (108) used an objective standard.
B. The Elonis Decision
Elonis v. United States, argued in late 2014 and decided in June 2015, concerned Facebook postings by Anthony Elonis containing threatening statements about his estranged wife, an amusement park where he was employed, an unspecified school, and an FBI agent. (109) He was charged with five counts of violating 18 U.S.C. [section] 875(c), which provides that "[w]hoever transmits in interstate or foreign commerce any communication containing any threat to kidnap any person or any threat to injure the person of another, shall be fined under this title or imprisoned not more than five years, or both."
At trial in federal district court Elonis argued that the statute should be applied using a subjective standard, through jury instructions specifying that the "government must prove that he intended to communicate a true threat." (110) The government, however, argued that "it doesn't matter what he thinks," (111) and the instructions that were actually provided to the jury stated that:
[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. (112)
After being convicted on four of the five counts, Elonis appealed to the Third Circuit, which upheld the conviction. He then petitioned the Supreme Court, arguing that the district court's use of an objective standard was improper. Elonis's petition presented the following question:
Whether, consistent with the First Amendment and Virginia v. Black, 538 U.S. 343 (2003), conviction of threatening another person requires proof of the defendant's subjective intent to threaten, as required by the Ninth Circuit and the supreme courts of Massachusetts, Rhode Island, and Vermont; or whether it is enough to show that a "reasonable person" would regard the statement as threatening, as held by other federal courts of appeals and state courts of last resort. (113)
In granting certiorari, the Supreme Court added (and ultimately only answered) a second question that was statutory as opposed to constitutional: "Whether, as a matter of statutory interpretation, conviction of threatening another person under 18 U.S.C. [section] 875(c) requires proof of the defendant's subjective intent to threaten." (114)
The Court's June 2015 decision, written by Chief Justice Roberts and joined by six other justices (and joined in part by a seventh, Justice Alito) reversed the Third Circuit. The Court addressed the lack of an explicit mens rea requirement in the text of the statute by noting that, consistent with its earlier rulings:
The fact that the statute does not specify any required mental state, however, does not mean that none exists. We have repeatedly held that "mere omission from a criminal enactment of any mention of criminal intent" should not be read "as dispensing with it." This rule of construction reflects the basic principle that "wrongdoing must be conscious to be criminal." (115)
The objective "reasonable person" standard used in the jury instructions, wrote the Court, "is a familiar feature of civil liability in tort law, but is inconsistent with 'the conventional requirement for criminal conduct--awareness of some wrongdoing.'" (116) The Court reasoned that "[h]aving liability turn on whether a 'reasonable person' regards the communication as a threat--regardless of what the defendant thinks--'reduces culpability on the all-important element of the crime to negligence,' and we 'have long been reluctant to infer that a negligence standard was intended in criminal statutes.'" (117) "Under these principles," concluded the Court, what the defendant thinks "does matter." (118) Having concluded that conviction under 18 U.S.C. [section] 875(c) did indeed require proof of the defendant's subjective intent to threaten, the Court did not reach the broader constitutional question of whether "conviction of threatening another person," outside the specific context of the statute at issue, "requires proof of the defendant's subjective intent to threaten."
The Elonis ruling was clear in stating that intent matters with respect to 18 U.S.C. [section] 875(c) and that the district court had therefore erred in providing jury instructions that did not address the defendant's state of mind. But beyond stating that a negligence standard was too low, the ruling was silent on what specific level of intent is required for conviction, leading Justice Alito to pen a separate opinion characterizing the decision as "certain to cause confusion and serious problems." (119) He continued:
Attorneys and judges need to know which mental state is required for conviction under 18 U.S.C. [section] 875(c)... but the Court provides only a partial answer. The Court holds that the jury instructions in this case were defective because they required only negligence in conveying a threat. But the Court refuses to explain what type of intent was necessary. (120)
Justice Alito argued that for conviction, "it must be shown that the defendant was at least reckless as to whether the transmission" was a threat. (121) "Someone who acts recklessly with respect to conveying a threat," explained Justice Alito, "necessarily grasps that he is not engaged in innocent conduct. He is not merely careless. He is aware that others could regard his statements as a threat, but he delivers them anyway." (122) "Accordingly," he concluded, "I would hold that a defendant may be convicted under [section] 875(c) if he or she consciously disregards the risk that the communication transmitted will be interpreted as a true threat." (123)
Justice Thomas also issued an opinion, and like Justice Alito, criticized the confusion he believed would arise from the Court's ruling:
This failure to decide throws everyone from appellate judges to everyday Facebook users into a state of uncertainty. This uncertainty could have been avoided had we simply adhered to the background rule of the common law favoring general intent.... Because the Court of Appeals properly applied the general-intent standard, and because the communications transmitted by Elonis were "true threats" unprotected by the First Amendment, I would affirm the judgment below. (124)
IV. TECHNOLOGY AND INTENT POST-ELONIS
In the Elonis decision, the Supreme Court held that, at least with respect to 18 U.S.C. [section] 875(c), intent matters. In addition, a reading of the opinion suggests that similar logic could apply to any statute that criminalizes threats (or, by extension, incitements to imminent lawless action) without also providing an explicit mens rea requirement. However, Elonis also leaves multiple questions unresolved. As Justice Alito observed, the Court declined to address the requisite level of intent for conviction under the statute in question. (125) And the Court did not reach the constitutional implications of intent.
A. Tweets, Algorithms, and Intent: Some Scenarios
In evaluating these and related questions, a set of hypothetical scenarios can help illustrate how communications technologies can add layers of complexity to the role of intent. To start, it is helpful to consider a non-technological example that provides a clear illustration of how knowledge of intent can be crucial in evaluating potentially criminal expression.
In a 2014 Columbia Law Review essay arguing that speaker intent is an "important factor in the question of whether speech is protected," (126) legal scholar Leslie Kendrick presents the example of an "unknowing inciter:" "A visitor wanders into an angry crowd while wearing a new shirt bearing a design that, unbeknownst to him, is in that context likely to trigger violence." (127) Kendrick rightly observes that "[i]t seems inappropriate to hold him responsible for the effects of his message on the actions of third parties, when he was unaware of and did not intend those effects." (128) Expression, as this example shows, can incite imminent lawless action even when a speaker's actions are entirely innocent.
Now consider a set of technology-focused scenarios, all based on variations on the theme of a tweet relating to imminent lawless action:
1. Person A intentionally composes and sends a tweet advocating imminent violence in a particular city. Person A knows that his or her Twitter followers include many people who are residents of the city in question and who have previously engaged in violent acts in that city.
This is the simplest case. Under Brandenburg, speech that "is directed to inciting or producing imminent lawless action and is likely to incite or produce such action" (129) is outside First Amendment protection. A tweet advocating imminent violence is certainly "directed to inciting or producing imminent lawless action." And, if it is knowingly sent (130) to people with a history of committing violent acts in the city, there is a strong argument that the tweet satisfies the second prong of the Brandenburg test of being "likely to incite or produce such action."
2. Person B sends a tweet praising Person A and urging people to take action based on Person A's tweets.
Is Person B's expression protected? This is a scenario where context and intent matter enormously. Under one variant of this scenario, if Person B sent her tweet after seeing Person A's violence-inciting tweet and concluding that she agreed with it and wanted to assist Person A in spurring imminent violence, Person B's tweet is unprotected speech. But consider a second variant: Suppose that Person B sent her tweet the day before Person A's tweet, and that Person A had never previously engaged in any expression advocating violence. Under that second variant, it would be wrong to subject Person B to criminal prosecution. After all, Person B's tweet, at the time it was issued, was clearly protected, and it would be unjust if Person A's later actions could lead to an after-the-fact reclassification of Person B's expression as unprotected.
Now consider a third variant: Suppose Person B sent her tweet the day after Person A's violence-inciting tweet, but without having been aware of A's tweet. Under this variant Person B would likely appear to an outside observer to be endorsing Person A's call to violence. This would be a misconception, however, arising from an unfortunate accident of timing. Knowing Person B's state of mind in this third variant, it is clear that her actions are no less innocuous than in the second variant above. Yet, absent an effort to explore that state of mind (which could be done for example, by examining records to show that she did not view the offending tweet from Person A before sending her own tweet), her risk of criminal prosecution and conviction would be much higher.
3. A bot (131) is programmed to scan all tweets and to retweet those that mention the city in question, appending "This sounds great!" to the text in the original tweet. The hot finds Person A's tweet, adds "This sounds great!" and automatically retweets it.
Under this scenario there is no question that the content of the bot's retweet was "directed to" inciting lawless action even though the bot itself clearly had no intent of any kind. And, if the bot's Twitter followers include people living in the city with a history of committing violence, the bot's retweet may be "likely to" produce lawless action. Thus, there is an argument that both prongs of the Brandenburg test are satisfied, and that the bot's retweet is unprotected expression.
But the argument should not end there. After all, the bot is just a computer running a simple program, and cannot reasonably be accused of acting with intent. The person who programmed the bot did act with intent, but the intent was innocuous: He or she was simply trying to retweet tweets mentioning the city, regardless of what other content those tweets might contain. In addition, because the programmer modified the original content before retransmitting it, he or she would arguably be outside the safe harbor provided by the Communications Decency Act of 1996 (132) and thereby exposed to prosecution.
