Hate Speech and Private Disassociation
by Michael Dorf
In my Verdict column and an accompanying blog post last week, I proposed that Congress pass a statute imposing on internet domain registration, hosting, and other companies that provide essential services common carrier obligations, but that the statute ought to allow such companies, if they so choose, not to provide services to purveyors of hate speech, as defined by the statute. In both the column and the blog post, I acknowledged that the constitutionality of such an exception is unclear, given that, as construed by the Supreme Court, the First Amendment protects hate speech. Here I want to consider the question I bracketed last week: Would a common-carrier obligation with such an exception be valid?
I confess that after devoting considerable thought to this question, I find myself torn. Here I'll explain why I find this question difficult, mostly by exploring examples that pull in opposite directions.
If one thinks that hate speech (appropriately defined) ought not to be protected, then the case is easy. One will then think that the government itself should be permitted to censor hate speech, so of course it can allow private companies to censor hate speech. That would be the result in most other constitutional democracies, where hate speech is not protected. I don't wish to take a position on that basic question here. Instead, I'll simply assume for purposes of this essay that the First Amendment forbids Congress itself from proscribing hate speech (on the internet or elsewhere).
Well, one might say, if Congress cannot proscribe hate speech, it cannot delegate to a private party the power to proscribe hate speech. But that characterization of the statute I'm hypothesizing is not quite right. Congress is not delegating any power to private companies. Prior to the statute, companies like GoDaddy, Google, and CloudFlare (the companies that denied service to the neo-Nazi site The Daily Stormer) were already entitled to refuse to provide services to purveyors of hate speech; the statute didn't change that; it merely restricted their freedom with respect to other kinds of speech.
Yet if there is no issue of delegation, there is nonetheless state action. Congress, in imposing common carrier obligations that protect some speech but not other speech, clearly acted. Nor is the fact that the sum of the action did not reduce the speech rights of purveyors of hate speech dispositive. In denying registration to offensive trademarks, 15 U. S. C. §1052(a), did not restrict speech. After all, a trademark itself restricts speech (of infringing users of the mark), so denying registration, if anything, allowed for greater free speech. Nonetheless, the Supreme Court, in Matal v. Tam, struck down the provision as content-based. So, one might think, the relevant question is not whether Congress shut down speech or enhanced it, but whether it used a content-based line. And in my hypothetical example of a common-carrier obligation with a hate-speech exception, it does.
That said, the fact that the statute does not itself censor but leaves to the companies the choice whether to permit hate-speech should count for something. Consider religion. Congress itself may not discriminate on the basis of religion. And Congress was under no obligation to forbid private employers from discriminating on the basis of religion. When Congress did impose such a non-discrimination obligation, it also included an exception allowing religious employers to discriminate on the basis of religion. To be sure, under the Hosanna-Tabor decision there are circumstances in which the Free Exercise and Establishment Clauses required that exception, but the statutory exception appears to be broader than constitutionally required. And one would think that the exception would be constitutionally valid even if it were not constitutionally required. Why? Because there is an obvious difference between the government itself discriminating on the basis of religion and the government exempting certain private actors from a statutory prohibition on religious discrimination.
Can we say the same thing about the exception in my hypothetical example? Maybe, but maybe the exception in Title VII is not quite the right analogy. Suppose that instead of authorizing religious institutions to discriminate in favor of co-religionists--which is, or at least could be, based in benign motives--Congress included an exemption in Title VII that permitted religious discrimination against one particular religion, Muslims say. Surely that would be unconstitutional religious discrimination by Congress, because, instead of allowing a category of discrimination to be specified by the regulated private actors, Congress itself would be allowing discrimination against only one particular group.
It could be argued that the exception allowing internet companies to exclude purveyors of hate speech is more like the anti-Muslim provision than the actual co-religionist exception in Title VII. After all, the free speech analogy to particular religions is particular content, and in my proposed statute the firms are given permission to discriminate against only one particular sort of message: hate speech.
But wait. In the example of the anti-Muslim exception to Title VII, it is hard to escape the conclusion that Congress itself is hostile to Muslims. Is it clear that Congress itself is hostile to hate speech in my proposed statute? Not necessarily.
Suppose that instead of (or in addition to) permission to the internet companies to deny service to purveyors of hate speech, Congress included an exception allowing such companies to deny service to websites that contain sexually explicit, even if not legally obscene, materials. Such an exception might be motivated by congressional hostility to sexual content, but it also might be motivated by something else: The goal of allowing companies that do not wish to associate with sexually explicit content--to run a G-rated company, as it were--to do so. Readers may have various reactions to this example, but my own intuition runs in its favor.
Where does that leave me? I regard the hate-speech exception as somewhat closer to the sexually explicit material exception than to the anti-Muslim discrimination exception, but I freely admit that others could reasonably disagree.
So . . . if I were a member of Congress, I think I would propose the common carrier statute with its hate speech exception, but I would include in the statute a nonseverability clause, making clear that if a court strikes down the exception, then the entire common carrier obligation goes away. Why? Because the current status quo more or less mirrors where I think the law should place it: Internet companies generally regard themselves as platforms for free speech, but they draw the line at hate speech. If, at some future time, internet companies start using their freedom (as non-state actors) from constitutional restraint in order to deny service to unpopular speech that falls well short of hate speech, I might then argue for a statute imposing common carrier obligations without exception.
