Trump's Unwarranted & Dangerous Attack on Google Inadvertently Raises a Serious Issue
by Michael Dorf
On Tuesday, President Trump tweeted that Google had "rigged" its search results to yield only "fake news" -- i.e., accurate stories from reputable sources that portray him in a negative light -- when one searches for news about him. He added that this "very serious situation . . . will be addressed." Later that day, Trump extended his warning to Facebook and Twitter, which he apparently believes are also rigging their algorithms to promote anti-Trump "fake news" at the expense of more pro-Trump sources.
The accusation is almost certainly nonsense. I say "almost certainly" because Google does not make its algorithm public, and so it is impossible to know for certain that deep within it there is no line of code that favors anti-Trump stories, but Google has certainly denied any such "rigging." Facebook and Twitter are a bit different. As social media sites, their algorithms give prominence to material shared and liked by others one likes and/or follows. Consequently, someone who is liberal will find stories that skew liberal; someone who is conservative will find stories that skew conservative; and someone who only likes cat videos will find stories that skew feline. But there appears to be no evidence that any of these tech companies is deliberately skewing the results--except to the extent that reality has a liberal bias.
Accordingly, in the current environment, one can best understand the Trump accusation and threat as an effort to chill speech. If Trump were to follow up on his threats--as Larry Kudlow has suggested he might--that would be a clear First Amendment violation, although proving that any regulatory action undertaken by the Trump team was motivated by the aim of censorship would be tricky. As I wrote last November, I thought that there was sufficient evidence that Trump's DOJ was seeking to block the AT&T/Time Warner merger as a result of Trump's attacks on CNN to warrant discovery. However, the district judge ultimately kept such evidence out. And the SCOTUS travel ban decision this past June suggests that we have entered an era in which judges feel it is not acceptable to admit that the current president routinely acts out of unconstitutional motives, even when he repeatedly says that's exactly what he's doing.
Suppose, however, that Trump were right. Suppose that Google and Facebook actually were deliberately skewing their algorithms to show users anti-Trump material and to suppress pro-Trump material. That would not be unconstitutional, of course, because Google and Facebook are private actors to which the First Amendment does not apply. But might the government be able to block them from doing so without violating their First Amendment rights? That is the serious issue that Trump has inadvertently raised. The answer is not entirely clear.
Let's start with a little background. Under the since-discarded "fairness doctrine," broadcasters were required to provide fair coverage of various sides of controversial issues as a condition of retaining their broadcast licenses. The rule was applied to a radio station that was ordered to give one Cook, who had been the object of severe criticism during a radio program that aired on the station, an opportunity to respond. The radio station objected that the application of the fairness doctrine to require it to air speech with which it disagreed violated the First Amendment.
In Red Lion v. FCC, the SCOTUS rejected the challenge and upheld the application of the fairness doctrine. Crucial to the Court's reasoning were the following propositions: (1) Electromagnetic spectrum is a scarce resource that the government must allocate or else risk it becoming unusable in the resulting cacophony; (2) that power to exclude means that licensees are, in some sense, acting on behalf of the government, so that private censorship is a kind of government censorship; and (3) the discretionary nature of the license also means that First Amendment rights in this context belong to the listeners, rather than the broadcasters.
Red Lion was decided in 1969. Five years later, in Miami Herald v. Tornillo, the Court reached more or less the opposite result with respect to a state right-of-reply statute applicable to newspapers. Although the Court in Miami Herald recognized the existence of media concentration--such that the typical metropolitan area has fewer newspapers than radio stations--it nonetheless found that a newspaper's de facto monopoly was an insufficient basis for the regulation. Oddly, even though the appellee's brief relied extensively on Red Lion, the Court in Miami Herald did not cite it at all.
The Reagan administration's FCC revoked the fairness doctrine, and the Court has not revisited the issue since Miami Herald, which leaves the law in a state of some confusion. Did Miami Herald overrule Red Lion sub silentio? If not, how should the cases be reconciled? The "scarcity rationale" is the usual candidate for harmonizing the cases, but it has been subject to extensive criticism. The best critique goes something like this: Granting that the scarcity of the electromagnetic spectrum justifies licensing, it hardly follows that such scarcity justifies content regulation.