Of the four levels of culpability in the Model Penal Code (133)--"purposeful" (also called intentional), "knowing," "reckless," and "negligent"--the highest two certainly do not apply. After all, the programmer certainly did not purposefully or knowingly cause the bot to specifically send the problematic tweet. (134) Assessing negligence and recklessness is more complex. In arguing that neither apply, the programmer would assert that he or she is blameless and cite the common practice in social networking of automatically retweeting tweets based on their inclusion of a specific keyword, without examining other content those tweets might contain.
However, there is a good counterargument that invoking "common practice" is insufficient, and that the programmer was therefore at least negligent since he or she should have known that blindly endorsing and retransmitting tweets could lead to some problematic endorsements. The failure to consider this possibility, the argument would continue, constituted negligence since it "involve [d] a gross deviation from the standard of care that a reasonable person would observe." (135) At least under tort law, the programmer would therefore be exposed to liability since a person who suffered harm due to violence spurred by the bot's tweet would have a cause of action.
Whether the programmer was also exposed to criminal liability would depend on the state of mind specified in (or inferred from) the statute in question. (136) A statute requiring purposeful or knowing action would not apply for the reasons discussed above. If, as implied by Elonis, negligence alone would be insufficient for criminal liability, (137) that leaves recklessness, which in turn would depend in part on the specific definition applied. It could be difficult to show that, as required under some definitions, (138) the programmer was aware of and consciously disregarded the risk. However, under a definition tied to the programmer's indifference regarding the consequences of his or her actions, recklessness would be easier to establish. (139)
4. A bot is programmed to compose tweets from scratch (as opposed to retweeting tweets composed and broadcast by other parties). The person who programmed the bot intended to cause it to issue tweets advocating peaceful street protests in response to a real-time analysis of news events. However, due to insufficiently careful programming, the bot issues a tweet that advocates violent protest.
Under this scenario the programmer was at least negligent. (140) After all, the programmer should have been aware of the risk that a computer programmed to compose tweets urging nonviolent protests might, due to insufficient care in programming, end up composing a tweet urging violent protests. But if negligence is not sufficient to establish liability under a statute criminalizing unprotected expression, were the programmer's actions reckless? As with scenario number three, the answer depends on the definition of recklessness applied as well as the programmer's state of mind as he or she developed the program.
One important source for insight into the programmer's state of mind is the program itself, and in particular the algorithms it employs to construct tweets. The design of algorithms that output phrases of text can span a spectrum. At one end of the spectrum, all possible output phrases are pre-written by a programmer, and the algorithm simply selects from this library of predetermined choices each time an output is generated. For this class of algorithms it is easy to avoid the possibility of a violence-inciting output by simply ensuring that no such phrases are present in the program's library of phrases.
At the other end of the spectrum are algorithms that compose sentences more like a human would, drawing from a vocabulary of individual words to assemble phrases based on word definitions and syntactic rules. This latter approach clearly offers the possibility for much more variety and creativity, but is also more complicated to program and involves greater risk of generating an output that is somehow mismatched to the input information. A programmer writing an algorithm to compose tweets from scratch would be well advised to add a module, downstream from the tweet-composing stage but upstream of the output, to ensure that any exhortations to take action are filtered to remove those that might be interpreted as advocating imminent violence. Failure to do this, particularly if the available vocabulary contains words that could incite violence if misused, could rise to the level of recklessness. (141)
5. Person D, who has repeatedly been arrested for vandalism in the past in a particular city, sees news reports that vandalism is occurring in the city and sends a tweet stating "Lots of vandalism occurring right now in [city name]; come down and play your role."
Tweeting only "Lots of vandalism occurring right now in [city name]" would clearly be protected, as that would simply be conveying a factual statement. But when the second clause, "come down and play your role," is added, the statement becomes more ambiguous, and context and intent are key to determining with this speech is protected. Suppose that, after issuing the tweet, Person D uses his mobile phone to call several friends, who then join him in committing acts of vandalism. With that additional context, the tweet can be reasonably understood as an incitement to imminent lawless action. But instead, suppose that Person D has recently foresworn crime and become active in promoting peaceful interventions to improve the community? Furthermore, suppose that after sending the tweet, he calls like-minded people and asks them to go to the area of unrest with him and talk people out of committing vandalism? In that case, the tweet is almost certainly not issued with intent to incite imminent lawless action.
This scenario also creates an additional possibility: Person D could have issued the tweet in good faith to try to stop the vandalism, but some people who saw the tweet instead could have nonetheless viewed it as a call to commit vandalism and acted accordingly. If that were to occur, Person D could be reasonably accused of negligence (under the argument that he should have known, given his past history, that his message would potentially be interpreted by some as a call to commit vandalism) and thereby be exposed to civil liability. Whether his actions rose to the level of recklessness would depend, as in scenario three, on factors including the specifics of Person D's state of mind and on the definition of recklessness applied.
B. Intent, Technology, and the First Amendment
As the above examples illustrate, context is key. And while context has mattered as long as there has been communication, technology and the culture surrounding its use have complicated matters. (142) Stated another way, intent, technology, and the First Amendment create a three-way intersection that individual Supreme Court rulings to date have considered only in part. The Elonis decision, for example, considered intent and (to a limited extent) (143) technology, but not the First Amendment. (144) Other cases have considered other portions of this intersection.
In Virginia v. Black the Court considered a Virginia statute criminalizing cross-burning and concluded that "while a State, consistent with the First Amendment, may ban cross burning carried out with the intent to intimidate, the provision in the Virginia statute treating any cross burning as prima facie evidence of intent to intimidate renders the statute unconstitutional in its current form." (145) This was a clear endorsement of the role of intent (and context) in setting the bounds of expression protected by First Amendment, though in a completely non-technological context.
In the 1997 Reno v. ACLU (146) decision, the Court considered technology and the First Amendment but not intent. Reno arose from a challenge to the anti-indecency provisions of the Communications Decency Act (CDA) of 1996. The Court found those provisions to be unconstitutional, writing that "[i]n order to deny minors access to potentially harmful speech, the CDA effectively suppresses a large amount of speech that adults have a constitutional right to receive and to address to one another." (147) Adopting language from the district court's ruling, the Court concluded that "'the content on the Internet is as diverse as human thought.' We agree with [the district court's] conclusion that our cases provide no basis for qualifying the level of First Amendment scrutiny that should be applied to this medium." (148)
It is tempting to conclude that in combination these cases largely put the issue to rest: Intent matters in defining the limits of constitutionally protected expression, including when that expression is facilitated by technology. That is a correct conclusion, but it is one that is incomplete. After all, simply stating that intent matters leaves open the question of what level of intent should be required (or proscribed) in prosecutions for allegedly unprotected expression. More fundamentally, it is a conclusion that, despite the rulings in Elonis and Reno, is at risk due to several factors--one technological and the other societal.
First, while Reno is a recent ruling on a jurisprudential time scale, in technological terms it is practically ancient history. Reno was argued and decided in an era when the Internet was still novel to many Americans. While dialup modem-based email services had become increasingly popular through the 1980s and early 1990s, it was not until in the mid-1990s, following the release of the Mosaic browser from Netscape Communications, that large numbers of Americans began to experience web browsing. The language in the Court's ruling captured a snapshot of the circa-1997 Internet that seems quaint today. The ruling noted, for example, that the number of Internet hosts had experienced "extraordinary growth," increasing from "about 300 in 1981 to approximately 9,400,000 by the time of the trial in 1996." (149) Today, by comparison, there are over one billion hosts on the Internet. (150) The Court also wrote that "[a]bout 40 million people used the Internet at the time of trial;" (151) today, that number has grown to well over 3 billion. (152)
Social networking sites as we think of them today did not exist in the mid-1990s. Mark Zuckerberg, who would go on to found Facebook in 2004, (153) was still in middle school when Reno was decided. Even Myspace, which preceded Facebook and would grow to briefly become the top social networking site, (154) was not founded until 2002. In 1997, Google had not yet been founded; instead, people performed web searches using search engines like Excite, AltaVista, and Lycos. And while mobile phones offering Internet access became commercially available in the late 1990s,155 it was not until well into the 2000s that wireless networks and mobile phone technology improved enough to enable design, production, and widespread adoption of smartphones more closely resembling those in widespread use today.
Depending on one's point of view, the enormous technology changes since Reno can be invoked to argue for increasing, decreasing, or retaining the role of intent in assessing whether expression is protected. This Article takes the position that, if anything, the greater potential for decontextualization given all the technological changes since 1997 creates even more of a need to consider intent when prosecuting allegedly unlawful expression. For all of the reasons explained above, the speed, complexity, and nature of digital communications have broadened the ways in which a speaker's original intent can decoupled from expression.
