Brief version
Advocates understand the idea that “code is speech” to create an impenetrable legal shield around anything built of programming code. When they do this they misunderstand, or misrepresent, free speech law (and rights law in general), which rarely creates such impenetrable shields, the principles that underlie that law, and the ways those principles should and might apply to code. The idea that government cannot regulate things because they are made of code cannot be right. That principle not only lacks support in most theories of freedom of speech, but is actively rebutted in the very case law that advocates claim to be marshaling in favor of their position. Further, in promoting this position, advocates misrepresent—in a manner it is hard not to see as willful—the very nature of the programming code they care so much about.
In an excellent piece in MIT Technology Review, law professor Neil Richards goes to the foundations of this question, and wisely calls the view that “code is speech” a “fallacy,” a “fantasy,” a “mistake,” and “wrong.” As a general take on the question I think this is more correct than not, although there are details that need to be gone into at some length. Code can and does have speech-like aspects. But in general, code is much closer to action than it is to speech. The demand that governments not regulate code becomes a demand that governments not regulate action—that is, that governments not regulate at all. That would be less troubling if there were not widespread, repeated, and effective demands for that proposition by the world’s most powerful moneyed interests. Further, the major target of government regulation—corporations—are also the major users of code in the world today. Those corporations, including Apple, whose filings in the #AppleVsFBI case are the most recent instance prompting these discussions, have a deep vested interest in blunting the ability of governments to regulate.
While its relation to speech may certainly be important part of many judicial and legislative actions regarding code, there is no general principle equating code and speech that can or should be relied on to structure those decisions. The cyberlibertarian understanding of “code is speech” contributes to a profoundly conservative assault on the rights of citizens, by depriving the state of the power to regulate and legislate against the corporations that exist only at the state’s pleasure in the first place. This is why “code is speech” has been so powerfully advocated for decades among crypto-anarchists and cypherpunks. Yet at least these groups are, for the most part, explicit about their desire to shrink governmental power and expand the power of capital. Today the view that “code is speech” is far more widespread, but it is no less noxious, than the explicit crypto-anarchist doctrine. Yet civil rights were not created for corporations. They are for individual citizens. They are supposed to protect us against abuses of power, not license them. The fact that Apple is today trying to sell products that openly display their rebuke to government oversight should frighten anyone to the left of Murray Rothbard. That today’s “privacy activists” and “free speech advocates” promote Apple’s actions as a realization of civil rights is an incredibly clear sign of the power of cyberlibertarianism to turn well-understood principles and rights against themselves, claiming to stand for the “little guy” while in fact doing the opposite.
Full version
“Code Is Speech” is one of many articles of faith that make up the cypherpunk and/or crypto-anarchist creed. The acceptance and repetition of those articles of faith among so many, even those who claim not to be on board with the overall cypherpunk ethos, is one way of cashing out the notion of cyberlibertarianism. As with the other cypherpunk articles of faith, a profoundly ambiguous slogan is taken to be a black-and-white statement of principle with a simple and clear meaning. The problem is that it is no such thing.
That simple meaning, as I infer it, would go something like this:
- The First Amendment to the US Constitution absolutely prohibits the US Government from regulating anything that can be said to be made of code: “Congress shall make no law.” Code is inherently the same kind of thing as political speech, and so the government may not legislate against or even regulate it. In particular, governments cannot legislate against the running of code.
This is how many advocates appear to understand the principle (e.g., the Electronic Frontier Foundation’s Executive Director in Time, EFF on its own blog, and a prominent “digital rights” activist/computer scientist). When I refer to “code is speech” in what follows, this is what I mean. By writing “code is not speech” I don’t mean to imply, as I make clear below, that code never has any speech-like features or should never be seen in that context by courts; I mean only that the above proposition is far more wrong than it is right. It is also deeply misleading and ultimately very destructive to democracy, and that should be no surprise: those who have been pushing this perspective the longest, and who arguably developed it, make no secret of their contempt for democratic governance.
The most recent invocation of “code is speech,” and a particularly telling one, is found in Apple’s February 25 motion to vacate the court order in the #AppleVsFBI case:
Under well-settled law, computer code is treated as speech within the meaning of the First Amendment. See, e.g., Universal City Studios, Inc. v. Corley, 273 F.3d 429, 449 (2d Cir. 2001); Junger v. Daley, 209 F.3d 481, 485 (6th Cir. 2000); 321 Studios v. Metro Goldwyn Mayer Studios, Inc., 307 F. Supp. 2d 1085, 1099–1100 (N.D. Cal. 2004); United States v. Elcom Ltd., 203 F. Supp. 2d 1111, 1126 (N.D. Cal. 2002); Bernstein v. Dep’t of State, 922 F. Supp. 1426, 1436 (N.D. Cal. 1996).
The Supreme Court has made clear that where, as here, the government seeks to compel speech, such action triggers First Amendment protections. (section B1; page 32)
The problems with this statement are surprisingly numerous given how brief it is. The case law is by no means “settled” for any number of reasons; to the degree that these cases even address specific issues under the “code is speech” penumbra, they are not the ones that Apple advocates; finally, courts have absolutely not made clear what Apple alleges they have, even in the narrow sense Apple claims—to the contrary, they have tended to hold the opposite position. In its filing Apple takes the general principle that “code is speech,” applies it in a new and legally unprecedented fashion, and then Apple and its supporters mock those who find any problems with the argument. This is typical of the imprecise and black-and-white way that digital enthusiasts think about “code is speech” (and other rights questions, for that matter). But it is both misleading and destructive.