In my Verdict column and an accompanying blog post last week, I proposed that Congress pass a statute imposing on internet domain registration, hosting, and other companies that provide essential services common carrier obligations, but that the statute ought to allow such companies, if they so choose, not to provide services to purveyors of hate speech, as defined by the statute. In both the column and the blog post, I acknowledged that the constitutionality of such an exception is unclear, given that, as construed by the Supreme Court, the First Amendment protects hate speech. Here I want to consider the question I bracketed last week: Would a common-carrier obligation with such an exception be valid?
I confess that after devoting considerable thought to this question, I find myself torn. Here I'll explain why I find this question difficult, mostly by exploring examples that pull in opposite directions.
If one thinks that hate speech (appropriately defined) ought not to be protected, then the case is easy. One will then think that the government itself should be permitted to censor hate speech, so of course it can allow private companies to censor hate speech. That would be the result in most other constitutional democracies, where hate speech is not protected. I don't wish to take a position on that basic question here. Instead, I'll simply assume for purposes of this essay that the First Amendment forbids Congress itself from proscribing hate speech (on the internet or elsewhere).
Well, one might say, if Congress cannot proscribe hate speech, it cannot delegate to a private party the power to proscribe hate speech. But that characterization of the statute I'm hypothesizing is not quite right. Congress is not delegating any power to private companies. Prior to the statute, companies like GoDaddy, Google, and CloudFlare (the companies that denied service to the neo-Nazi site The Daily Stormer) were already entitled to refuse to provide services to purveyors of hate speech; the statute didn't change that; it merely restricted their freedom with respect to other kinds of speech.
Yet if there is no issue of delegation, there is nonetheless state action. Congress, in imposing common carrier obligations that protect some speech but not other speech, clearly acted. Nor is the fact that the sum of the action did not reduce the speech rights of purveyors of hate speech dispositive. In denying registration to offensive trademarks, 15 U. S. C. §1052(a), did not restrict speech. After all, a trademark itself restricts speech (of infringing users of the mark), so denying registration, if anything, allowed for greater free speech. Nonetheless, the Supreme Court, in Matal v. Tam, struck down the provision as content-based. So, one might think, the relevant question is not whether Congress shut down speech or enhanced it, but whether it used a content-based line. And in my hypothetical example of a common-carrier obligation with a hate-speech exception, it does.
That said, the fact that the statute does not itself censor but leaves to the companies the choice whether to permit hate-speech should count for something. Consider religion. Congress itself may not discriminate on the basis of religion. And Congress was under no obligation to forbid private employers from discriminating on the basis of religion. When Congress did impose such a non-discrimination obligation, it also included an exception allowing religious employers to discriminate on the basis of religion. To be sure, under the Hosanna-Tabor decision there are circumstances in which the Free Exercise and Establishment Clauses required that exception, but the statutory exception appears to be broader than constitutionally required. And one would think that the exception would be constitutionally valid even if it were not constitutionally required. Why? Because there is an obvious difference between the government itself discriminating on the basis of religion and the government exempting certain private actors from a statutory prohibition on religious discrimination.
Can we say the same thing about the exception in my hypothetical example? Maybe, but maybe the exception in Title VII is not quite the right analogy. Suppose that instead of authorizing religious institutions to discriminate in favor of co-religionists--which is, or at least could be, based in benign motives--Congress included an exemption in Title VII that permitted religious discrimination against one particular religion, Muslims say. Surely that would be unconstitutional religious discrimination by Congress, because, instead of allowing a category of discrimination to be specified by the regulated private actors, Congress itself would be allowing discrimination against only one particular group.
It could be argued that the exception allowing internet companies to exclude purveyors of hate speech is more like the anti-Muslim provision than the actual co-religionist exception in Title VII. After all, the free speech analogy to particular religions is particular content, and in my proposed statute the firms are given permission to discriminate against only one particular sort of message: hate speech.
But wait. In the example of the anti-Muslim exception to Title VII, it is hard to escape the conclusion that Congress itself is hostile to Muslims. Is it clear that Congress itself is hostile to hate speech in my proposed statute? Not necessarily.
Suppose that instead of (or in addition to) permission to the internet companies to deny service to purveyors of hate speech, Congress included an exception allowing such companies to deny service to websites that contain sexually explicit, even if not legally obscene, materials. Such an exception might be motivated by congressional hostility to sexual content, but it also might be motivated by something else: The goal of allowing companies that do not wish to associate with sexually explicit content--to run a G-rated company, as it were--to do so. Readers may have various reactions to this example, but my own intuition runs in its favor.
Where does that leave me? I regard the hate-speech exception as somewhat closer to the sexually explicit material exception than to the anti-Muslim discrimination exception, but I freely admit that others could reasonably disagree.
So . . . if I were a member of Congress, I think I would propose the common carrier statute with its hate speech exception, but I would include in the statute a nonseverability clause, making clear that if a court strikes down the exception, then the entire common carrier obligation goes away. Why? Because the current status quo more or less mirrors where I think the law should place it: Internet companies generally regard themselves as platforms for free speech, but they draw the line at hate speech. If, at some future time, internet companies start using their freedom (as non-state actors) from constitutional restraint in order to deny service to unpopular speech that falls well short of hate speech, I might then argue for a statute imposing common carrier obligations without exception.