The natural reaction of prosecutors will be to oppose requirements to prove intent, just as they did in Elonis, because it raises their burden. But it does not raise it very much. Today, nearly everyone, including anyone who might be accused of engaging in criminal expression, leaves a digital trail, and usually an extensive one. For crimes involving threats or incitements to imminent lawless action, in addition to the allegedly unlawful communication itself, there will often be a wealth of other accompanying digital evidence that can be examined to shed light on the defendant's state of mind. As Professor Jennifer Rothman noted in proposing an "intent prong" for evaluating threats, factors to be considered can include "whether it was likely that the threat would reach the target or his associates," "whether the speaker told others of his intentions," and "whether the speaker made prior threats." (156) Given the decline in storage costs and the growth in cloud-based services over the last fifteen years, all of these factors are much easier to examine today than in 2001, since far more communications are stored for longer periods of time, and stored in a manner (such as in the cloud-based digital backups of social media companies) that are more protected from after-the-fact deletion by a defendant and therefore accessible to a prosecutor in possession of a warrant.
It is also important to acknowledge some counterarguments. Technology has made it easier to issue statements such as threats more quickly and more anonymously than before. In addition, including intent in the inquiry gives defendants an incentive to claim that their statements are "works of art," just as Elonis did. (157) As Justice Alito observed, this could "grant a license to anyone who is clever enough to dress up a real threat in the guise of rap lyrics, a parody, or something similar." (158) These are reasonable concerns, and the proper response is to make sure the level of intent is not set so high as to make conviction for criminal threats (and incitements to imminent lawless action, etc.) a practical impossibility. (159) In addition, there are various steps that could be taken to assess the credibility of a defendant's "works of art" claim including the specificity of the threat, its likely meaning in light of the defendant's other statements and actions, and the defendant's prior history. A person who had never previously published online lyrics but suddenly claimed, in response to a criminal threat prosecution, that he had become a lyricist at the moment the allegedly threatening statements were issued would not have much credibility.
There is also a second, societal factor to consider. Giving insufficient consideration to the role of intent (or dispensing with it altogether) in criminal prosecutions involving allegedly unprotected expression would have an impact that goes far beyond the criminal justice system. If speaker intent is deemed to be unimportant in court, it may be deemed unimportant in many other places, and legitimate speech will be chilled accordingly. Even after the Supreme Court confirmed the importance of intent in Elonis, there is evidence that purely objective standards for proscribing speech are alive and well, particularly on public university campuses. (160) For instance, in November 2015, University of Missouri police issued an e-mail to the university community asking people who witness incidents of "hurtful speech" to call police "immediately." (161) The e-mail went on to note that if the speakers were students, the university's "Office of Student Conduct can take disciplinary action." (162) It is alarming that a public university police department would ask people to report "hurtful" speech, which could include an enormous range of unambiguously protected but potentially "hurtful" statements, such as "your hometown sports team is no good." And it is noteworthy that the phrasing of the request from the university police, using "hurtful," suggested an objective standard that places primacy on a listener's interpretation instead of on the intent of the speaker.
It is tempting to dismiss the growth in these sorts of speech infringements, applied without consideration of speaker intent, as a phenomenon of concern primarily with respect to universities--more particularly public universities, which have constitutional obligations. But as attorney Wendy Kaminer has written:
[W]hat happens on campus doesn't stay on campus. Students graduate. They become faculty members, administrators, political candidates, and state or federal regulators. Considering the fact that campus speech restrictions and a culture of censorship date back decades, it's not surprising that some policy-makers and opinion leaders approaching middle age, as well as students, support restrictions on whatever they deem hateful, bigoted, or generally offensive speech. (163)
Kaminer's observation that these restrictions are based on an objective standard ("whatever they deem hateful, bigoted, or generally offensive") underscores the extent to which intent is often absent from the inquiry. In short, issues of when and how to limit expression--and what role speaker intent should play in those decisions--impact enormous numbers of people, not just those in the criminal justice system and not just those on university campuses.
Based on Supreme Court jurisprudence, it is tempting to assume that the central role of intent in delimiting the scope of protected expression--including technology-facilitated expression--is not at risk. This, after all, is a reasonable conclusion suggested by an examination of the Supreme Court decisions in Reno v. ACLU in 1997, Virginia v. Black in 2003, and Elonis v. United States in 2015. But those decisions are certainly not the last word. Future cases, including some that reach the Supreme Court, will likely include arguments noting (correctly) that the communications technology landscape has changed immensely since Reno v. ACLU, and asserting (incorrectly) that under a full analysis of the complexities of today's technology environment, which the Supreme Court did not perform in Elonis, a narrowing of First Amendment freedoms is needed. Along precisely those lines, one legal scholar has recently opined that, in light of technology changes, "Those who regard free speech as fundamental need to consider whether legal principles that arose centuries ago make sense in the age of Snapchat." (164)
This Article takes the position that, to put it mildly, freedom of expression still "makes sense" today, and ensuring that freedom, particularly in light of digital communications technology advances that show no sign of slowing, requires proper consideration of another legal principle from centuries ago: mens rea. Expression in the next decade, and in the decade after that, will be even more complex, faster, more efficient, and more easily decontextualized than what we have today. Algorithms and artificial intelligence used in communications will also become much more powerful. These changes will continue to broaden the ways in which a speaker's original intent can be left behind, even as his or her expression propagates through cyberspace. And technology-enabled communication will likely advance in ways that are counterintuitive. After all, few people in the 1990s could have imagined that after the turn of the century, as bandwidth and storage became more plentiful than ever before, a service (Twitter) limiting messages to only 140 characters would play a major role in social movements in the United States and worldwide. (165)
It would be unfortunate if the remarkable communications advances of the twenty-first century were to lead to a narrowing of First Amendment protections. To avoid that outcome, this Article has argued that communication technology advances have made it more important than ever to examine speaker intent when criminalizing expression. Proper consideration of intent can protect speakers from being punished for lawful speech, and it should be an important component of protecting those targeted by unlawful speech. The same technology trends that are making communications so much more pervasive and efficient are also making it easier to examine intent, and thereby to protect people targeted by expression falling outside the bounds of the First Amendment.
(1.) Alex Cocotas, Kids Send A Mind Boggling Number Of Texts Every Month, BUSINESS INSIDER (Mar. 22, 2013), http://www.businessinsider.com/chart-of-the-day-number-of-texts-sent-2013-3 [https://perma.cc/3F27-WJ3A].
(2.) Id. As reported in the survey, adults aged twenty-five to thirty-four sent an average of thirty-seven daily text messages; those aged thirty-five to forty-four sent an average of twenty-seven daily texts; those aged forty-five to fifty-four sent an average of seventeen daily texts; and those fifty-five and older sent an average of eight daily texts.
(3.) For a summary and analysis of automated tweets, see Stefanie Haustein et al., Tweets as Impact Indicators: Examining the Implications of Automated Bot Accounts on
Twitter, ARXIV.ORG, http://arxiv.org/abs/1410.4139 [https://perma.cc/YK3P-XQJJ] (last visited Jan 3, 2016).
(4.) "Speaker" as used in this Article has a broad meaning, generally referring to a person who engages in expression, whether verbally, in writing, or through some other means.
(5.) "Active" refers to "monthly active users"--those who use a service at least once every month.
(6.) See Number of Monthly Active Twitter Users in the United States From 1st Quarter 2010 to 4th Quarter 2015 (in Millions), Statista, http://www.statista.com/statistics/274564/monthly-active-twitter-users-in-the-united-states/ [https://perma.cc/HE3U-EZJ6] (last visited Mar. 5, 2016). Of course, as discussed infra, not all of these users are people. Some Twitter accounts are associated with companies or other entities.
(7.) The statement that tweets can be "read by anyone in the world with an Internet connection" refers to "public tweets." Twitter also offers an option of setting up an account to send "protected tweets," which are visible only to followers approved by the sender. See About Public and Protected Tweets, TWITTER, https://support.twitter.com/articles/14016 [https://perma.cc/F8L5-VCYW] (last visited Feb. 6, 2016).
(8.) As of late 2015, Instagram had about 400 million monthly active users, of whom about 25 percent were located in the U.S. See Arjun Kharpal, Facebook's Instagram Hits 400M Users, Beats Twitter, CNBC (Sept. 23, 2015), http://www.cnbc.com/2015/09/23/instagram-hits-400-million-users-beating-twitter.html [https://perma.cc/93WL-3J2Q].
(9.) Number of Monthly Active Facebook Users in the United States and Canada as of 4th Quarter 2 015 (in Millions), STATISTA, http://www.statista.com/statistics/ 247614/number-of-monthly-active-facebook-users-worldwide/ [https://perma.ee/ D9DA-KNDH] (last visited Mar. 5, 2016). These data show 219 million monthly active Facebook users in the United States and Canada as of Q4 2015. Prorating based on the relative populations of the United States (approximately 290 million people age fifteen and over) and Canada (approximately thirty-two million people age fifteen and over) gives an estimate of just under 200 million United States monthly active Facebook users.
(10.) Archiving is not limited only to digital expression. Verbal expression in public spaces is also much more likely than in the past to be recorded, and therefore more easily available for after-the-fact scrutiny.
(11.) 135 S. Ct. 2001 (2015).
(12.) The Court reversed the Third Circuit, holding that, for conviction under 18 U.S.C. [section] 875(c), what the defendant thinks "does matter." Id. at 2011.