I’ll explain why in four parts: 1) the law is not settled in any sense, but represents a patchwork of cases heard in various courts taking up different aspects of a complex issue, several of them remarkably inconclusive, and none of them taking the issue head-on mentioned above as the main understanding of “code is speech”; 2) government can and does, in fact, regulate speech; 3) in general, it is clear that across the board, code is not speech in the same sense that ordinary speech (or even expressive works of art and other media) is speech, because the primary purpose of code is to take action, and action, speaking generally, is exactly what government can and should regulate; 4) the specific argument Apple makes is entirely novel, demonstrating the deep ambiguity of the phrase “code is speech,” and rests on much shakier legal ground than are some of the other claims under that heading. This overreaching reinterpretation is typical of the ways digital evangelists try to turn legal principles against themselves to their own advantage. Finally, I’ll briefly explain why “code is speech” diminishes civil rights rather than strengthening them.
i. The Law Is Not Settled
First, and simplest, despite what Apple and EFF and other advocates say, the law about the relationship of code and speech is not settled. It is not true that the cases that have been decided so far add up to a clear articulation of principle, because they approach small parts of the problem from disparate angles; and the most central elements of that principle have never been addressed head-on. The issue raises fundamental questions about the nature of the First Amendment in its application to a new form of technology. Such questions can only approach being “settled” if and when the Supreme Court issues a ruling on a case, or at least affirms a lower court ruling by failing to issue cert on a petition for appeal. That has never happened in any of these cases. Only two cases have reached the Supreme Court that address aspects of this question at all: Brown and Sorrell. Neither of these cases takes the question head-on, and for good reason. Apple doesn’t even include them in its litany of cases. (see Appendix 1 for summaries of what each of the cases say).
The lower court cases, some of which get closer to the general question of code and speech, cannot possibly be said to have settled the law. For example, Bernstein v DOJ—the case that even EFF claims is the one that “established” the view that “code is speech”—which is arguably the case that took the general question of code and its relation to speech most directly, was vacated during the appeals process when the government decided not to enforce the relevant regulations due to changing facts: it is no longer considered valid precedent at all. Further, Apple refers to a lower court ruling in Bernstein v Dept of State in 1996 (as does EFF), but the later appellate rulings actually don’t fully support the reasoning the lower court offers, and raise serious questions about the core claims advocates extract from the cases (see Appendix 2). Finally, the cases all address different aspects of the relationship of code to speech, and several of them, to the extent they try to extract a general principle from the issue, actually argue against what “code is speech” advocates want the cases to say. A more accurate assessment of the legal landscape in the US would be: no court has ever issued a general opinion that could lead to the principle that “code is speech.” In Appendix 1 I step through each of the cases Apple cites, as well as the other relevant cases, and show in detail that they do not touch the core question, and that they do not settle many of the open legal questions, let alone the core question of the relationship of code to speech.
The relationship of code and speech needs to be examined thoroughly, in detail, with experts on all aspects of that relationship weighing in—not exclusively tech industry lawyers, engineers, corporate lobbyists and advocates with a vested and one-sided interest in magnifying their own power and the power of those they represent. The blanket belief that “code is speech” leads to rank overextensions of rights talk like the view that “data is speech” (Bambauer 2014), or that Google’s search engine results deserve First Amendment protection (Volokh and Falk 2012, ably dissected in Grimmelman 2014 and Pasquale 2012a, 2012b); it also fuels the truly disturbing First Amendment shield Google has tried to create around the EU “right to be forgotten” ruling. It is possible and desirable to think carefully about what the principles of freedom of the press and freedom of speech are supposed to mean and how those mesh with new technologies; Richards (2016) and Tutt (2012) give good overviews of some of the issues there, and the great First Amendment scholar Jack Balkin has been thinking about these questions for several decades (see e.g. Balkin 2004). But there is no “settled law” whatsoever about what that relationship should be.
ii. Government Can and Does Regulate Speech
Even if code were speech across the board—which it is not—it simply is not the case that the First Amendment means, in US law, or for that matter, freedom of speech laws elsewhere, that “if it’s speech, the government can’t pass laws about it.” Freedom of speech is usually taken as a prohibition on censorship or “prior restraint” on bona fide speech, of which the core exemplar is explicit political speech. Richards is particularly good on this topic. Yet what is so often overlooked in these discussions is that the First Amendment does not create a blanket prohibition on censorship; rather, actually settled case law has taken it for over half a century (including the period when most of the significant First Amendment case law has been developed) to mean that, in various contexts, different tests must be applied to determine whether or not a given law does or does not violate the First Amendment. These are typically framed in terms of levels of scrutiny, where scrutiny refers to the burden required for the government to overcome First Amendment protections. Scrutiny tests are used not just for the First Amendment, but in many circumstances where fundamental rights are implicated.
For example: it’s long been settled that “content-neutral” prohibitions on speech—that is, laws that prohibit any speech, regardless of content, in certain venues or at certain times—encounter a lower level of scrutiny than would censorship of political editorials. In general, for the government to censor a political editorial, it must meet the test of “strict scrutiny,” meaning that if the government can demonstrate a “compelling government interest” in censoring the content (and meet a few other criteria), it may be found constitutional. In general, for the government to issue content-neutral speech restrictions—for example, when a local government issues noise limits for certain periods of time, or declares certain parts of the city off-limits for protest—the standard is much lower, “intermediate scrutiny.” These “scrutiny” tests are all over First Amendment law and other rights law; they have become the standard way of mediating between rights interests and government power since at least the 1960s, in part due to considerations arising from the Equal Protection Clause of the 14th Amendment (for a more detailed account see Siegel 2006). Simply saying that something is covered by the First Amendment does not mean that government is unable to regulate it. For kinds of speech that are less central, lower tests apply—“intermediate scrutiny,” and in the least central instances, the “rational basis test.” I’ve added more information about scrutiny tests, including the fact that they are a routine part of legal education taught in introductory Constitutional Law classes (Appendix 5).
One doesn’t have to look far to see the court system making this abundantly clear, even in the very cases Apple and others cite as if it supports their position. This topic is addressed with particular force in the lower court ruling in Universal Studies v. Corley, in a part of the decision that was not overturned by subsequent rulings of the Appeals Court:
All modes of expression are covered by the First Amendment in the sense that the constitutionality of their “regulation must be determined by reference to First Amendment doctrine and analysis.” Regulation of different categories of expression, however, is subject to varying levels of judicial scrutiny. Thus, to say that a particular form of expression is “protected” by the First Amendment means that the constitutionality of any regulation of it must be measured by reference to the First Amendment. In some circumstances, however, the phrase connotes also that the standard for measurement is the most exacting level available. (Universal Studios v Reimerdes at IIIa [14])
In both the Universal Studios v Reimerdes and Universal Studies v. Corley decisions the judges go to great lengths to indicate the levels of scrutiny that various kinds of speech regulation require. They repeatedly reject the opinion that simply labeling something “speech” means government must keep its hands off. Further, it’s clear, as Neil Richards says, that the general trend of these tests is to put political speech by individuals at the core, demanding the highest level of scrutiny to justify legislation, and other forms of speech are seen in relation to it.
