INTRODUCTION
A century of developments in communications technology has done wonders for listeners. In the not-so-distant past, the listener had few choices regarding the speech they heard. An urban listener might find a range of choice in libraries or bookstores, or among private associations’ meetings or periodicals. But mostly listeners heard what others happened to say, on the job, at church, in the neighborhood, and on the street-corner. Today, listeners have more choices than ever, and more ability to choose which speech to hear and which not to hear.
First Amendment doctrine, for its part, has occasionally shielded these listeners’ choices about whom to listen to, even over speakers’ choices about whom to speak to. I largely agree with the authors in this symposium who argue that it should do so more often. The appeal is evident. Communication goes most smoothly with a willing speaker and a willing listener. And with listeners now awash in speech options, someone must choose among them. The listener seems obviously superior, as chooser, to many looming alternatives: the stalker, the fake-news purveyor, the oligopolistic media platform, or—most antithetical to the First Amendment—the government.
Yet this Commentary urges caution in so fully embracing listener’s choice that it becomes a paradigm of First Amendment jurisprudence. One can easily move from recognizing the advantages of willing listening (and speaking) to identifying the “core” of constitutionally protected speech as not just a “joint activity”—to use Ashutosh Bhagwat’s illuminating term—but mutually consensual, affirmatively chosen by both speaker and listener. One might even begin seeing something constitutionally suspect about speech that lacks a consenting listener. On this interpretation, the fact that the speaker on his soapbox in Washington Square Park annoys some passersby is, while inevitable, still unfortunate—a sacrifice made for the sake of the other, more willing passersby.
Yet in a not-too-distant future, this fact might be “correctible” by technology, assisted by artificial intelligence and mass data collection. Every webpage you visit might seamlessly filter out any content you have elected not to see; or software might elect for you, based on your eyes’ patterns of lingering. When you step onto the street, you might wear digital glasses to blur out all written speech that you have deemed unwelcome, and digital headphones to cancel all noise except oral speech you have deemed welcome.
While this seems a paradise for listeners’ choice, it would not be one for listeners’ interests, another traditional First Amendment concern. The speech that we choose to hear does not always make us better off. Certainly, a listener’s choice is one aspect of their interests. And speech they vehemently reject—such as harassing speech—will rarely be in their interest. But a gray zone exists between chosen listening and coerced listening, which I will call unchosen listening. Some unchosen listening, I will argue, is a desirable part of being an autonomous person and citizen.
I will also argue that it is desirable for the broader society of which listeners are a part—including its collective knowledge, culture, and, especially, democratic institutions. These societal interests underlie the First Amendment, too, as Alan Chen also discusses in this symposium. Perhaps for these reasons, unchosen listening has been revered in First Amendment doctrine. As the Supreme Court declared in Terminiello v. Chicago, speech “may indeed best serve its high purpose when it induces a condition of unrest, creates dissatisfaction with conditions as they are, or even stirs people to anger. Speech is often provocative and challenging. It may strike at prejudices and preconceptions and have profound unsettling effects as it presses for acceptance of an idea.”
Indeed, the most fiercely protected First Amendment speech—perhaps its real “core” case—involves non-consensual listeners: speech in the public forum. This means citizens trying to move or persuade fellow citizens, and especially those none too thrilled to listen. It is—for those familiar with the First Amendment landmarks—Abrams’s communist pamphlets thrown from the roof, Kovacs’s pro-labor commentary amplified from his truck, Mosley’s sign decrying race discrimination, Cohen’s jacket protesting the draft. And it is arguably the only speech government has a (currently recognized) positive obligation to protect—by preserving the openness of traditional public forums—under the First Amendment.
While the Court has seldom expounded on the theme of unchosen listening, its silence invites less doctrinally bound theorizing. This Commentary thus offers a brief normative case for resisting a listener’s-choice paradigm (and maintaining more of a public forum paradigm).
I. The Case for unchosen listening
What speech will a listener hear, if they mostly choose that speech themselves or have it selected for them based on algorithmic predictions of their preferences? Chances are, it will be fairly insular and homogeneous. That is, it will be familiar, comfortable, and unchallenging; and it will come from like-minded speakers and a small number of sources. Its insularity and homogeneity will be reinforced to the extent that the listener chooses with greater granularity—that is, chooses not just among sources or topics but among specific speech from each source or specific views on each topic. The result might resemble “The Daily Me” envisioned by Cass Sunstein. While not all listeners will choose like this, and few will do so all the time, enough listeners will likely do so in enough cases to affect the overall patterns of communication in society.