(13.) The Elonis ruling stated that a negligence standard was improper, writing: "we 'have long been reluctant to infer that a negligence standard was intended in criminal statutes.'" Id. (quoting Rogers v. United States, 422 U.S. 35, 47 (1975)). However, the Court declined to specify which standard--for example, recklessness, knowledge, or purpose--should apply.
(14.) To be clear, this Article is not arguing that the statements for which Elonis was criminally charged deserve First Amendment protection. Had the jury in his trial been properly instructed regarding mens rea, he may well have been convicted in a manner that would have withstood Supreme Court review.
(15.) See, e.g., Notable & Quotable: Unfree Speech on Campus, WALL St. J. (Oct. 22, 2015), http://www.wsj.com/articles/notable-quotable-unfree-speech-on-campus-1445555707 [https://perma.cc/E7FG-WHY3] (reporting that in a recent survey of 800 college students across the country commissioned by the William F. Buckley Jr. Program at Yale, "[o]ne-third of the students polled could not identify the First Amendment as the part of the Constitution that dealt with free speech. Thirty-five percent said that the First Amendment does not protect 'hate speech,' while 30 percent of self-identified liberal students say the First Amendment is outdated").
(16.) For example, defamation raises its own distinct complexities regarding intent, which could be further complicated given the rapid remixing and redistribution of content that often occurs on social networks. For public figures (including but not limited to public officials), establishing defamation requires showing "actual malice." See Gertz v. Robert Welch, Inc., 418 U.S. 323, 345 (1974); N.Y. Times, v. Sullivan, 376 U.S. 254, 279-80 (1964). For private persons, a negligence standard is often used. See, e.g., RESTATEMENT (SECOND) OF TORTS [section] 558 (AM. LAW INST. 1977) ("To create liability for defamation there must be [among other things]... fault amounting at least to negligence on the part of the publisher."). While the threshold mental state for establishing liability in relation to defamation can be lower than that required in criminal prosecutions related to other forms of expression (for instance, threats or incitements to imminent lawless action), the underlying premise of this Article--that speaker intent matters, and particularly so in light of the complexities introduced by modern communications technology--applies in both criminal and civil contexts.
(17.) While the Internet was invented in the 1960s, see Elizabeth Palermo and Lauren Cox, Who Invented the Internet?, LIVESCIENCE (Jan. 15, 2014), http://www.livescience.com/42604-who-invented-the-internet.html [https://perma.cc/T87W-M2G6], and the first commercial cellular telephone services began in the mid-1980s, see Pace Quickens In Cellular Radio Market, N.Y. TIMES (Feb. 14, 1984), http://www.nytimes.com/1984/02/14/business/pace-quickens-incellular-radio-market.html [https://perma.cc/4JDA-BRQR], it was not until the 1990s that mobile phones and the Internet began to experience mass commercial adoption. This Article focuses largely on the communications and related technologies that have experienced widespread adoption in the United States in the past two decades, including the Internet, increasingly powerful search engines, basic mobile phones and later smartphones, social networking, cloud computing, and the increasing use of algorithms and artificial intelligence in relation to digital communications.
(18.) Of course, in the pre-digital era there was also plenty of opportunity for people to actively curate the communications they would receive, for example, through choosing to subscribe to a particular newspaper or magazine, or to tune a television or radio to a particular station.
(19.) "Instagram Direct" allows a sender to send a photo to one (or more) people selected from a list of followers. See Introducing Instagram Direct, INSTAGRAM (Dec. 12, 2013), http://blog.instagram.com/post/69789416311/instagram-direct [https://perma.cc/2WTB-MNHY].
(20.) See, e.g., Slater Tow, Improved Friend Lists, FACEBOOK (Sep. 13, 2011, 12:59 PM), https://www.facebook.com/notes/facebook/improved-friend-lists/10150278 932602131/ [https://perma.cc/4JS8-XDD4].
(21.) A public tweet is viewable by anyone, but it is more likely to be viewed by the followers of the person (or machine) doing the tweeting, than by people who are not followers.
(22.) See About Public and Protected Tweets, TWITTER, https://support.twitter.com/ articles/14016 [https://perma.cc/F8L5-VCYW] (last visited Feb. 6, 2016).
(23.) A tweet containing the text "@username" will result in an autogenerated message to the person with the account name "username" stating that he or she has been mentioned in the tweet.
(24.) On Twitter and other social networks, "hashtags," which are words preceded by a "#" sign, are often included in messages to help people find posts related to a topic of interest. For example, social networking posts about the movie Star Wars: The Force Aivakens often include the hashtag #theforceawakens.
(25.) Among other things, variations in how data is routed raise important issues regarding extraterritoriality. Consider the example of two people, both located outside the United States, who are exchanging electronic communications. Those communications might be carried solely within that country using facilities and services owned by non-U.S. entities. Alternatively, those communications might physically pass through the United States (if, for example, the communications are utilizing a cloud services provider that passes all traffic through a servicer in the United States), or they might stay outside the United States but nonetheless pass through overseas servers managed by a U.S. entity. In the event that the communications were allegedly unlawful, each of these scenarios would raise distinct legal issues.
(26.) U.S. states with Google data centers include Alabama, Georgia, Iowa, North Carolina, Oklahoma, Oregon, and South Carolina. International Google data center locations include Belgium, Chile, Finland, Ireland, the Netherlands, Singapore, and Taiwan. See Data Center Locations, Google, https://www.google.com/about/ datacenters/inside/locations/index.html [https://perma.cc/CEZ9-DXWF] (last visited Dec. 27, 2015).
(27.) The Newest Addition to the Facebook Data Center Fleet: Fort Worth, FACEBOOK (July 7, 2015), http://newsroom.fb.com/news/2015/07/the-newest-addition-to-the-facebook-data-center-fleet-fort- worth/ [https://perma.cc/SFN4-38FD].
(28.) An Amazon Web Services "availability zone" is located within a specific geographic region and contains "one or more discrete data centers, each with redundant power, networking and connectivity, housed in separate facilities." AWS Global Infrastructure, Amazon Web Services, https://aws.amazon.com/ about-aws/global-infrastructure/ [https://perma.cc/K7FU-MLRF] (last visited Dec. 27, 2015).
(29.) See, e.g., Laurence Zuckerman, With Internet Cachet, Not Profit, A New Stock Is Wall St.'s Darling, N.Y. TIMES (Aug. 10, 1995), http://www.nytimes.com/1995/ 08/10/us/with-intemet-cachet-not-profit-a-new-stock-is-wall-st-s-darling.html [https://perma.cc/TY49-3YFE].
(30.) An algorithm is a set of steps that move towards a goal. Algorithms are not necessarily complex, and they are not necessarily implemented on computers. For example, cooking recipes are examples of algorithms. In the context of computing and communications technology, algorithms are often used to analyze enormous amounts of data, extract patterns or information that would not have been easily accessible without the aid of a computer, and to take action based on that information. See John Villasenor, In Defense of Algorithms, SLATE (Dec. 1, 2015), http://www.slate.com/articles/technology/future_tense/2015/12/in_defense_of_the _algorithms_that_guide_tasks_technical_and_mundane.html [https://perma.cc/ ZPE6-Q4CS],
(31.) Artificial intelligence (AI) is a broad term that admits multiple definitions. As used here, AI refers to techniques in which computers engage in tasks often associated with the concept of intelligence, including learning based on past experience and, more generally, creating new algorithms as opposed to being limited to running the specific algorithms they were originally programmed to perform.
(32.) A "bot", short for robot, is a computer program that performs a task--typically one that is highly repetitive--in a manner that interacts with a digital service much as a human might, though typically much faster. With respect to a service like Twitter, bots can be used to automatically compose and issue new tweets or to retweet tweets sent by others.
(33.) A retweet "is a re-posting of someone else's [tjweet.... Retweets look like normal Tweets with the author's name and username next to it/' but also include "the name of the user who retweeted the [tjweet." See FAQs about Retweets (RT), TWITTER, https://support.twitter.com/articles/77606 [https://perma.cc/H2KGCPKL] (last visited Feb. 6, 2016).
(34.) Jeff Haden, 8 Great Twitter Tools You Aren't Using, INC. (May 30, 2012), http://www.inc.com/jeff-haden/8-great-twitter-tools-you-arent-using.html [https://perma.cc/4LVK-VX5T].
(35.) Matt Egan, Robots write thousands of news stories a year, but not this one, CNNMONEY (June 11, 2015, 10:08 AM), http://money.cnn.com/2015/06/ll/media/ robots-journalists-media-jobs/index.html [https://perma.cc/NDJ3-95N9].
(36.) AeroVironment tops Street 2Q forecasts, YAHOO! FINANCE (Dec. 8, 2015, 4:18 PM), http://finance.yahoo.com/news/aerovironment-tops-street-2q-forecasts211838062.html [https://perma.cc/D9ET-AQBC].