Apple’s filing in #AppleVsFBI overlooks this fact entirely—Apple writes as if the fact that “code is speech” simply blocks all action by the government. It doesn’t. Even if a Court were to apply the strict scrutiny test to a case like this one, it’s entirely plausible that using a legal warrant to “speak” by writing some computer code in the service of the investigation of a crime where many people were killed and many more were injured would be properly viewed as “compelling government interest” and a “narrowly tailored measure.” Although it is fashionable to dismiss everything the government says in this case and others like it, it’s notable that the Department of Justice makes exactly this point in its own filing (Appendix 4).
iii. In General, Code Is More Properly Viewed as Action than as Speech
This is truly the heart of the matter. Advocates love to gloss over the fact that the cases mentioned here look at one or another aspect of code, but actively reject the wholesale equation of code and speech. The reasons for this are obvious. Code clearly does have some speech-like qualities. In certain limited contexts, code can be used to express ideas between people, whether through the archetypal example of “Perl poetry” or the more prosaic argument that publication of Daniel Bernstein’s Snuffle code was meant to convey the idea of his encryption algorithm to other programmers. This is what advocates love to harp on, and some at times talk as if the fact that code can have expressive features itself triggers full First Amendment protections for any and all code. This is, simple, false.
The digital revolution is characterized by fallacious arguments that take the form “completely different and exactly the same”: a phenomenon is exactly the same as some existing phenomenon X, so critical objections to it are invalid; the phenomenon is completely different from X, which is why consumers would want it and developers will build and sell it. This is particularly true with regard to code. Code is one of the most explosively new phenomena society has ever encountered, particularly in the past 50 years. It has literally transformed enormous parts of the world. It is deeply connected to existing phenomena (especially formal logic and mathematics), but it is different from all of them. Code is why computers exist. Code is one of the main reasons that computers are remarkable, one of the main reasons we have a “digital revolution.” That is why it is incredibly disingenuous and self-serving that the most vigorous proponents of the transformative power of the digital—and in this sense of code—are the very same people telling us that code is so much like “language” (no doubt, because it is largely made of language and language-like symbols) that the law must treat it as if it were language
This is most easily seen when we think about the main use to which code is put. The reason we have and pay so much attention to code is because it is executed. Execution is not primarily a form of communication, not even between humans and machines. Execution is the running of instructions: it is the direct carrying out of actions. It is adding together two numbers, or multiplying ten of them, or looping back to perform an operation again, or incrementing a value. This is what code does. Lots of code does not even look like language at all—it looks, if anything, like giant arrays of numbers that mean very little to anyone but the most highly-trained programmer. All of it executes, or at least can be executed. That is what it is for. That is what it does. That’s what makes it different from lots of other things, like most language and expressive media, all of whose primary function is to convey thoughts and ideas and feelings between persons.
Code can and often does serve these expressive functions. But this fact has been used, often very cynically, by digital advocates to push the legal system into misapprehending expression as the primary purpose of code, which it is not. Then, decision based in part or in whole on these expressive functions are taken as “proof” of some sort that code cannot be regulated. This is cynical because the advocates know that what they mean by code is not the freedom to express ideas with code, but to run it. Thus the courts hear “code is speech” as a doctrine about the fact that code can be used to express ideas between people, and that it therefore can’t be restricted on First Amendment grounds. Engineers and the EFF and others hear these decisions as resulting in a dictum that the government cannot restrict the running of programs because running programs is like speaking. This is just, as near as one can be about legal matters, false on its face. And it results in abhorrent doctrine: it says that as long as a corporate actor takes action using programming code, the government cannot restrict it because of the First Amendment.
It’s important to note that two of the cases most often cited in this context, Bernstein and Universal Studios, both make this point. Bernstein explicitly restricts itself to the publication of code because code in execution is so different from ordinary forms of expression, and the dissent in Bernstein does not even accept this (Appendix 2); the court in Universal Studios considers at length “the functional aspects of code”—that is, its use in execution—and considers them to raise different questions from its expressive characteristics (Appendix 3).
iv. Apple’s Argument Expands the Idea of “Code Is Speech” in an Unprecedented and Antidemocratic Manner
Apple claims that “code is speech” is settled law; it isn’t. Then it claims that “code is speech” prohibits the government from compelling Apple to write code because that would be compelling it to speak. But that assertion rests on a something that is settled law: that the government can compel corporations to speak, much differently from how it treats individuals (see “Compelled Speech”). In its filings and press releases, Apple refers to itself as if it were a natural person. Natural persons have many more First Amendment (and other Bill of Rights) protections than do corporations, although the boundaries between the two are more porous and murky than they should be (see Greenwood’s essays, and Pollman 2011). Many lawyers and legal scholars (and even dissenting Supreme Court justices) feel that the twin leaps of logic that come to fruition in Citizens United and Hobby Lobby—that corporations are people and that money spent (by corporations) is speech and that corporations can have religious beliefs—depend on truly untenable extensions of constitutional rights to corporations. But even there, the Court has not ruled that corporations have exactly the same rights under the First Amendment that natural persons do.
Further, Apple nowhere mentions that the question of corporate compelled speech is one that the Supreme Court has adjudicated, largely against the position it takes. Corporations are artificial entities that only exist at the pleasure and license of the government. In the very act of incorporation, corporations agree to participate in certain aspects of law and regulation that natural persons do not have to.
Corporations can be compelled to speak. For example, corporations are not allowed to promulgate false advertising. This is different from the fact that individuals can sue after the fact for libel and slander. The FTC is allowed to censor advertisements it deems to be false, along with other tools at its disposal. In addition, the Court has long ruled that the FDA and other regulatory bodies can demand that products be labeled for potential harm to human life (e.g. poison labels on pesticides, warning labels on pharmaceuticals and tobacco products), for informational purposes related only generally to health (nutrition information on food), and even for purposes of general engineering safety and operability (such as the engineering labels that Apple itself includes inside its products and in the literature surrounding them—in other words, Apple is already being “compelled” to speak millions of times each day, much more publicly than it would have been in #AppleVsFBI, and does not complain about it).