Assuming this is true, the following discussion identifies five significant interests that can be undermined when listening becomes too tailored to a listener’s choices: preference optimization, error avoidance, innovation, self-evolution, and tolerance. Each interest has an individual dimension: for listeners themselves. Each also has a collective dimension: for a society in which the total pool of speech consumed is determined largely by listeners’ choices.
A. Optimizing Preferences
Even listener-chosen speech does not always satisfy listener preferences. Speech often defies expectations. Sometimes we have a “meh” reaction to a movie by our favorite director, or are spellbound by a movie we expected to bore. Our expectations are sometimes even insufficient for choice. I may have almost no sense of whether I would like a movie from an unfamiliar genre, because I have not experienced anything like it before. After all, the indicators of content that are available prior to hearing speech—for example, its speaker, venue, general subject, or reviews—are necessarily limited.
With ever more personalized listening technology, our mistaken choices—or mistaken predictions of our choices by algorithms—can lock us into inferior patterns of listening. Mistaken choices are easily corrected when they produce dissatisfaction, but less so when they produce merely suboptimal satisfaction. Say that I have watched and liked one Fellini film and one Spielberg film. Perhaps I am directed, by choice or algorithm, toward more Spielberg films rather than Fellini films. I continue to like Spielberg films and hence grow my love of Spielberg. I may never know the counterfactual––whether I would have preferred the Fellini films. The best chance of exiting a suboptimal equilibrium is exposure to unchosen speech, or at least less-precisely-chosen speech. Optimizing, therefore, involves reflective equilibrium between chosen and unchosen exposure.
A collective cultural danger lurks here, too, at least insofar as listeners rely on algorithmic services to make these finer speech selections for them. Because these services operate within a market economy, their corporate makers may face incentives to nudge listeners toward more popular, hence more easily satisfiable, preferences. Forced to rely on the same services to find listeners, speakers—including musicians, writers, filmmakers, journalists, and others—may also feel pressure to create content to win the algorithm’s game, as many have already. The result may be a blander culture and public sphere, with fewer options over which to optimize our preferences.
B. Avoiding Error
Listening primarily to chosen speech may proliferate error. The first problem is that it may skew our available evidence. If, as assumed, we choose to listen to speakers who share our beliefs, including false ones, then they will be unlikely to offer evidence against those beliefs. They may even pass on new false beliefs based on old ones. These harms are well-documented within “epistemic bubbles,” that is, discussion spaces in which certain views are systematically excluded.
A second problem is that too much chosen listening may objectionably narrow the range of subjects on which we possess evidence. People may choose to listen to little or no speech concerning decisions they will or must take, including voting. For instance, research suggests that the most promising ways to inform voters may involve incorporating speech that they do not choose alongside speech they do—such as entertainment.
Yet a third problem is that a steady diet of chosen speech may dull our capacity to think itself. As John Stuart Mill observed 166 years ago, reasoning is better stimulated and optimized by speech that is uncomfortable and challenging. But chosen speech is, by assumption, typically the opposite. Chosen speech can exacerbate self-favoring and in-group-favoring cognitive biases. Chosen speech may not even offer reasons, because it does not seek to persuade. Chosen speech can, by repeating or just assuming our beliefs, reinforce and bury them so deeply that we have trouble getting the “mental distance” necessary to scrutinize them—to question them and their justification.
C. Innovation
Unchosen listening can stimulate not just rational thought, but more creative and generative forms of thought that do not merely analyze evidence but propose new ideas or solutions.
Creativity is by its nature unexpected. It often involves making connections among facts, ideas, disciplines, cultures, and so forth that others—including oneself—had never thought to draw. We are thus sometimes likelier to arrive at an innovative solution not by continuing to think directly about the problem—constrained by our previous expectations—but by allowing our mind to wander and randomly associate. The revolutionary inventions, discoveries, insights, and movements that have been prompted by contact with unusual contexts, other disciplines, or other cultures are countless. Consider an eclectic sampling. Clinical psychology has been seriously influenced by Buddhist religious practices. Impressionism in painting was largely inspired by Japanese ukiyo-e prints; and the painting of Jackson Pollock, darling of the Supreme Court’s own First Amendment cases, was likely inspired by Native American sandpainting. The QR code was suggested by the Go board, while the foldable shape of heart stents was suggested by Japanese origami. Numerous engineering innovations, from aeronautics to robotics, have come from observing animals. The latest monumental computer science invention, large language models, grew from insights in both linguistics and cognitive science. And this symposium’s own Aziz Huq uses observations in sociology for thinking about digital speech.