(38.) Id. Automated Insights, Inc., is a company that provides software that can be used to automatically generate textual descriptions on a range of topics, including crime trends, real estate descriptions, and election results. Examples can be seen at https://automatedinsights.com/examples/ [https://perma.cc/VW4B-VCYK] (last visited Feb. 6, 2016). Another company that provides solutions for creating auto-generated articles is Narrative Science Inc.
(39.) David Koenig, FedEx tops Street 2Q Forecasts, helped by online shopping, ASSOCIATED PRESS (Dec. 16, 2015, 6:33 PM), http://bigstory.ap.org/arti cle/d338ca9b3ef950be7082c8925bb07a9c/fedex-tops-street-2q-forecasts-helpedonline-shopping [https://perma.cc/RM9B-3TJV].
(41.) See Watts v. United States, 394 U.S. 705 (1969).
(42.) See Brandenburg v. Ohio, 395 U.S. 444 (1969per curiam).
(43.) Mens rea has been a recurring topic in law review publications. See, e.g., Eugene J. Chesney, Concept of Mens Rea in the Criminal Law, 29 Am. Inst. CRIM. L. & CRIMINOLOGY 627 (1939discussing the historical evolution of mens rea in English, Roman, and American law); Paul T. Crane, "True Threats" and the Issue of Intent, 92 Va. L. Rev. 1225, 1277 (2006arguing that "only when the speaker has the specific intent to threaten should he be punished for making a true threat"); Lawrence Crocker, Justice in Criminal Liability: Decriminalizing Harmless Attempts, 53 OHIO St. L. J. 1057, 1059 (1992arguing that the "actual imposition [of a potential criminal] upon his victim or society" is more important in evaluating criminal liability than the defendant's state of mind); George P. Fletcher, The Theory of Criminal Negligence: A Comparative Analysis, 119 U. PA. L. Rev. 401 (1971discussing negligence in criminal liability); Eric A. Johnson, Rethinking the Presumption of Mens Rea, 47 WAKE FOREST L. REV. 769 (2012discussing the application of mens rea to the multiple elements in a criminal statute); Dan M. Kahan, Ignorance of Law Is an Excuse--But Only for the Virtuous, 96 MICH. L. REV. 127 (1997discussing the role of morality in law, particularly in relation to prosecutions in which the defendant was unaware of a relevant criminal statute); Leslie Kendrick, Essay, Guilty Minds, 114 COLUM. L. REV. 1255, 1295 (2014writing that speaker intent matters, and that "this conclusion can be explained by an autonomy-based account of freedom of speech"); Jennifer E. Rothman, Freedom of Speech and True Threats, 25 HARV. J.L. & PUB. POL'Y 283, 289 (2001proposing the addition of an "intent prong" to the [widely-used-in-2001] reasonable listener test); Francis B. Sayre, Public Welfare Offenses, 33 COLUM. L. Rev. 55 (1933explaining the historical context of mens rea and coining the term "public welfare offenses" to describe crimes prosecuted without consideration of state of mind); Leonid Traps, "Knowingly" Ignorant: Mens Rea Distribution in Federal Criminal Law after Flores-Figueroa, 112 COLUM. L. REV. 628, 628-29 (2012discussing the "distributive" model of mens rea, involving applying "the mens rea term in a statute to every subsequent element of the offense" in light of Flores-Figueroa v. United States, 556 U.S. 646 (2009)); Douglas R. Young, Rethinking the Specific-General Intent Doctrine in California Criminal Law, 63 CAL. L. REV. 1352 (1975discussing specific and general intent in California law); Federal Threats Statute--Mens Rea and the First Amendment--Elonis v. United States, 129 HARV. L. REV. 331, 339 (2015writing that "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").
(44.) Edward Coke, The Third Part of the Institutes of the Laws of England 6,107(1669).
(45.) Fowler v. Padget, (1798) 101 Eng. Rep. 1103, 1106 (T.R.).
(46.) 4 William Blackstone, Commentaries *21.
(47.) 342 U.S. 246 (1952). Morissette involved a scrap iron seller who collected spent U.S. government shell casings he believed to be abandoned. He was convicted under a statute (18 U.S.C. [section] 641) prohibiting "knowingly convert[ing]" government property to his use. The Supreme Court reversed, finding that the jury had not been properly instructed to consider Morissette's intent in taking the casings. See Morissette, 342 U.S. at 275.
(48.) Morissette, 342 U.S. at 251.
(49.) Id. at 251-252.
(50.) Id. at 250.
(51.) 471 U.S. 419.
(52.) Id. at 433.
(53.) 511 U.S. 513 (1994).
(54.) Id. at 524.
(55.) 513 U.S. 64 (1994).
(56.) Id. at 73.
(57.) 511 U.S. 600 (1994).
(58.) Id. at 619.
(60.) 530 U.S. 255 (2000).
(61.) Id. at 269 (quoting X-Citement Video, Inc., 513 U.S. at 72).
(62.) 135 S. Ct. at 2010 (quoting Carter v. United States, 530 U.S. 255, 269 (2000)).
(63.) The growth in "public welfare statutes" occurred not only in the United States but also in England. See Francis B. Sayre, Public Welfare Offenses, 33 COLUM. L. Rev. 55, 56, 63 (1933).
(64.) Id. at 55.
(65.) Id. at 56.
(66.) Id. at 68.
(67.) Technological change was not the only contributing factor in the growth of strict liability crimes, nor was it the original one. Sayre traces the growth of public welfare offenses to mid-nineteenth century laws relating to sales of liquor and adulterated food. See id. at 63-66.
(68.) Id. at 68-69. Two decades later, the Supreme Court's dicta in Morissette provided a similar explanation for the growth of strict liability crimes:
The industrial revolution multiplied the number of workmen exposed to injury from increasingly powerful and complex mechanisms, driven by freshly discovered sources of energy, requiring higher precautions by employers. Traffic of velocities, volumes and varieties unheard of came to subject the wayfarer to intolerable casualty risks if owners and drivers were not to observe new cares and uniformities of conduct. Congestion of cities and crowding of quarters called for health and welfare regulations undreamed of in simpler times. Wide distribution of goods became an instrument of wide distribution of harm when those who dispersed food, drink, drugs, and even securities, did not comply with reasonable standards of quality, integrity, disclosure and care. Such dangers have engendered increasingly numerous and detailed regulations which heighten the duties of those in control of particular industries, trades, properties or activities that affect public health, safety or welfare.
Morissette v. United States, 342 U.S. 246, 253-54 (1952). In light of these changes, "courts have turned to construing statutes and regulations which make no mention of intent as dispensing with it and holding that the guilty act alone makes out the crime." Id. at 256.
(69.) Sayre recognized this, writing that if faced with "inundations of cases of petty violations, the lower criminal courts would be physically unable to examine the subjective intent of each defendant, even were such determination desirable. As a matter of fact it is not; for the penalty in such cases is so slight that the courts can afford to disregard the individual in protecting the social interest." Sayre, supra note 63, at 69-70.
(70.) Jack Thompson, Bobby Unser Convicted On Wilderness Law, CHI. Trib. (June 13, 1997), http://articles.chicagotribune.com/1997-06-13/sports/9706130068_1_ bobby-unser-fined-federal-judge [https://perma.cc/RU6K-KD5K]. Unser was convicted under 16 U.S.C. [section] 551 and 36 C.F.R. [section] 261.16(a). He was fined $75, though 16 U.S.C. [section] 551 allows for penalties of "a fine of not more than $500 or imprisonment for not more than six months, or both." Unser appealed to the U.S. Court of Appeals for the Tenth Circuit, which upheld the conviction. United States v. Unser, 165 F.3d 755 (10th Cir. 1999), cert, denied, 528 U.S. 809 (1999).
(71.) Under 16 U.S.C. [section] 1372 it is generally unlawful to "take any marine mammal in waters or on lands under the jurisdiction of the United States." However, there are "Exemptions for Alaskan Natives" codified at 16 U.S.C. [section] 1371(b). The Alaska man was a Native American and thus permitted under the law to take the otters. However, his subsequent resale to a non-Native American led to his prosecution. The fact that the seller believed he was selling to a Native American in compliance with the law was deemed irrelevant given the lack of a mens rea requirement in the statute. He pled guilty, paid a $1,000 fine, and received two years of probation. See Gary Fields and John R. Emshwiller, As Federal Crime List Grows, Threshold of Guilt Declines, WALL ST. J. (Sept. 27, 2011), http://www.wsj.com/articles/ SB10001424053111904060604576570801651620000 [https://perma.cc/93ZZ-7VDR].
(72.) John Villasenor, If You Accidentally Break the Law, Are You a Criminal?, VICE NEWS (Dec. 16, 2015), http://www.vice.com/read/if-you-accidentally-break-the-law-are-you-a-criminal [https://perma.cc/3C9G-3GGT ]. The Migratory Bird Treaty Act, for example, makes it unlawful to "transport or cause to be transported ... any migratory bird" but makes no mention of the intent of the person doing the transporting. See 16 U.S.C. [section] 703(a2012).
(73.) See generally HARVEY A. SlLVERGLATE, THREE FELONIES A DAY: How THE Feds Target the Innocent (2009).