Apple’s public portrayal of itself as a natural person whose rights to speak are being infringed by the government—and the way advocates unquestioningly repeat this representation—shows the danger of accepting “code is speech” as a general principle. Its effect is not at all what it appears to be. That makes sense: the speech of individuals is already protected at a very profound level by First Amendment jurisprudence, and if and when it appears that an individual is “speaking” through code—as at least the majority opinion in Bernstein found—courts have adequate resources to confront that issue. Just because “code is speech” is so vague on the one hand, and so badly grounded in actual free speech doctrine on the other, it leads to objectionable restatements of legal principle that are easily exploited by those who have the strongest vested interest in controlling how governments regulate power.
v. “Code Is Speech” Diminishes Civil Rights
Richards, Goodman, and others (Boston Globe, Fein and Gifford) have done good work in nailing this down, so I’m not going to belabor it, except to make a relatively obvious point that is repeatedly echoed in the case law. Most code is action. A huge amount of code is promulgated by corporations, not individuals. The effect of embracing “code is speech” is to say that governments cannot regulate what corporations do. That might seem like hyperbole, but it is 100% on board with the Silicon Valley view of the world, the overt anarcho-capitalism that many of its leaders embrace, and the covert cyberlibertarianism that so many more accept without fully understanding its consequences. It is profoundly anti-democratic. This is part of what makes it so confounding that so many see Apple as some kind of civil rights actor, especially when its avowed mission is to sell products that block the serving of legal, targeted warrants, and when it already makes such outrageous statements regarding the corporate taxes it clearly would owe if not for the existing successful capture of regulatory and legislative bodies that enable the kinds of corporate inversions and other tax dodges we see on display in the Panama leaks, among many other places.
The most tireless legal scholar on this point is Hofstra professor Daniel J. H. Greenwood, whose works, cited in the bibliography below, I cannot recommend enough. Greenwood (1998, 2005, 2013; Greenfield, Greenwood and Jaffe 2007) has long argued that “nothing in the structure or language of the bill of rights suggests that the traditional rights of American citizens apply to corporations” (Greenwood 2013, 14). Coates (2015), Miller (2011), Pollman (2011), and quite a few others have worked on it as well. To most scholars who don’t have a vested financial interest in the success of one company or another (e.g. those who don’t work directly for corporations or for corporate-funded think tanks), the encroachment of corporations into rights language has been one of the signal failures of US democracy. This is not to say corporations never should have any rights, or even that the notion of “artificial person” is entirely bankrupt (it seems to do important work, for example, in the rights of both corporations and natural persons to act as equal parties in contracts and to use the civil courts to adjudicate breaches of contract), but that the general expansion of corporate personhood and identification of corporations as the locus of constitutional rights is among the most significant dangers facing democratic governance today.
Consider this truly jarring statement regarding #AppleVsFBI from EFF’s Executive Director, Cindy Cohn:
The Supreme Court has rejected requirements that people put “Live Free or Die” on their license plates or sign loyalty oaths, and it has said that the government cannot compel a private parade to include views that organizers disagree with. That the signature and code in the Apple case are implemented via technology and computer languages rather than English makes no difference. For nearly 20 years in cases pioneered by EFF, the courts have recognized that writing computer code is protected by the First Amendment.
EFF is mostly staffed by lawyers. Cohn is an attorney who has a long pre-EFF history of working for civil and human rights—and actually worked on Bernstein, which EFF persists in mischaracterizing in several critical ways. Yet rather than making clear the fact that she is making a truly novel argument about a corporation being compelled to speak—or, to be much more honest, to take action—she purposely blurs the lines between between code as action and as speech, and between individuals and corporations. She writes that “the FBI should not force Apple to violate its beliefs,” but the only case that the Supreme Court has ever decided that even suggests that corporations have beliefs is the horribly right-wing 2014 Hobby Lobby decision, which nobody outside of far-right ideologues should endorse, and which itself depends on the fact that Hobby Lobby is a family-owned private corporation, not a public company like Apple. It is fine to endorse this view, I suppose, but to frame it in terms of loyalty oaths is really dirty pool. This is right-wing politicking of the highest order, demanding that corporations be extended the full panoply of rights which the framers and almost all non-technology and non-right-wing thinkers have always thought apply only to natural persons. That it can somehow be mounted in terms of “human rights” and “freedom” is really shocking. As a principle, “code is speech” does not represent a natural extension of rights, but rather a significant curtailing of rights, by putting ordinary actions outside the penumbra of legal regulation. Hopefully, should the matter ever be fully adjudicated by the Supreme Court, sanity will prevail (which is obviously asking a lot), and this will be made as clear as it should be.

Image Source: ShutterStock/Mclek via canmua.net
Bibliography
- Balkin, Jack. 2004. “Digital Speech and Democratic Culture: A Theory of Freedom of Expression for the Information Society.” NYU Law Review 79 (Apr 2004), 1-55, http://www.yale.edu/lawweb/jbalkin/telecom/digitalspeechanddemocraticculture.pdf.
- Bambauer, Jane R. 2014. “Is Data Speech?,” Stanford Law Review 66:1 (Jan 2014), 57-120, http://www.stanfordlawreview.org/sites/default/files/66_Stan._L_Rev_57_Bambauer.pdf.
- The Boston Globe. 2016. “Apple vs. FBI Is Not a Free Speech Matter.” Editorial (Mar 27). http://www.bostonglobe.com/opinion/editorials/2016/03/27/apple-fbi-not-free-speech-matter/bUTouf0mfLy435IB94ikEI/story.html.
- Chopra, Samir. 2016. “Apple’s ‘Code Is Speech’ Argument, The DeCSS Case, And Free Software.” SamirChopra.Com (Mar 3). https://samirchopra.com/2016/03/03/apples-code-is-speech-argument-the-decss-case-and-free-software/.
- Coates, John C., IV. 2015. “Corporate Speech and the First Amendment: History, Data, and Implications.” Constitutional Commentary 30:2 (Summer 2015), 223-276. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2566785.