Creativity thus seems likeliest to be prompted by exposure to speech that is unexpected, or at least unguided by our own choices. Unchosen speech can offer new concepts and frameworks of thinking that supplement or integrate with existing ones. Or it can simply interrupt habitual frameworks—generating another (creative) form of mental distance—and thus open the mind to the potential relevance of unexpected inputs. By contrast, chosen speech can lock in certain expectations of relevance.
One might expect collective, rather than individual, innovation to be fueled through highly personalized listening. It might segregate most listeners into speech subcultures, within which they could refine their own insular ideas—like members of a species separated among islands and continuing to evolve. But then a few innovators sampling from those radically diverse subcultures might produce innovations of even greater enormity. Yes, it seems just as plausible that, so long as enough listening is chosen, occasional sampling by a greater number and diversity of listeners would provide offsetting benefits of more sophisticated—if slightly less radically distinct—subcultures and more frequent innovation.
D. Self-Evolution
Persons arguably have a fundamental autonomy interest in being able not just to make up their minds, but to change them, too. This ability to change applies to their beliefs, as described above, but also their ends—their important goals, values, commitments, and so forth—and their tastes. I do not mean the mere freedom to zig when one previously zagged. Certain factors beyond our control inevitably change over time: our external circumstances vary, and our set of experiences and beliefs expands. In response, it is often appropriate or even necessary to change ourselves in order to cohere, adapt, and grow. Otherwise, we become frozen into a course determined not by us, but by the “dead hand” of our past self. Indeed, one might say that autonomous self-development just is the continual evolution of our self in response to new, external inputs.
It is almost always possible to change our ends, at least insofar as nothing physically prevents us. But pursuing our ends is often habitual, or even inertial. For us to have a meaningful opportunity to change, we must occasionally gain mental distance—of yet another sort—sufficient to reflect directly on our existing ends and their continued desirability, with adequate understanding of alternatives.
Listening only to chosen speech impedes this mental distance. Because we as listeners choose speech based on who we are, that speech is unlikely either to prompt critical self-reflection or to present an adequate range of alternatives for whom we could become. It may even passively reenforce our present ends and hinder our evolution. By contrast, consuming speech that is unchosen and hence more unfamiliar or even genuinely surprising is likelier to enable self-transformation.
A society that lacks such dynamism among its members, too, seems destined to remain trapped in outdated ways of life. As its membership grows and shifts, and the world changes, it may not evolve—culturally, politically, or legally—to meet new needs. As a staunch advocate of “[a] constitutional regime fearful of political entrenchment and dedicated to continual adaptation,” Justice Oliver Wendell Holmes saw protection of specifically “the expression of opinions that we loathe” as indispensable.
E. Tolerance and Persuadability
In a pluralistic society, we must cooperate across irresolvable differences and disagreements. We must work with and take into account the interests of those with values, beliefs, and tastes significantly different from our own (let us call them our counterparts). Sometimes we must reach compromises with our counterparts, even at partial expense of our own ends. This demands tolerance toward our counterparts, that is, engaging and respecting them as rational agents and persons as we engage with them. We must, so far as possible, attempt to understand and make sense of their claims, identify issues of agreement, and persuade them on issues of disagreement.
Highly personalized listening can be expected to decrease tolerance. Humans have a well-documented proclivity to exhibit negative personal attitudes toward those who hold views—especially moral and political views—with which they strongly disagree. We may also interpret the speech of our counterparts in uncharitable ways, thereby missing any merit in it. But these effects can be countered, the more often we hear counterparts speak. The more we so listen, the likelier we become to succeed in the social cooperative tasks mentioned above, and to empathize with our counterparts.
Tolerance is useful within any social group, but its usefulness is heightened within a democracy. For in a democracy, we exercise coercive power mutually over each other. As Jonathan Gingerich has argued, all participants in a democratic system, and especially potential losers, must have a genuine opportunity to persuade their fellow citizens on political issues. This logically entails that citizens must open themselves, even ever so slightly, to the possibility of being persuaded, in light of their own fallibility and the basic rationality of other citizens—at least enough to actively listen. For any opportunity for a speaker to persuade, however slight, dies if all listeners tune out.
One might go further. Elected officials within a democracy owe a duty to their constituents to hear out their political concerns, no matter how much the officials would prefer not to. One might by extension argue that voters, as joint rulers over one another, have an obligation to hear out the political concerns of at least some of their counterparts. As Justice Louis Brandeis once declared, “public discussion is a political duty.”