(74.) The United States, which has less than 4.5% of the world's population, now has over 20% of the world's prison population. See Michelle Ye Hee Lee, Does the United States really have 5 percent of the world's population and one quarter of the world's prisoners?, WASH. POST (Apr. 30, 2015), https://www.washingtonpost.com/ news/fact-checker/wp/2015/04/30/does-the-united-states-really-have-five-percentof-worlds-population-and-one- quarter-of- the-worlds-prisoners/ [https://perma.cc/ 6YVM-FW22]. An April 2014 Congressional Research Service report found that the "number of inmates under the [Bureau of Prisons'] jurisdiction has increased from approximately 25,000 in FY1980 to over 219,000 in FY2013." See NATHAN James, Cong. Research Serv., R42937, The Federal Prison Population Buildup: Overview, Policy Changes, Issues, and Options 2 (2014). As Alexandra Natapoff has observed, an additional factor contributing to high incarceration rates is that "due process and rule-of-law wane at the bottom of the penal pyramid where offenses are pettiest and defendants are poorest" and misdemeanants are "routinely incarcerated--despite the absence of a prison sentence--as they wait in jail for their cases to be adjudicated." Alexandra Natapoff, Misdemeanors, 85 S. Cal. L. Rev. 1313, 1316 (2012).
(75.) The House Judiciary Committee created a bipartisan task force to study over-criminalization in 2013. See Press Release, U.S. House of Representatives Committee on the Judiciary, House Judiciary Committee Reauthorizes Bipartisan Over-Criminalization Task Force (Feb. 5, 2014), available at https://judiciary.house.gov/press-release/house-judiciary-committee-reauthorizesbipartisan-over-criminalization- task- force/ [https://perma.cc/D8T6-7NH4].
(76.) H.R. 4002, 114th Cong. (2015), https://www.govtrack.us/congress/bills/114/ hr4002 [https://perma.cc/2EMS-S3SK] (last visited Feb. 6, 2016).
(77.) Id. [section] 11.
(78.) S. 2298, 114th Cong. (2015), https://www.govtrack.us/congress/bills/114/ s2298 [https://perma.cc/9S8Y-H7DE] (last visited Feb. 6, 2016).
(79.) S. 2298 contains a complex list of exceptions. One of the exceptions is in essence a safety valve providing that the "willfully" standard would not apply if doing so "would lessen the degree of mental culpability that the Government is required to prove with respect to that element under... precedent of the Supreme Court." Id. [section] 2.
(80.) Id. "Willfully" as defined in S. 2298 "means that the person acted with knowledge that the person's conduct was unlawful." Id.
(81.) Given the incompatibility between the specific mens rea provisions in the two bills, in their current form only one of the bills could become law.
(82.) The Obama Administration has been strongly opposed to mens rea reform initiatives. A Justice Department spokesperson said the House bill (H.R. 4002) "would create confusion and needless litigation, and significantly weaken, often unintentionally, countless federal statutes." Zach Carter, House Bill Would Make It Harder To Prosecute White-Collar Crime, HUFFINGTON POST (Nov. 15, 2015), http://www.huffingtonpost.com/entry/white-collar-crime-prosecution_ 564a2336e4b06037734a2f84?fspxpqfr [https://perma.cc/E7RF-WFCB]. A White House official said the bill "would undermine public health and safety, including laws that protect our environment and ensure food and drug safety." Zach Carter, White House Comes Out Against Effort To Block White-Collar Crime Prosecutions, HUFFINGTON POST (Nov. 19, 2015), http://www.huffingtonpost.com/entry/white collar-crime-white-house-response_564dd06be4b00b7997f95240 [https://perma.cc/ VA7X-U9YQ].
(83.) By contrast, state level mens rea reform initiatives may have more success than federal initiatives. In December 2015 the Governor of Michigan signed into law Michigan House Bill 4713. Jack Spencer, Gov. Snyder Signs Bill That Makes 'Guilty Mind' Necessary For Convictions, MICH. CAPITOL CONFIDENTIAL (Dec. 24, 2015), http://www.michigancapitolconfidential.com/22031 [https://perma.cc/477DR645]. The bill provides, with some exceptions, that:
if statutory language defining an element of a criminal offense that is related to knowledge or intent or as to which mens rea could reasonably be applied neither specifies culpability nor plainly imposes strict liability, the element of the offense is established only if a person acts with intent, knowledge, or recklessness.
Mich. Comp. Laws [section] 9(3) (2015).
(84.) All of those criminal prosecutions involved statutes that predate the Internet-era: The relevant portions of the Migratory Bird Treaty Act (40 Stat. 755) were enacted in 1918; the Marine Mammal Protection Act (Pub. L. No. 92-522, 86 Stat. 1027) was enacted in 1972; and the Wilderness Act (Pub. L. No. 88-577, 78 Stat. 890) was enacted in 1964.
(85.) That said, under some circumstances the act of making a sale, or transporting an animal, or choosing a particular mode of transport could certainly be a form of expression.
(86.) Arguably the most important such federal criminal statute in this regard is 18 U.S.C. [section] 875(c), which was at issue in Elonis and provides that "[w]hoever transmits in interstate or foreign commerce any communication containing any threat to kidnap any person or any threat to injure the person of another, shall be fined under this title or imprisoned not more than five years, or both." 18 U.S.C. [section] 875(c) (2012).
(87.) In 1939, a federal statute criminalizing threats was enacted (Pub. L. No. 76-76, 53 Stat. 742 (1939), originally codified at 18 U.S.C. [section] 338), but that statute was limited to threats conveyed using the U.S. Post Office. By contrast, 18 U.S.C. [section] 875(c), which was enacted in 1948 when Title 18 was created, was much broader, and in its original form criminalized threats transmitted "in interstate commerce." Pub. L. No. 80-772, 62 Stat. 741 (1948). In 1986, "in interstate commerce" was further broadened to "in interstate or foreign commerce." Pub. L. No. 99-646, 100 Stat. 3614 (1986).
(88.) The use of "or" here is not meant to suggest that objective and subjective standards are of necessity mutually exclusive. It is possible to have a framework in which conviction requires showing both a subjective intent to threaten as well as showing that the content of the threat was such at that a reasonable person would fear imminent physical harm. This is what the Ninth Circuit did in United States v. Bagdasarian, 652 F.3d 1113 (9th Cir. 2011).
(89.) See Paul T. Crane, "True Threats" and the Issue of Intent, 92 Va. L. Rev. 1225 (2006), for a detailed explanation of the differences between subjective and objective standards, their respective subcategories, and how (up through the mid2000s) those standards have been applied in the courts. Another detailed discussion of these standards, dating from before the 2003 Virginia v. Black decision, is found in Jennifer E. Rothman, Freedom of Speech and True Threats, 25 Harv. J.L. & PUB. POL'Y 283 (2001).
(90.) There are two subcategories of subjective standard. The most common form of subjective standard asks a jury to assess whether a defendant intended that a statement (or other form of expression) be perceived as a threat. See Crane, supra note 89, at 1235-36. The second, less common form of the subjective standard asks whether the speaker intended to actually carry out the threat. Id.
(91.) Objective tests can be further partitioned into subcategories. The "reasonable speaker" test evaluates whether "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 harm." United States v. Kosma, 951 F.2d 549, 557 (3d Cir. 1991). A "reasonable listener" test turns on, in the words of the Eighth Circuit, "whether the recipient of the alleged threat could reasonably conclude that it expresses a determination or intent to injure presently or in the future." United States v. Dinwiddie, 76 F.3d 913, 925 (8th Cir. 1996) (internal quotation marks omitted). There is also a third subcategory of objective standard that examines how a reasonable person who is neither the speaker nor the person named in the threat would perceive it.
(92.) While the constitutional question of intent is intertwined with the question of defining "true threats," it is not limited to that question. For example, as discussed in Part IV, infra, analogous inquiries can also be made with respect to speech that is alleged to constitute an incitement to imminent lawless action.
(93.) 394 U.S. 705. In Watts the Court wrote that conviction under 18 U.S.C. [section] 871(a) "requires the Government to prove a true 'threat.'" Id. at 708.
(94.) 538 U.S. 343 (2003).
(95.) Id. at 359-60 (internal quotation marks and citations omitted).
(96.) Crane, supra note 89, at 1226.
(97.) Black, 538 U.S. at 359-60.
(99.) In reviewing a conviction under 18 U.S.C. [section] 879(a)(3), the Ninth Circuit wrote "[b]ecause the true threat requirement is imposed by the Constitution, the subjective test set forth in Black must be read into all threat statutes that criminalize pure speech." United States v. Bagdasarian, 652 F.3d 1113,1117 (9th Cir. 2011). The Ninth Circuit also stated that in addition to the subjective standard, some statutes require an objective standard to be applied as well: "[W]ith respect to some threat statutes, we require that the purported threat meet an objective standard in addition [to a subjective standard], and for some we do not." Id.