- Cohn, Cindy. 2016. “Code Is Free Speech.” Time (Mar 17). http://time.com/4248928/code-is-free-speech/.
- “Compelled Speech.” IT Law Wiki. http://itlaw.wikia.com/wiki/Compelled_speech.
- Crocker, Andrew and Jamie Williams. 2016. “Deep Dive: Why Forcing Apple to Write and Sign Code Violates the First Amendment.” EFF Deeplinks (Mar 3). https://www.eff.org/deeplinks/2016/03/deep-dive-why-forcing-apple-write-and-sign-code-violates-first-amendment.
- Dame-Boyle, Alison. 2015. “EFF at 25: Remembering the Case that Established Code as Speech.” EFF Deeplinks (Apr 16). https://www.eff.org/deeplinks/2015/04/remembering-case-established-code-speech.
- Fein, Ron, and Danny Gifford. 2016. “Apple, the FBI, and the Real Meaning of the First Amendment.” The Daily Caller (Mar 4). http://dailycaller.com/2016/03/04/apple-the-fbi-and-the-real-meaning-of-the-first-amendment/.
- Garrett, Brandon L. 2014. “The Constitutional Standing of Corporations.” University of Pennsylvania Law Review 163, 95-164. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2330972.
- Goodman, Ellen. 2016. “Big Pharma, Tobacco, Tech—How the First Amendment Is Being Abused.” The Guardian (Mar 18). http://www.theguardian.com/technology/2016/mar/18/apple-first-amendment-defense-big-pharma-tobacco-tech-companies?CMP=share_btn_tw.
- Greenfield, Kent, Daniel J.H. Greenwood, and Erik S. Jaffe. 2007. “Should Corporations Have First Amendment Rights?” Seattle University Law Review 30. 875-894. http://digitalcommons.law.seattleu.edu/cgi/viewcontent.cgi?article=1909&context=sulr.
- Greenwood, Daniel J.H. 2013. “Do Corporations Have a Constitutional Right to Bear Arms? and Related Puzzles in Post-National Jurisprudence.” Draft posted to https://people.hofstra.edu/Daniel_J_Greenwood/pdf/2dADRAFT.pdf.
- Greenwood, Daniel J.H. 2005. “Markets and Democracy: The Illegitimacy of Corporate Law.” University of Missouri-Kansas City Law Review 74:1. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=747385.
- Greenwood, Daniel J.H. 1998. “Essential Speech: Why Corporate Speech Is Not Free.” Iowa Law Review 83. 995-1070. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=794785.
- Grimmelman, James. 2014. “Speech Engines.” Minnesota Law Review 98:3. 868-952. http://www.minnesotalawreview.org/wp-content/uploads/2014/02/Grimmelmann_MLR.pdf.
- Kanovitz, Jacqueline R. 2010. Constitutional Law. Twelfth Edition. New Providence, NJ: Matthew Bender & Co/LexisNexis.
- Miller, Darrell A. H. 2011. “Guns, Inc.: Citizens United, McDonald, and the Future of Corporate Constitutional Rights.” New York University Law Review 86 (Oct 2011). 887-. http://www.nyulawreview.org/sites/default/files/pdf/NYULawReview-86-4-Miller.pdf.
- Park, Alex. 2014. “10 Supreme Court Rulings—Before Hobby Lobby—That Turned Corporations Into People.” Mother Jones (Jul 10). www.motherjones.com/politics/2014/07/how-supreme-court-turned-corporations-people-200-year-saga.
- Pasquale, Frank. 2012a. “Automated Arrangement of Information: Speech, Conduct, and Power.” Concurring Opinions (Jun 25). http://concurringopinions.com/archives/2012/06/automated-arrangement-of-information-speech-conduct-and-power.html.
- Pasquale, Frank. 2012b. “Search as Speech: Two Scenarios.” Concurring Opinions (May 29). http://concurringopinions.com/archives/2012/05/search-as-speech-two-scenarios.html.
- Pollman, Elizabeth. 2011. “Reconceiving Corporate Personhood.” Utah Law Review 4. 1629-1675. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1732910.
- Richards, Neil. 2016. “Apple’s ‘Code=Speech’ Mistake.” MIT Technology Review (Mar 1). https://www.technologyreview.com/s/600916/apples-code-speech-mistake/.
- Siegel, Stephen A. 2006. “The Origin of the Compelling State Interest Test and Strict Scrutiny.” The American Journal of Legal History 48:4 (Oct). 355-407. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=934795.
- Tribe, Laurence H. 2015. “Dividing Citizens United: The Case v. the Controversy.” Constitutional Commentary 30:2 (Summer). 463-494. http://ssrn.com/abstract=2575865.
- Tutt, Andrew. 2012. “Software Speech.” 65 Stan. L. Rev. Online 73 (Nov 15). http://www.stanfordlawreview.org/online/software-speech.
- US Department of Justice. 2016. “Government’s Reply In Support Of Motion To Compel And Opposition To Apple Inc.’s Motion To Vacate Order.” US District Court for the Central District of California (Mar 10). https://www.scribd.com/doc/303738452/Gov-t-Response-to-Apple.
- Volokh, Eugene, and Daniel M. Falk. 2012. “First Amendment Protection for Search Engine Search Results.” Journal of Law, Economics, and Policy 8:4 (2012). 883-899. http://www2.law.ucla.edu/volokh/searchengine.pdf.
Appendices
APPENDIX 1: The Cases Do Not Add Up to “Settled Law”
US Court Cases Cited By Apple in Its #AppleVsFBI Motions
- Universal City Studios, Inc. v. Corley, 273 F.3d 429, 449 (2d Cir. 2001); this case is sometimes referred to as Universal Studios v Reimerdes; the Corley name got attached only on the 2001 appeal. This case is different from the encryption cases, because it asks whether the First Amendment prohibits the government from regulating the distribution of a software program designed to circumvent legal copyright protections (under the DMCA). Unlike some of the other cases, this one does not turn on the First Amendment question. Although it is used by advocates as if it does, in fact the language in this string of cases, especially the appeal, rebuts both the legal and factual arguments about “code is speech” much more than it supports them. Links: Wikipedia; decisions: Corley (appeal), Reimerdes (lower court).