II. Limitations and Implications
Having made the general case for unchosen listening, I offer a few brief remarks to situate it within contemporary free speech doctrine.
Hopefully it is clear that I do not mean to reject listener choice as a critical First Amendment concept. Listener choice is rightly viewed by courts as definitive in certain limited domains, such as the home. Arguably courts should, even most of the time, defer to the listener’s choice, as the least objectionable, content-neutral proxy for their interests. After all, listeners’ choices usually converge with their personal interests. Respecting listeners’ choices also tends, through market mechanisms, to generate decently diverse media options, which are good for their democratic interests.
As mentioned at the outset, I am also inclined to believe that the First Amendment does not protect a right to coerce listeners. The final scene of A Clockwork Orange is not a good look for the Constitution. Some speech, such as hundreds of creepy Facebook messages, may be constructively coercive, because of its repetition, intimidation, or substantial disruption of normal activities. Other speech may be coercive because of time-constrained psychological vulnerabilities of a listener. But listening is not coerced simply because unrequested, unaccepted, or even undesired. I am inclined to believe that even speech that ignores a listener’s affirmative assertion of unwillingness is not coerced, absent such exacerbating factors. Speech is not sex. Nor is listening coerced if the asserter had some minimal exposure to the speech on a prior occasion—the glimpse before we “avert our eyes” or click to delete.
One of the main venues for speech of this sort is the public forum. Of course, speech in these spaces works better—to spark correction, creativity, transformation, or tolerance—when the listener is game. Indeed, the more actively a listener resists, the less likely speech will achieve any such benefits. But even if most listeners avert their eyes against unwanted speech in public, they at least gain minimal exposure to the message and the fact that someone wants to speak it. Even that regular exposure can heighten tolerance and reduce resistance toward other, perhaps more articulate, unchosen speech.
Virtual forums can prevent even such minimal exposure to speech. With digital listener’s-choice technology, you need not avert your eyes whenever you see an unwelcome message, but can often effectively swipe right to pre-screen all similar messages. Indeed, some messages can be silenced not because you affirmatively resisted them, but because messages you liked better were prioritized over them. Yet media platforms may increasingly turn to listener’s choice as a means of filtering principle that both satisfies users and evades responsibility.
Aggressive use of listener-chosen filtering may even portend the demise of the traditional public forum. The area of publicly owned space available for talking appears to have been shrinking for decades, as the Court has recognized. Listeners, willing and unwilling, have fled those spaces even faster since the digital era. Research shows that pedestrians now walk faster, and interact less often, in public spaces. Many factors seem to exacerbate this trend: ever more densely developed urban areas, constrained public budgets, perceived rises in crime, architectural features designed to make spaces less welcoming (likely targeted at the homeless), and increasing stretches of everyday life spent online. When we talk to strangers, it is thus mostly in private, digitally moderated spaces, to the willing. Even when we do find ourselves conversing in physical spaces, severe geographical political segregation means that our interlocutors will often still be likeminded. All of this stymies the serious engagement across difference that is critical to the democratic process.
Admittedly, the public forums of old were never ideal sites for reasoned discourse. Soapbox speakers are typically better at stirring passions than parsing arguments. Yet the solution should not be doctrinal abandonment of the public forum, and the commitment to unchosen listening that it represents. Rather, courts should stand ready to approve the expansion of spaces in which we encounter at least some unbidden speech—if necessary, privately owned ones—and encourage reason-giving within them. That would be the First Amendment paradise.
Unchosen Listening
INTRODUCTION
A century of developments in communications technology has done wonders for listeners.1By “listeners,” I will mean consumers of speech in all forms. In the not-so-distant past, the listener had few choices regarding the speech they heard. An urban listener might find a range of choice in libraries or bookstores, or among private associations’ meetings or periodicals. But mostly listeners heard what others happened to say, on the job, at church, in the neighborhood, and on the street-corner. Today, listeners have more choices than ever, and more ability to choose which speech to hear and which not to hear.
