My last post set out the case that the International Covenant on Civil and Political Rights (ICCPR) offered strong protections to online speech on social media. Let's turn now to assessing that case. That case depended on Article 19 of the ICCPR which established both a broad right to free speech and a tripartite test for restrictions on speech by governments. Some have argued that the vagueness prong of the tripartite test would invalidate many “hate speech” restrictions. Let’s imagine social media companies adopt ICCPR in total. Would Article 19’s tripartite test in fact invalidate restrictions on speech rights online?

The “would” in my question reflects a subjunctive mood. I am assuming that ICCPR is not in fact now applied to speech on social media. But that may be a false assumption.

In 2019, the U.N.’s Special Rapporteur on the Promotion and Protection of the Right to Freedom of Opinion and Expression argued that social media companies should apply international human rights on their platform including the tripartite test. David Kaye, the Special Rapporteur in question, has extensive knowledge of social media content moderation; he has published a well-regarded book on the topic. His normative call for action to the companies suggests they have not in fact adopted the tripartite test in their internal moderation. The experience of social media cannot tell us much about the empirical success or failure of the tripartite test.

The Special Rapporteur also notes the tripartite test obligates states that ratified the ICCPR. Accordingly, governments should have extensive experience applying the tripartite test. Apparently they lack such experience since the Special Rapporteur and the UN Human Rights Committee continue to note failures by national governments to abide by international human rights. Indeed the Special Rapporteur and the UN Human Rights Committee evaluate how well governments follow international human rights law. But they are not courts enforcing rights against recalcitrant as well as compliant malefactors.

Perhaps we can find something similar to the tripartite test in an individual nation. The test does look somewhat like the “strict scrutiny” test in American constitutional law. Courts apply “strict scrutiny” when the government restricts speech based on its content. Such restrictions may be valid only if they further a “compelling government interest” and are narrowly tailored to achieve that end. Is that a stringent test? The renowned law professor Gerald Gunther once claimed strict scrutiny was “strict in theory, fatal in fact.” In other words, when judges applied the test, they had in effect decided to strike down a government law or action. Was Gunther right?

An impressive authority doubted Gunther’s claim. In Adarand Constructors v. Pena (1995), Justice Sandra Day O'Connor's majority opinion sought “to dispel the notion that strict scrutiny is 'strict in theory, but fatal in fact.’" That remark prompted law professor Adam Winkler to test Gunther’s claim empirically. How often in fact was strict scrutiny fatal to government action?

Winkler found that between 1990 and 2003, the U.S. Supreme Court applied strict scrutiny 12 times while the lower federal courts did so in a conclusive, final ruling 447 times. Winkler found that in 30 percent of these decisions a court upheld a government action. Overall, strict scrutiny seemed risky to the government but hardly fatal.

But our interest is more focused. Does strict scrutiny protect speech from government regulation? Winkler found that almost half his dataset involved strict scrutiny being applied to speech restrictions. Overall, government regulation of speech survived in 22 percent of the cases examined. That may seem like a poor performance, but strict scrutiny was “stricter” in speech cases than in any other category in Winkler’s database. Winkler also broke down the “speech” category of 222 cases into 7 subcategories. He found that two categories were almost “fatal in fact.” The categories of “public forum viewpoint discrimination” and “sign ordinances” saw exactly 1 government regulation survive scrutiny in 46 cases, a survival rate of 2 percent. A third category “indecency regulation” was different: one-third of government regulations of speech survived strict scrutiny. His four other categories are not clearly relevant to the task at hand.

Let’s drill down a bit in each category. Viewpoint discrimination is “is a form of content discrimination particularly disfavored by the courts.” As Winkler explained, “The typical viewpoint discriminatory public forum restriction bars a political or religious group from using a school or library's public rooms, speaking at city council or other governmental meetings, displaying art in government buildings or spaces, or taking out advertisements on buses or ‘adopt-a-highway’ programs.” Courts evidently found it difficult to justify such direct limits on core speech. The “sign ordinance” category involves local governments regulating yard signs in order to maintain “a visually pleasing aesthetic in neighborhoods and traffic control,” a rather uncompelling government interest. The indecency category involved “the obviously compelling governmental interest of protecting minors from harms thought to stem from access to sexually explicit materials.” Nonetheless, “courts usually held that less restrictive alternatives were available that posed fewer hurdles for adults' constitutionally protected access to indecent material.” So even with children said to be in the balance, strict scrutiny had teeth.

In the American speech context, “strict scrutiny” is far from fatal to government regulations. But when content discrimination or a minor government interest is in play, such tests can be all but fatal to government overreach. What can we expect if international norms like the tripartite test are applied to social media? Many social media limits on speech are likely to be vague or more harmful to speech than alternative policies. They will fail the tripartite test outlined by the Special Rapporteur. But the test must be applied by an institution that gives priority to Article 19’s declaration favoring free expression. Does such an institution exist? Not yet. If such a body were created, the tripartite test could be a useful tool in fact (and not just theory) to vindicate free expression online if its members cared a lot about free expression.