- Junger v. Daley, 209 F.3d 481, 485 (6th Cir. 2000); this case builds largely on the Bernstein rulings that were eventually overruled and then vacated. Further, like so many of the cases mentioned here, language in the decision rebuts rather than supports the expansive reading of “code is speech.” This case is also about the publication (and execution?) of encryption software. Links: Wikipedia; case archive.
- 321 Studios v. Metro Goldwyn Mayer Studios, Inc., 307 F. Supp. 2d 1085, 1099–1100 (N.D. Cal. 2004); another case (like Universal Studios) about the DMCA, and another case that rebuts rather than supports the general “code is speech” equivalence: as Wikipedia puts it, the court “did not agree that enforcing the DMCA in this case would regulate computer code on the basis of content. The court held that only the functional element of the computer code was barred, and so the DMCA did not suppress the code based on its content. As such, the court applied an intermediate scrutiny standard in evaluating the restriction of speech in this case.” And it’s important to add: 321 Studios, the litigant who made the First Amendment argument, lost the case: “The court held that both of DVD Copy Plus and DVD-X Copy violated the DMCA and that the DMCA was not unconstitutional. The court enjoined 321 Studios from manufacturing, distributing, or otherwise trafficking in any type of DVD circumvention software.” Links: Wikipedia.
- United States v. Elcom Ltd., 203 F. Supp. 2d 1111, 1126 (N.D. Cal. 2002); this case may be the biggest stretch of all those listed. The case was a criminal prosecution of software developer Dmitry Sklyarov under the DMCA. Sklyarov was acquitted by the jury, so there was no appeal possible. There is no judicial decision to refer to, and it is not at all clear that the case has anything to do with the First Amendment. Links: Wikipedia.
- Bernstein v. Dep’t of State, 922 F. Supp. 1426, 1436 (N.D. Cal. 1996): this case is more properly referred to as Bernstein v. U.S. Dept. of Justice, 176 F.3d 1132 (9th Cir.1999) because “the Ninth Circuit ordered that this case be reheard by the en banc court, and withdrew the three-judge panel opinion.” It’s no accident that Apple cites the 1996 decision, as do advocates like EFF, because that decision says something about “code is speech” that is actually superseded in the 1999 appeals decision that overrules the 1966 decision (see Appendix 2). Further, this entire stream of cases was vacated in 2003 and was never fully tested in the courts; it has no binding force: “On October 15, 2003, almost nine years after Bernstein first brought the case, the judge dismissed it and asked Bernstein to come back when the government made a ‘concrete threat.’” Although Bernstein is the case “code is speech” advocates cite most routinely as if it supports their case (it does not), it is the least like settled law among all of them. Links: Wikipedia; EFF archive; Bernstein’s archive.
Relevant US Supreme Court Cases Not Cited By Apple in Its #AppleVsFBI Motions
- Sorrell v. IMS Health Inc., No. 10-779 131 S.Ct. 2653 (2011). Apple does not cite this case, which is odd since it is the only one to have advanced to the Supreme Court, the only one to have turned directly on First Amendment questions, and the only case to possibly have what is usually meant by the phrase “settled law.” Of course this case is not directly about code: it is about whether governments can prevent corporations from selling data they have collected. Like Citizens United, this case was written by the right wing of the Court, and along with that case is frequently referred to by everyone to their left as a truly disturbing extension of First Amendment rights to corporations. Links: Wikipedia; decision.
- Brown v. Entertainment Merchants Ass’n, 564 U.S. 08–1448 (2011). The other Supreme Court case that appears to touch on code as speech. In Brown, the Supreme Court ruled 7-2 that video games deserve the same protections as do any other forms of cultural expression. Andrew Tutt summarizes the holding: “As a threshold matter, the Court had to decide whether video games were speech. Rather than reach beyond video games to software generally, the Court zeroed in on video games and held that they were speech because they communicated ideas through familiar literary devices. The Court reasoned that video games were speech because they expressed ideas in familiar ways: ‘Like the protected books, plays, and movies that preceded them, video games communicate ideas—and even social messages—through many familiar literary devices (such as characters, dialogue, plot, and music) and through features distinctive to the medium (such as the player’s interaction with the virtual world).’” This sidesteps rather than addresses the main code-as-speech question: it is hard to argue that cultural products made with computer code are any different from any other cultural products, so they deserve identical kinds of protection to those other products. The case is clearly about things built with code, rather than code itself; Tutt is particularly good on the fundamental differences between the two. Links: Wikipedia; decision.
- Citizens United v. Federal Election Commission, No. 08-205, 558 U.S. 310 (2010). The case in which the Supreme Court famously ruled that laws constraining campaign-related expenditures, even when those expenditures were made by corporations, violated the First Amendment. Relevant to this case because even the ACLU sided with the majority view here, which is typically summarized as “money is speech.” Citizens United extends, and to some extent depends upon, the expansion of First Amendment rights to entities other than natural persons (see e.g. Park 2014 for a catalog of some of the other relevant cases), but even those reasonable First Amendment advocates who agree with the basic thrust of the decision did much more damage than necessary to the Constitutional fabric due to “opportunistic overreach”: “In Citizens United, the Court was presented with a narrow question about the constitutionality of campaign finance rules as applied to a nonprofit’s on-demand video, but it transformed the case into an opportunity to rule with a broad brush, putting essentially all future regulation of campaign finance in conspicuous jeopardy” (Tribe 2015, 476-7). Links: Wikipedia; decision.
- Burwell v. Hobby Lobby, 573 U.S. ___ (2014). In some ways even more than Citizens United this is the case that should disturb civil liberties advocates with regard to the “code is speech” claim. In Hobby Lobby the Court decided that corporations—although, in this case, a very specific kind of corporation that is closely held by a family—can have the kind of religious beliefs that the First Amendment was intended to protect under the free exercise clause. Almost all commentators, even some who support Citizens United, feel that in Hobby Lobby the court went far beyond the question put to it in the case. As one prominent legal commentator puts it, the “unfounded claims” made by the majority in Hobby Lobby “disregarded the fundamental feature of state corporate law: separation of ownership from the entity”: “Far from being ‘quite beside the point,’ legal separateness is the point of creating a corporation,” (Garrett 2014, 145-6). In her dissent, Justice Ruth Bader Ginsburg wrote: “the exercise of religion is characteristic of natural persons, not artificial legal entities” (Hobby Lobby at 2794, Ginsburg, J., dissenting). It’s notable how much the “code is speech” defense mounted by Apple portrays corporations as natural persons with the kinds of beliefs, attitudes and rights that many outside the far right think only attach to natural persons. Links: Wikipedia; slip opinion.
