Should We Regulate Speech to Reduce “Noise”?

​Information consumers no doubt face more “noise” in public channels than ever. A new article by Prof. Justin (Gus) Hurwitz suggests a novel, but misguided, approach to regulating speech – one that reflects a new spirit of hubris in many legal, academic, and political circles.
 
The article, dubbed “Noisy Speech Externalities,” presents something of a counterargument to Justice Holmes’ dissent in Abrams v. United States, which introduced the idea that more speech is the best answer to bad speech – what he called a “free trade in ideas.” In light of recent scholarship, free and unregulated speech seems to some to be too quaint for our times.
“Consumers of information face a glut of information that overwhelms their ability to process it all,” Hurwitz argues. Excessive “noise,” he suggests, harms the underlying goal of allowing unfettered speech because people become too confused to filter out the legitimate information.
 
Hurwitz has a solution to this “market failure” in speech. He wants to utilize an EPA-like pollution control framework that would require social networks to use “best available” content moderation technologies that would filter out speech that muddies the public discourse. Who decides whether some speech should be silenced as “pollution” or “noise” remains conspicuously unclear.
 
We should, it seems, just leave those questions to the “experts” and their “best available” algorithms.
 
Hurwitz draws on Claude Shannon’s “information theory,” which posits that a signal-to-noise ratio governs the extent to which a listener may distinguish good information from bad information. In other words, the more noise, the less discernible a signal. No question, this is the reality of social media today.
 
What Hurwitz fails to reckon with is the practical application of his theoretical framework. In arguing that excessive speech constitutes a market externality meriting a gag order, Hurwitz suggests that post-grad tech workers should be the arbiters of our civil discourse, with virtually no constraints on their power. This, he says, should be the contingent basis upon which Section 230’s liability shield is applied.
 
Speech theorists of this sort seem to lack a basic appreciation for the efficacy of speech over time. Surely, the revolutionary mobs who protested the Stamp Act were “noisy.” So, too, were civil rights demonstrators whose cacophonous outcries muddled the signal-to-noise ratio in the 1960s.
 
Passionate, inflamed speech may result in short-term discomfort; in the long run, it’s still the only way to build consensus for or against a starkly new proposition. Witness the shifting debate on the origins of covid, or the efficacy of masks and lockdowns. Applying outmoded “command-and-control” regulatory solutions is sure to repress valuable insights.
 
This paper has glimmers of the thinking of Marxist philosopher Herbert Marcuse. In his influential essay, Repressive Tolerance, Marcuse argued for the censorship of ideas and the repression of people who advocate positions that could stymie what to him was the clear Marxist direction of history. “Suppression of the regressive ones,” Marcuse wrote, “is a prerequisite for the strengthening of the progressive ones.”
 
Can one concoct a turn of phrase more thoroughly Orwellian than “repressive tolerance”? In a similar though less strident vein, Hurwitz suggests that social media organizations should make First Amendment policy decisions based on their own vague – and likely biased – interpretations of what constitutes a vibrant marketplace of ideas with an appropriate noise level.
 
There may well be a better answer on social media discourse and content moderation. The best answer of all is quality education. Better schooling is the missing piece needed to make consumers more discerning. The best information filters are the ones we carry in our heads.