First Amendment doctrine, for its part, has occasionally shielded these listeners’ choices about whom to listen to, even over speakers’ choices about whom to speak to.2E.g., Hill v. Colo., 530 U.S. 703, 716 (2000); Rowan v. Post Office Dept., 397 U.S. 728, 738 (1970); Frisby v. Schultz, 487 U.S. 474, 477–78 (1988). I largely agree with the authors in this symposium who argue that it should do so more often.3See generally e.g., Ashutosh Bhagwat, Respecting Listeners’ Autonomy: The Right to Be Left Alone, 98 S. Cal. L. Rev. 1129 (2025); James Grimmelmann, Listeners’ Choices Online, 98 S. Cal. L. Rev. 1231 (2025); Helen Norton, Fear and Free Speech, 98 S. Cal. L. Rev. 1351 (2025). The appeal is evident. Communication goes most smoothly with a willing speaker and a willing listener. And with listeners now awash in speech options, someone must choose among them. The listener seems obviously superior, as chooser, to many looming alternatives: the stalker, the fake-news purveyor, the oligopolistic media platform, or—most antithetical to the First Amendment—the government.
Yet this Commentary urges caution in so fully embracing listener’s choice that it becomes a paradigm of First Amendment jurisprudence. One can easily move from recognizing the advantages of willing listening (and speaking) to identifying the “core” of constitutionally protected speech as not just a “joint activity”—to use Ashutosh Bhagwat’s illuminating term4Bhagwat, supra note 3, at 1135—but mutually consensual, affirmatively chosen by both speaker and listener.5See Grimmelmann, supra note 3, at 1281–82. Grimmelmann also has thoughtfully discussed the limits of listener choice, given the risks of irrational and uninformed choice, as well as the importance of other values. See James Grimmelmann, Listeners’ Choices, 90 U. Colo. L. Rev. 365, 372–73 (2019). One might even begin seeing something constitutionally suspect about speech that lacks a consenting listener. On this interpretation, the fact that the speaker on his soapbox in Washington Square Park annoys some passersby is, while inevitable, still unfortunate—a sacrifice made for the sake of the other, more willing passersby.6See Bhagwat, supra note 3, at 1143–44; Grimmelmann, supra note 3, at 1233. To be clear, Bhagwat explicitly acknowledges the need to maintain the diversity and friction of public discourse. Bhagwat, supra note 3, at 1143. I just believe the importance of that need is more cleanly understood without too much emphasis on the willingness of listeners in general.
Yet in a not-too-distant future, this fact might be “correctible” by technology, assisted by artificial intelligence and mass data collection. Every webpage you visit might seamlessly filter out any content you have elected not to see; or software might elect for you, based on your eyes’ patterns of lingering. When you step onto the street, you might wear digital glasses to blur out all written speech that you have deemed unwelcome, and digital headphones to cancel all noise except oral speech you have deemed welcome.7Thanks to Jeremy Gartland and Eugene Volokh for inspiring this thought experiment.
While this seems a paradise for listeners’ choice, it would not be one for listeners’ interests, another traditional First Amendment concern.8See, e.g., Citizens United v. F.E.C., 558 U.S. 310, 354 (2010); Red Lion Broad. Co. v. F.C.C., 395 U.S. 367, 390 (1969). The speech that we choose to hear does not always make us better off.9I assume that a listener’s interests are not inevitably reflected in their choices. The former can come apart from the latter, either because the listener’s interests are objective in some sense, or else because they are subjective but higher-order or long-term such that they do not determine every choice. Certainly, a listener’s choice is one aspect of their interests. And speech they vehemently reject—such as harassing speech—will rarely be in their interest. But a gray zone exists between chosen listening and coerced listening, which I will call unchosen listening. Some unchosen listening, I will argue, is a desirable part of being an autonomous person and citizen.
I will also argue that it is desirable for the broader society of which listeners are a part—including its collective knowledge, culture, and, especially, democratic institutions. These societal interests underlie the First Amendment, too, as Alan Chen also discusses in this symposium.10Alan K. Chen, Pluralism and Listeners’ Choices Online, 98 S. Cal. L. Rev. 1387, 1395 (2025). Perhaps for these reasons, unchosen listening has been revered in First Amendment doctrine. As the Supreme Court declared in Terminiello v. Chicago, speech “may indeed best serve its high purpose when it induces a condition of unrest, creates dissatisfaction with conditions as they are, or even stirs people to anger. Speech is often provocative and challenging. It may strike at prejudices and preconceptions and have profound unsettling effects as it presses for acceptance of an idea.”11Terminiello v. Chicago, 337 U.S. 1, 4 (1949).