APPENDIX 2: Bernstein Explicitly Restricts Itself to the Publication and Expressive Features of Code
The part of a lower court ruling EFF and others love to cite, but that is superseded by appeals:
This court can find no meaningful difference between computer language, particularly high-level languages as defined above, and German or French….Like music and mathematical equations, computer language is just that, language, and it communicates information either to a computer or to those who can read it…
-Judge Patel, April 15, 1996
This seems mistaken for many reasons, among them that it creates an entirely new principle that music, math equations, and code are just language, which is obviously false. It also relies on the completely false equivalency between “programming languages” and “[human] languages,” another characteristic digital fallacy which requires separate treatment, but if you want some sense of why this is wrong, ask a linguist.
From the much more limited appeals ruling by Betty Fletcher:
Cryptographers use source code to express their scientific ideas in much the same way that mathematicians use equations or economists use graphs. Of course, both mathematical equations and graphs are used in other fields for many purposes, not all of which are expressive. But mathematicians and economists have adopted these modes of expression in order to facilitate the precise and rigorous expression of complex scientific ideas. Similarly, the undisputed record here makes it clear that cryptographers utilize source code in the same fashion. (Fletcher 1999 at 4233)
Judge Nelson’s dissent is even stronger, and I think it’s correct, and I hope that any future rulings, especially by the Supreme Court, follow its reasoning. Nelson writes that he is
inevitably led to conclude that encryption source code is more like conduct than speech. Encryption source code is a building tool. Academics and computer programmers can convey this source code to each other in order to reveal the encryption machine they have built. But, the ultimate purpose of encryption code is, as its name suggests, to perform the function of encrypting messages. Thus, while encryption source code may occasionally be used in an expressive manner, it is inherently a functional device. (Nelson, dissent, 4245-6)
APPENDIX 3: Both Universal Studios Rulings Say Unambiguously that Code Can Be Regulated
Universal Studios v Reimerdes:
These considerations suggest that the DMCA as applied here is content neutral, a view that draws support also from City of Renton v. Playtime Theatres, Inc. The Supreme Court there upheld against a First Amendment challenge a zoning ordinance that prohibited adult movie theaters within 1,000 feet of a residential, church or park zone or within one mile of a school. Recognizing that the ordinance did “not appear to fit neatly into either the ‘content based or the ‘content-neutral’ category,” it found dispositive the fact that the ordinance was justified without reference to the content of the regulated speech in that the concern of the municipality had been with the secondary effects of the presence of adult theaters, not with the particular content of the speech that takes place in them. As Congress’ concerns in enacting the anti-trafficking provision of the DMCA were to suppress copyright piracy and infringement and to promote the availability of copyrighted works in digital form, and not to regulate the expression of ideas that might be inherent in particular anti-circumvention devices or technology, this provision of the statute properly is viewed as content neutral. (at 21)
And further,
This analysis finds substantial support in the principal case relied upon by defendants, Junger v. Daley. The plaintiff in that case challenged on First Amendment grounds an Export Administration regulation that barred the export of computer encryption software, arguing that the software was expressive and that the regulation therefore was unconstitutional. The Sixth Circuit acknowledged the expressive nature of computer code, holding that it therefore was within the scope of the First Amendment. But it recognized also that computer code is functional as well and said that “[t]he functional capabilities of source code, particularly those of encryption source code, should be considered when analyzing the governmental interest in regulating the exchange of this form of speech.” Indeed, it went on to indicate that the pertinent standard of review was that established in United States v. O’Brien, the seminal speech-versus-conduct decision. Thus, rather than holding the challenged regulation unconstitutional on the theory that the expressive aspect of source code immunized it from regulation, the court remanded the case to the district court to determine whether the O’Brien standard was met in view of the functional aspect of code. (at 24)
APPENDIX 4: DoJ Response to Apple’s #AppleVsFBI Filing
On the difference between corporate & individual speech, & quoting two of the exact cases Apple cites to show that they do not support its argument:
Apple’s claim is particularly weak because it does not involve a person being compelled to speak publicly, but a for-profit corporation being asked to modify commercial software that will be seen only by Apple. There is reason to doubt that functional programming is even entitled to traditional speech protections. See, e.g., Universal City Studios, Inc. v. Corley, 273 F.3d 429, 454 (2d Cir. 2001) (recognizing that source code’s “functional capability is not speech within the meaning of the First Amendment”). “[T]hat [programming] occurs at some level through expression does not elevate all such conduct to the highest levels of First Amendment protection. Doing so would turn centuries of our law and legal tradition on its head, eviscerating the carefully crafted balance between free speech and permissible government regulation.” United States v. Elcom Ltd., 203 F. Supp. 2d 1111, 1128-29 (N.D. Cal. 2002). (US DoJ 2016, p32)
On the question of which scrutiny standard should apply:
Even if, despite the above, the Order placed some burden on Apple’s ability to market itself as hostile to government searches, that would not establish a First Amendment violation because the Order “promotes a substantial government interest that would [otherwise] be achieved less effectively.” Rumsfeld, 547 U.S. at 67. There is no question that searching a terrorist’s phone—for which a neutral magistrate has found probable cause—is a compelling government interest. See Branzburg v. Hayes, 408 U.S. 665, 700 (1972) (recognizing that “the investigation of a crime” and “securing the safety” of citizens are “fundamental” interests for First Amendment purposes). As set forth above, the FBI cannot search Farook’s iPhone without Apple’s assistance, and Apple has offered no less speech-burdensome manner for providing that assistance.