Moreover, even if the scarcity of electromagnetic spectrum once justified something like a fairness doctrine, spectrum scarcity no longer is salient, given the possibility of streaming virtually any possible content over the Internet. Hence, even if we think of Google and Facebook as monopolies, they are monopolies in the way that the Miami Herald was a monopoly in Miami--and that's not a sufficient basis for content regulation of the sort that Trump supposes.
But maybe it should be a sufficient basis. Maybe the Court in Miami Herald was wrong. We might think that a de facto monopoly combined with real barriers to entry should suffice for the government to require a firm--even a newspaper--to act as a common carrier, at least in some respects.
Even if so, however, I would distinguish between a search engine like Google and a social media site like Facebook. Suppose it were true that Google skewed results in favor of anti-Trump sites. Pro-Trumpers could create or patronize a pro-Trump (or neutral) search engine. Perhaps one of the existing alternatives (Bing, Yahoo!, etc.) would do the job. Granted, these alternatives aren't nearly as popular as Google, but the popularity of Google rests mostly on its superiority. That superiority rests a bit on network effects, but not nearly as much as the value of Facebook rests on network effects.
Many people are unhappy with Facebook for all sorts of good and bad reasons, but they continue to use Facebook, because that's where their friends and relatives are. The success of Facebook is more a story of path-dependence and network effects than is the success of Google. Facebook is like the QWERTY keyboard or VHS, whereas Google is a superior product.
Or at least so it appears to me. Maybe I'm wrong. If I'm right, however, that perhaps ought to have First Amendment implications. If one thinks that Red Lion was justifiable and that it should have continuing vitality at least in those domains in which the market--left to its own devices--will yield insufficiently diverse points of view for reasons having nothing to do with the popularity of such points of view, then there would be better reason for the government to require evenhandedness of a network-effect-based quasi-monopolist like Facebook than an entity like Google.
But of course, all of this supposes--without any basis in fact--that Google and Facebook have been deliberately attaching an anti-Trump skew to their algorithms. Back here in reality, the Trump proposal is simply more of his usual dangerous demagoguery attacking the free press.
On Tuesday, President Trump tweeted that Google had "rigged" its search results to yield only "fake news" -- i.e., accurate stories from reputable sources that portray him in a negative light -- when one searches for news about him. He added that this "very serious situation . . . will be addressed." Later that day, Trump extended his warning to Facebook and Twitter, which he apparently believes are also rigging their algorithms to promote anti-Trump "fake news" at the expense of more pro-Trump sources.
The accusation is almost certainly nonsense. I say "almost certainly" because Google does not make its algorithm public, and so it is impossible to know for certain that deep within it there is no line of code that favors anti-Trump stories, but Google has certainly denied any such "rigging." Facebook and Twitter are a bit different. As social media sites, their algorithms give prominence to material shared and liked by others one likes and/or follows. Consequently, someone who is liberal will find stories that skew liberal; someone who is conservative will find stories that skew conservative; and someone who only likes cat videos will find stories that skew feline. But there appears to be no evidence that any of these tech companies is deliberately skewing the results--except to the extent that reality has a liberal bias.
Accordingly, in the current environment, one can best understand the Trump accusation and threat as an effort to chill speech. If Trump were to follow up on his threats--as Larry Kudlow has suggested he might--that would be a clear First Amendment violation, although proving that any regulatory action undertaken by the Trump team was motivated by the aim of censorship would be tricky. As I wrote last November, I thought that there was sufficient evidence that Trump's DOJ was seeking to block the AT&T/Time Warner merger as a result of Trump's attacks on CNN to warrant discovery. However, the district judge ultimately kept such evidence out. And the SCOTUS travel ban decision this past June suggests that we have entered an era in which judges feel it is not acceptable to admit that the current president routinely acts out of unconstitutional motives, even when he repeatedly says that's exactly what he's doing.
Suppose, however, that Trump were right. Suppose that Google and Facebook actually were deliberately skewing their algorithms to show users anti-Trump material and to suppress pro-Trump material. That would not be unconstitutional, of course, because Google and Facebook are private actors to which the First Amendment does not apply. But might the government be able to block them from doing so without violating their First Amendment rights? That is the serious issue that Trump has inadvertently raised. The answer is not entirely clear.