(100.) In a case involving 18 U.S.C. [section] 844(h)(1), the Tenth Circuit wrote that to fall outside the First Amendment, a "threat must be made 'with the intent of placing the victim in fear of bodily harm or death.'" United States v. Magleby, 420 F.3d 1136, 1139 (10th Cir. 2005) (quoting Black, 538 U.S. at 360). One of the statutes at issue in Magleby, 18 U.S.C. [section] 844(h)(1), addressed the use of "fire or an explosive to commit any felony which may be prosecuted in a court of the United States." On appeal Magleby argued that 18 U.S.C. [section] 844(h)(1) was unconstitutional. However, the Tenth Circuit affirmed the conviction, writing that "in view of the doubtful success of this First Amendment challenge to [section] 844(h)(1), we hold that failure to raise the challenge on direct appeal did not constitute ineffective assistance of counsel." Id. at 1144-45 (internal citations omitted).
(101.) State supreme courts using a subjective standard include Massachusetts, Rhode Island, and Vermont. Petition for Writ of Certiorari at 20, Elonis v. United States, 135 S. Ct. 2001 (2015) (No. 13-983) [hereinafter Elonis Petition for Writ of Certiorari].
(102.) Elonis arose from a decision in the Third Circuit, which had upheld a conviction of Elonis under 18 U.S.C. [section] 875(c) applied using an objective standard.
(103.) In a case involving the same statute at issue in Elonis (18 U.S.C. [section] 875(c)), the Fourth Circuit wrote that "In determining whether a statement is a 'true threat/ we have employed an objective test so that we will find a statement to constitute a 'true threat' 'if an ordinary reasonable recipient who is familiar with the context... would interpret [the statement] as a threat of injury.'" United States v. White, 670 F.3d 498, 508-09 (4th Cir. 2012) (quoting United States v. Armel, 585 F.3d 182,185 (4th Cir. 2009)). The Fourth Circuit considered and rejected the assertion by the defendant, White, that conviction under 18 U.S.C. [section] 875(c) requires showing intent.
(104.) In a 2012 decision, the Sixth Circuit wrote that "[t]o convict under [section] 875(c), a jury need conclude only that 'a reasonable person (1) would take the statement as a serious expression of an intention to inflict bodily harm (the mens rea), and (2) would perceive such expression as being communicated to effect some change or achieve some goal through intimidation (the actus reus).'" United States v. Jeffries, 692 F.3d 473, 479 (6th Cir. 2012) (quoting United States v. Alkhabaz, 104 F.3d 1492,1495 (6th Cir. 1997)), cert, denied, 134. S. Ct. 59 (2013).
(105.) In a 2005 decision regarding 18 U.S.C. [section] 875(c), the Seventh Circuit wrote that a true threat is 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 a statement as a serious expression of an intention to inflict bodily harm upon or to take the life of [another individual]." United States v. Stewart, 411 F.3d 825, 828 (7th Cir. 2005) (quoting United States v. Khorrami, 895 F.2d 1186, 1192 (7th Cir. 1990)). Guilt, according to the Seventh Circuit, "is not dependent upon 'what the defendant intended, but whether the recipient could reasonably have regarded the defendant's statement as a threat."' Stewart, 411 F.3d at 828 (quoting United States v. Schneider, 910 F.2d 1569, 1570 (7th Cir. 1990)).
(106.) In a 2011 decision regarding 18 U.S.C. [section] 875(c), the Eighth Circuit wrote that a true threat is "a statement that a reasonable recipient would have interpreted as a serious expression of an intent to harm or cause injury to another." United States v. Mabie, 663 F.3d 322, 330 (8th Cir. 2011) (quoting Doe v. Pulaski Cty. Special Sch. Dist., 306 F.3d 616, 624 (8th Cir. 2002) (en banc)), cert, denied, 133 S. Ct. 107 (2012).
(107.) The Eleventh Circuit wrote in 2013 that "[18 U.S.C.] [section] 875(c) is a general-intent offense that requires the Government to show... [that] the communication would be construed by a reasonable person as a serious expression of an intent to inflict bodily harm or death." United States v. Martinez, 736 F.3d 981, 983 (11th Cir. 2013).
(108.) See Elonis Petition for Writ of Certiorari, supra note 101, at 20.
(109.) As noted previously, this Article is not asserting that Elonis's statements deserved constitutional protection. The statements at issue were odious and likely unlawful. If prosecutors in his trial had endeavored to show intent rather than arguing that it was irrelevant, and if the jury had been properly charged regarding mens rea, Elonis may well have been convicted under the standard that the Supreme Court later determined was proper.
(110.) Elonis, 135 S. Ct. at 2007.
(113.) See Elonis Petition for Writ of Certiorari, supra note 101, at (I).
(114.) 134 S. Ct. 2819 (2014) (granting certiorari).
(115.) Elonis, 135 S. Ct. at 2009 (quoting Morissette v. United States, 342 U.S. 246, 252 (1952)).
(116.) Id. at 2011 (quoting United States v. Dotterweich, 320 U.S. 277, 281 (1943)).
(117.) Id. (quoting Rogers v. United States, 422 U.S. 35, 47 (1975); United States v. Jeffries, 692 F.3d 473, 484 (6th Cir. 2012) (Sutton, J., dubitante)).
(119.) Id. at 2013 (Alito, J., concurring in part and dissenting in part).
(120.) Id. at 2013-14.
(121.) Id. at 2015.
(123.) Id. at 2016.
(124.) Id. at 2018 (Thomas, J., dissenting).
(125.) Justice Alito noted that while the Court indicated that the level of mens rea required should be at least recklessness, it is unclear whether the right level is recklessness itself or something higher like knowledge or purpose:
[T]he Court refuses to explain what type of intent was necessary. Did the jury need to find that Elonis had the purpose of conveying a true threat? Was it enough if he knew that his words conveyed such a threat? Would recklessness suffice? The Court declines to say. Attorneys and judges are left to guess.... This will have regrettable consequences.... If purpose or knowledge is needed and a district court instructs the jury that recklessness suffices, a defendant may be wrongly convicted. On the other hand, if recklessness is enough, and the jury is told that conviction requires proof of more, a guilty defendant may go free. We granted review in this case to resolve a disagreement among the Circuits. But the Court has compounded--not clarified--the confusion.
Id. at 2013-14 (Alito, J., concurring in part and dissenting in part).
(126.) Leslie Kendrick, Guilty Minds, 114 COLUM. L. REV. 1255,1295 (2014).
(127.) Id. at 1270-71.
(128.) Id. at 1271. Kendrick also cites several examples of real-world events analogous to this scenario. See, e.g., Malia Wollan, Fresno State Loves Its Bulldogs, but So Does a Gang, N.Y. TIMES (Nov. 7, 2013), http://www.nytimes.com/ 2013/11/1O/sports/ncaafootball/fresno-adopts-its-college-team-but-so-does-agang.html [https://perma.cc/X6MN-HMJU].
(129.) Brandenburg v. Ohio, 395 U.S. 444, 447 (1969) (per curiam).
(130.) In this scenario, by construction, Person A knows that he or she is sending the tweet to people with a history of committing violence. More generally, however, followers on Twitter raise interesting questions regarding intent because Twitter account holders have the option of looking to see who their followers are, but are not obligated to do so. As a result, there can be substantial uncertainty regarding how much knowledge Twitter users have about the specific composition of their audiences. Of course, uncertain knowledge of one's audience is not uniquely an artifact of digital technology: A person giving a speech may have only partial knowledge regarding the composition of the audience. The differences, however, are that on Twitter the list of followers is easily accessible to any "speaker" (or anyone else) who cares to look and that tweets are more likely to be viewed by followers, but can also be viewed by anyone else.
(131.) As noted previously, a "bot," short for robot, is a computer program that performs a task--typically one that is highly repetitive--in a manner that emulates human behavior. With respect to a service like Twitter, bots can be used to automatically issue tweets.
(132.) Under the Communications Decency Act, "[n]o provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider." 47 U.S.C. [section] 230 (c)(1) (2012).
(133.) Model Penal Code [section] 2.02(2) (Am. Law Inst. 1962).
(134.) It could be argued that the programmer acted with purpose in programming the bot to issue tweets in the general sense. However, with respect to "purposeful" or "knowing," the more pertinent question is whether the programmer purposely or knowingly caused the bot to send the specific tweet at issue. The answer to that question is clearly "no."
(135.) The full definition of "negligently" in the Model Penal Code is:
A person acts negligently with respect to a material element of an offense when he should be aware of a substantial and unjustifiable risk that the material element exists or will result from his conduct. The risk must be of such a nature and degree that the actor's failure to perceive it, considering the nature and purpose of his conduct and the circumstances known to him, involves a gross deviation from the standard of care that a reasonable person would observe in the actor's situation.
Model Penal Code [section] 2.02(2)(d) (Am. Law Inst. 1962).
(136.) See, e.g., Elonis, 135 S. Ct. at 2009 ("The fact that the statute does not specify any required mental state, however, does not mean that none exists.").
(137.) The Elonis Court wrote that "we 'have long been reluctant to infer that a negligence standard was intended in criminal statutes.'" Id. (quoting Rogers v. United States, 422 U.S. 35, 47 (1975)). However, it is also worth noting that with respect to state statutes relating to actions creating a risk of death or great bodily injury, negligence is sometimes considered criminal. For example, in a case involving a statute aimed at protecting children, a California appeals court wrote that the statute "proscribes 'willfully' causing or permitting a child to be placed in a situation likely to produce great bodily harm or death. We hold that the standard of conduct condemned by the statute is that of criminal negligence." People v.