For all of these reasons, Apple’s First Amendment claim must fail. (US DoJ 2016, p34)
APPENDIX 5: Scrutiny Tests in Constitutional Jurisprudence
One of several principles involved in #AppleVsFBI that, unlike “code is speech,” deserves the label of “settled law” is the application of so-called “scrutiny tests” to cases involving possible governmental breaching of fundamental civil rights. While it is most familiarly applied in First Amendment cases, it is used across the board in a wide variety of rights cases, and is frequently associated with the Equal Protection clause of the 14th Amendment (though its connection to that clause is not entirely clear; see Siegel 2006 for a thorough analysis).
To show how basic scrutiny tests are to US rights law, here’s a characteristic passage from a Con Law textbook I was able to find online:
The determination that “speech” is involved is just the beginning. It means that the case will be decided under the First Amendment. However, it does not guarantee the outcome. The right to speak is not absolute. A society in which the government was powerless to restrain citizens from speaking at any time or place, on any subject, however loudly they pleased, would be an insufferable place to live. The First Amendment does not strip the government of power to regulate speech; it prohibits the government from “abridging freedom of speech.” Deciding when a restriction “abridges freedom of speech” is what First Amendment jurisprudence is about; this determination calls for complex value judgments. (Kanovitz 2010, 45; my emphasis)
These value judgments are made through the use of a well-established concept that one never reads in the pro-“code is speech” evangelism: “scrutiny.” Scrutiny is part of what Con Law students also learn in their first year. It refers to the kinds of tests courts must apply to determine whether a law or regulation is allowable. The closer a speech act is to the core case—core political speech, such as an editorial endorsing a political candidate or passing of a law—the higher the level of scrutiny that the courts must apply. This highest level is called “strict scrutiny”; in these cases, the government must show a “compelling governmental interest” in the goal of the regulation, and that the regulation is “narrowly tailored” to accomplish this goal. Even in cases of core political speech, regulations that meet the “strict scrutiny” tests have passed and will continue to pass Supreme Court muster. The most familiar example is in content-based prohibitions on political speech within a certain distance of polling places on election days. Remember, that is an example of the most important kind of speech on anyone’s philosophy, and yet Courts have routinely ruled that, within these very limited contexts, actual prohibition on speech is fully legal.
Here’s material from a Con Law resource at the University of Missouri-Kansas that goes into some detail about the varying application of scrutiny tests with regard to the Equal Protection Clause:
Legislation frequently involves making classifications that either advantage or disadvantage one group of persons, but not another. States allow 20-year-olds to drive, but don’t let 12-year-olds drive. Indigent single parents receive government financial aid that is denied to millionaires. Obviously, the Equal Protection Clause cannot mean that government is obligated to treat all persons exactly the same–only, at most, that it is obligated to treat people the same if they are “similarly circumstanced.”
Over recent decades, the Supreme Court has developed a three-tiered approach to analysis under the Equal Protection Clause.
Most classifications, as the Railway Express and Kotch cases illustrate, are subject only to rational basis review. Railway Express upholds a New York City ordinance prohibiting advertising on commercial vehicles–unless the advertisement concerns the vehicle owner’s own business. The ordinance, aimed at reducing distractions to drivers, was underinclusive (it applied to some, but not all, distracting vehicles), but the Court said the classification was rationally related to a legitimate end. Kotch was a tougher case, with the Court voting 5 to 4 to uphold a Louisiana law that effectively prevented anyone but friends and relatives of existing riverboat pilots from becoming a pilot. The Court suggested that Louisiana’s system might serve the legitimate purpose of promoting “morale and esprit de corps” on the river. The Court continues to apply an extremely lax standard to most legislative classifications. In Federal Communications Commission v Beach (1993), the Court went so far as to say that economic regulations satisfy the equal protection requirement if “there is any conceivable state of facts that could provide a rational basis for the classification.” Justice Stevens, concurring, objected to the Court’s test, arguing that it is “tantamount to no review at all.”
Classifications involving suspect classifications such as race, however, are subject to closer scrutiny. A rationale for this closer scrutiny was suggested by the Court in a famous footnote in the 1938 case of Carolene Products v. United States (see box at left). Usually, strict scrutiny will result in invalidation of the challenged classification–but not always, as illustrated by Korematsu v. United States, in which the Court upholds a military exclusion order directed at Japanese-Americans during World War II. Loving v Virginia produces a more typical result when racial classifications are involved: a unanimous Supreme Court strikes down Virginia’s miscegenation law.
The Court also applies strict scrutiny to classifications burdening certain fundamental rights. Skinner v Oklahoma considers an Oklahoma law requiring the sterilization of persons convicted of three or more felonies involving moral turpitude (“three strikes and you’re snipped”). In Justice Douglas’s opinion invalidating the law we see the origins of the higher-tier analysis that the Court applies to rights of a “fundamental nature” such as marriage and procreation. Skinner thus casts doubt on the continuing validity of the oft-quoted dictum of Justice Holmes in a 1927 case (Buck v Bell) considering the forced sterilization of certain mental incompetents: “Three generations of imbeciles is enough.”
The Court applies a middle-tier scrutiny (a standard that tends to produce less predictable results than strict scrutiny or rational basis scrutiny) to gender and illegitimacy classifications.
There is nothing to the claim that the fundamental rights listed in the Bill of Rights or the UDHR simply block governments from all legislation and regulation. Among other things, such categorical principles would make it impossible to adjudicate cases where rights and responsibilities come into conflict—which, again contrary to the rhetoric of crypto-anarchists and other black-and-white thinkers, happens more often than not. Wikipedia links: Strict Scrutiny; Intermediate Scrutiny
5 Trackbacks
[…] David Golumbia, “Code Is Not Speech.” […]
[…] his blog Uncomputing, Virginia Commonwealth University professor David Golumbia tackles the question of code as speech, placing the matter appropriately at the heart of our current political paradox in which we feel […]
[…] the discourse of encryption advocates, a project I’ve addressed as well in pieces like “Code Is Not Speech” and “Tor, Technocracy, Democracy.” It’s a piece I haven’t before […]
[…] of people for abetting the vicious crime of human trafficking,” Golumbia, who advocates against code as a form of speech, told me. “Every other kind of media is held liable for participating in crimes like this, and we […]
[…] of people for abetting the vicious crime of human trafficking,” Golumbia, who advocates against code as a form of speech, told me. “Every other kind of media is held liable for participating in crimes like this, and we […]