Let's start with a little background. Under the since-discarded "fairness doctrine," broadcasters were required to provide fair coverage of various sides of controversial issues as a condition of retaining their broadcast licenses. The rule was applied to a radio station that was ordered to give one Cook, who had been the object of severe criticism during a radio program that aired on the station, an opportunity to respond. The radio station objected that the application of the fairness doctrine to require it to air speech with which it disagreed violated the First Amendment.
In Red Lion v. FCC, the SCOTUS rejected the challenge and upheld the application of the fairness doctrine. Crucial to the Court's reasoning were the following propositions: (1) Electromagnetic spectrum is a scarce resource that the government must allocate or else risk it becoming unusable in the resulting cacophony; (2) that power to exclude means that licensees are, in some sense, acting on behalf of the government, so that private censorship is a kind of government censorship; and (3) the discretionary nature of the license also means that First Amendment rights in this context belong to the listeners, rather than the broadcasters.
Red Lion was decided in 1969. Five years later, in Miami Herald v. Tornillo, the Court reached more or less the opposite result with respect to a state right-of-reply statute applicable to newspapers. Although the Court in Miami Herald recognized the existence of media concentration--such that the typical metropolitan area has fewer newspapers than radio stations--it nonetheless found that a newspaper's de facto monopoly was an insufficient basis for the regulation. Oddly, even though the appellee's brief relied extensively on Red Lion, the Court in Miami Herald did not cite it at all.
The Reagan administration's FCC revoked the fairness doctrine, and the Court has not revisited the issue since Miami Herald, which leaves the law in a state of some confusion. Did Miami Herald overrule Red Lion sub silentio? If not, how should the cases be reconciled? The "scarcity rationale" is the usual candidate for harmonizing the cases, but it has been subject to extensive criticism. The best critique goes something like this: Granting that the scarcity of the electromagnetic spectrum justifies licensing, it hardly follows that such scarcity justifies content regulation.
Moreover, even if the scarcity of electromagnetic spectrum once justified something like a fairness doctrine, spectrum scarcity no longer is salient, given the possibility of streaming virtually any possible content over the Internet. Hence, even if we think of Google and Facebook as monopolies, they are monopolies in the way that the Miami Herald was a monopoly in Miami--and that's not a sufficient basis for content regulation of the sort that Trump supposes.
But maybe it should be a sufficient basis. Maybe the Court in Miami Herald was wrong. We might think that a de facto monopoly combined with real barriers to entry should suffice for the government to require a firm--even a newspaper--to act as a common carrier, at least in some respects.
Even if so, however, I would distinguish between a search engine like Google and a social media site like Facebook. Suppose it were true that Google skewed results in favor of anti-Trump sites. Pro-Trumpers could create or patronize a pro-Trump (or neutral) search engine. Perhaps one of the existing alternatives (Bing, Yahoo!, etc.) would do the job. Granted, these alternatives aren't nearly as popular as Google, but the popularity of Google rests mostly on its superiority. That superiority rests a bit on network effects, but not nearly as much as the value of Facebook rests on network effects.
Many people are unhappy with Facebook for all sorts of good and bad reasons, but they continue to use Facebook, because that's where their friends and relatives are. The success of Facebook is more a story of path-dependence and network effects than is the success of Google. Facebook is like the QWERTY keyboard or VHS, whereas Google is a superior product.
Or at least so it appears to me. Maybe I'm wrong. If I'm right, however, that perhaps ought to have First Amendment implications. If one thinks that Red Lion was justifiable and that it should have continuing vitality at least in those domains in which the market--left to its own devices--will yield insufficiently diverse points of view for reasons having nothing to do with the popularity of such points of view, then there would be better reason for the government to require evenhandedness of a network-effect-based quasi-monopolist like Facebook than an entity like Google.
But of course, all of this supposes--without any basis in fact--that Google and Facebook have been deliberately attaching an anti-Trump skew to their algorithms. Back here in reality, the Trump proposal is simply more of his usual dangerous demagoguery attacking the free press.