Peabody, 119 Cal. Rptr. 780, 780 (Ct. App. 1975). The court then provided a definition of criminal negligence, stating that conviction under the statute
requires proof of criminal negligence which means that the defendant's conduct must amount to a reckless, gross or culpable departure from the ordinary standard of due care; it must be such a departure from what would be the conduct of an ordinarily prudent person under the same circumstances as to be incompatible with a proper regard for human life.
Id. at 783.
(138.) Conscious disregard is found in the definition of "reckless" in the Model Penal Code and in one of the definitions in Black's Law Dictionary. The full definition of "recklessly" in the Model Penal Code is:
A person acts recklessly with respect to a material element of an offense when he consciously disregards a substantial and unjustifiable risk that the material element exists or will result from his conduct. The risk must be of such a nature and degree that, considering the nature and purpose of the actor's conduct and the circumstances known to him, its disregard involves a gross deviation from the standard of conduct that a law-abiding person would observe in the actor's situation.
Model Penal Code [section] 2.02(2)(c) (Am. Law Inst. 1962) (emphasis added). In Black's Law Dictionary, one of the two alternative definitions of "recklessness" is "[c]onduct whereby the actor does not desire harmful consequence but nonetheless foresees the possibility and consciously takes the risk." BLACK'S Law DICTIONARY 1385 (9th ed. 2009) (emphasis added).
(139.) In Black's Law Dictionary, one of the two definitions of "recklessness" is "[t]he state of mind in which a person does not care about the consequences of his or her actions." BLACK'S Law DICTIONARY 1385 (9th ed. 2009).
(140.) In this example, culpability of at least negligence is inherent in the construction of the scenario, since the tweet advocating imminent lawless action was "due to insufficiently careful programming."
(141.) It might be argued that the programmer could eliminate this risk by simply ensuring that no potentially problematic words (like "vandalize") were in the vocabulary in the first place. However, doing so would mean the algorithm would be unable to use those words in other contexts that raise no First Amendment concerns (for example, in a factual statement like "the store was vandalized yesterday").
(142.) For another discussion on the importance of context in online communications, see Brief for the American Civil Liberties Union, et al. as Amici Curiae Supporting Petitioner at 25-26, Elonis v. United States, 135 S. Ct. 2001 (2015) (No. 13-983), explaining that:
[O]nline communications can easily become decontextualized by third parties. A speaker might send an email to one person, only to see that person forward the message to dozens of others or post it on a public mailing list. Or a speaker may post a comment on his own Facebook profile page, intending it to be seen only by those friends he has allowed to view his page, and later find that one of those friends has taken a screen-capture of his comments and posted the image to an entirely different website. These actions, completely beyond the control of the speaker, place the speaker's statements in front of audiences that the speaker had no expectation or intent to reach. Further, such decontextualization circumvents any effort by a speaker to provide additional context, outside the plain words of the statement, that would make the non-threatening intent of the statement clear.
(143.) While technology-related arguments featured prominently in the amicus briefs, see, e.g., Brief for the National Network to End Domestic Violence, et al. as Amici Curiae Supporting Respondent, Elonis v. United States, 135 S. Ct. 2001 (2015) (No. 13-983); Brief for the American Civil Liberties Union, supra note 142, the Elonis majority opinion itself treated the technology aspects of the case as largely incidental, giving them little attention in the ruling.
(144.) Or at least not explicitly, since Elonis was decided on purely statutory grounds. That said, First Amendment issues figured prominently in the amicus briefs and were reflected in the constitutional question that the Court declined to reach.
(145.) 538 U.S. 343, 347-48 (2003).
(146.) 521 U.S. 844 (1997).
(147.) Id. at 874.
(148.) Id. at 870 (quoting Reno v. ACLU, 929 F. Supp. 824, 842 (E.D. Pa. 1996)).
(149.) Id. at 850.
(150.) Number of worldwide internet hosts in the domain name system (DNS) from 1993 to 2015 (in millions), Statista, http://www.statista.com/statistics/264473/number-of-internet-hosts-in-the-domain- name- system/ [https://perma.cc/6U87-U42P] (last visited May 3, 2016).
(151.) Reno, 521 U.S. at 850.
(152.) See ICT Facts and Figures: The World in 2015, INTERNATIONAL TELECOMMUNICATIONS UNION, https://www.itu.int/en/ITU-D/Statistics/Documents/facts/ ICTFactsFigures2015.pdf [https://perma.cc/EV73-SM5U] (last visited Dec. 14, 2015) (stating that "[globally 3.2 billion people are using the Internet").
(153.) Sarah Phillips, A brief history of Facebook, THE GUARDIAN (Jul. 25, 2007), http://www.theguardian.com/technology/2007/jul/25/media.newmedia [https://perma.cc/RYP2-M46A].
(154.) Myspace reached its peak in the mid-2000s, when Facebook was still primarily used by college students. See Felix Gillette, The Rise and Inglorious Fall of Myspace, BLOOMBERG BUSINESS (June 22, 2011), http://www.bloomberg.com/bw/ magazine/content/11_27/b4235053917570.htm#pl [https://perma.cc/7FLC-BAAR].
(155.) See, e.g., Youssef M. Ibrahim, Cell Phones Make Nokia a World Player, N.Y. TIMES (Aug. 13, 1997), http://www.nytimes.com/1997/08/13/business/cell-phones-make-nokia-a-world-player.html [https://perma.cc/XXU3-S9L6] (stating that the Nokia 9000 Communicator was designed to be "the world's first movable, pocket-sized office. The device would allow conference calls, receive and send faxes, handle E-mail and cruise the Internet--all from a moving car, train or sidewalk cafe."); see also Stephen Manes, Forget Mousetraps! Build Us a Better Web Phone, N.Y. TIMES (Mar. 10, 1998), http://www.nytimes.com/1998/03/10/science/personal-computers-forget-mousetraps-buiId-us-a- better-web-phone.html [https://perma.cc/V4R2-6E7W] (stating that "[prototypes for several phones with built-in Web browsers were shown at the Consumer Electronic Show in January 1997, but none of them actually arrived on the market last year").
(156.) Jennifer E. Rothman, Freedom of Speech and True Threats, 25 Harv. J.L. & PUB. POL'Y 283, 334 (2001).
(157.) "Elonis also claims his threats were constitutionally protected works of art." Elonis v. United States, 135 S. Ct. 2001, 2016 (2015) (Alito, J., concurring in part and dissenting in part).
(159.) Justice Alito argued in his Elonis opinion that the proper standard for 18 U.S.C. [section] 875(c) was recklessness: "Someone who acts recklessly with respect to conveying a threat necessarily grasps that he is not engaged in innocent conduct. He is not merely careless. He is aware that others could regard his statements as a threat, but he delivers them anyway." Id. at 2015.
(160.) Speech restrictions are rife on both public and private university campuses, though it is with respect to public universities, which have constitutional obligations, that the concerns are most acute.
(161.) David Graham, When Campus Hate-Speech Rules Go Further Than the Law, THE ATLANTIC (Nov. 10, 2015), http://www.theatlantic.com/politics/archive/ 2015/11/hate-speech-on-campus/415200/ [https://perma.cc/Y4XF-G4ER].
(162.) Id. The e-mail message from the university police also said that "cases of hateful or hurtful speech are not crimes," though that caveat is weak when accompanied, as it was in this case, with a statement that the university might nonetheless "take disciplinary action." See id.
(163.) First Amendment Protections on Public College and University Campuses: Hearing Before the Subcomm. on the Constitution and Civil Justice of the H. Comm. on the Judiciary, 114th Cong. 74 (2015) (statement of Wendy Kaminer).
(164.) Eric Posner, ISIS Gives Us No Choice but to Consider Limits on Speech, SLATE (Dec. 15, 2015), http://www.slate.com/articles/news_and_politics/view_ from_chicago/2015/12/isis_s_online_radicalization_efforts_present_an_unprecede nted_danger.single.html [https://perma.cc/V7T3-A39Y].
(165.) See, e.g., Mirae Yang, The Collision of Social Media and Social Unrest: Why Shutting Down Social Media is the Wrong Response, 11 Nw. J. TECH. & INTELL. PROP. 707 (2013).
John Villasenor, Professor of Electrical Engineering, Public Policy, and Management, UCLA; Visiting Professor of Law, UCLA; National Fellow, The Hoover Institution; Nonresident Senior Fellow, The Brookings Institution. The Author gratefully acknowledges the feedback and advice of Professor Catherine Crump of the UC Berkeley School of Law, Professor Dov Fox of the University of San Diego School of Law, and Professor Jennifer Rothman of Loyola Law School.
|Printer friendly Cite/link Email Feedback|
|Publication:||Harvard Journal of Law & Public Policy|
|Date:||Jun 22, 2016|
|Previous Article:||Linguistic relativism and the decline of the rule of law.|
|Next Article:||The FCC's transaction reviews and First Amendment risks.|