Indeed, the most fiercely protected First Amendment speech—perhaps its real “core” case—involves non-consensual listeners: speech in the public forum.12I mean by “core” only, as stated, the most stringently protected; various theories of the First Amendment might interpret the theoretical core—the cases best serving First Amendment values—differently. This means citizens trying to move or persuade fellow citizens, and especially those none too thrilled to listen. It is—for those familiar with the First Amendment landmarks—Abrams’s communist pamphlets thrown from the roof, Kovacs’s pro-labor commentary amplified from his truck, Mosley’s sign decrying race discrimination, Cohen’s jacket protesting the draft.13Abrams v. United States, 250 U.S. 616, 630 (1919); Kovacs v.Cooper, 336 U.S. 77, 79 (1949); Police Dept. of Chi. v. Mosley, 408 U.S. 92, 93 (1972); Cohen v. California, 403 U.S. 15, 16 (1971). And it is arguably the only speech government has a (currently recognized) positive obligation to protect—by preserving the openness of traditional public forums—under the First Amendment.14See Hague v. Comm. Indus. Org., 307 U.S. 496, 513, 515–16 (1939).
While the Court has seldom expounded on the theme of unchosen listening, its silence invites less doctrinally bound theorizing. This Commentary thus offers a brief normative case for resisting a listener’s-choice paradigm (and maintaining more of a public forum paradigm).15I take my case to be congenial and supplementary to Alan Chen’s remarks on a similar theme in this symposium. See generally Chen, supra note 10.
I. The Case for unchosen listening
What speech will a listener hear, if they mostly choose that speech themselves or have it selected for them based on algorithmic predictions of their preferences?16 Currently, these predictions often track not our choices about content but predictions of what content we will click through to––based on past listening data––or spend time listening to. I largely set aside here the enormous discrepancies in preferences that might result. For a discussion in the music context, see Liz Pelly, Mood Machine: The Rise of Spotify and the Costs of the Perfect Playlist 103 (2025) (describing how music streaming services like Spotify do not actually satisfy listeners’ preferences, but rather those of their “slightly off data double”). I also set aside the potentially grave issue of how listeners’ preferences might themselves be manipulated by media or other corporations. Chances are, it will be fairly insular and homogeneous. That is, it will be familiar, comfortable, and unchallenging; and it will come from like-minded speakers and a small number of sources.17See, e.g., Florian Arendt, Temple Northup & Lindita Camaj, Selective Exposure and News Media Brands: Implicit and Explicit Attitudes as Predictors of News Choice, 22 Media Psych. 526, 540 (2019) (finding that consumers’ positive or negative attitudes toward news media brands predict their content choice decisions); Eytan Bakshy, Solomon Messing & Lada Adamic, Exposure to Ideologically Diverse News and Opinion on Facebook, 348 Science 1130, 1131–32 (2015). Its insularity and homogeneity will be reinforced to the extent that the listener chooses with greater granularity—that is, chooses not just among sources or topics but among specific speech from each source or specific views on each topic.18Thanks to Joseph Blocher for alerting me to potential problems of more granular choosing. The result might resemble “The Daily Me” envisioned by Cass Sunstein.19Cass R. Sunstein, #Republic: Divided Democracy in the Age of Social Media 1 (2018) (citing Nicholas Negroponte, Being Digital 153 (1995)). While not all listeners will choose like this, and few will do so all the time, enough listeners will likely do so in enough cases to affect the overall patterns of communication in society.
Assuming this is true, the following discussion identifies five significant interests that can be undermined when listening becomes too tailored to a listener’s choices: preference optimization, error avoidance, innovation, self-evolution, and tolerance.20A full theory, and more space than I have, would be needed to identify and adjudicate the range of conflicts that arise between listeners’ interests and choices, as well as to determine when listener’s choosing becomes over-tailored. Each interest has an individual dimension: for listeners themselves. Each also has a collective dimension: for a society in which the total pool of speech consumed is determined largely by listeners’ choices.
A. Optimizing Preferences
Even listener-chosen speech does not always satisfy listener preferences.21Cf. Grimmelmann, Listeners’ Choices, supra note 5, at 1237 (echoing similar concerns that some listeners may make mistakes about their own preferences). Speech often defies expectations. Sometimes we have a “meh” reaction to a movie by our favorite director, or are spellbound by a movie we expected to bore. Our expectations are sometimes even insufficient for choice. I may have almost no sense of whether I would like a movie from an unfamiliar genre, because I have not experienced anything like it before. After all, the indicators of content that are available prior to hearing speech—for example, its speaker, venue, general subject, or reviews—are necessarily limited.
With ever more personalized listening technology, our mistaken choices—or mistaken predictions of our choices by algorithms—can lock us into inferior patterns of listening. Mistaken choices are easily corrected when they produce dissatisfaction, but less so when they produce merely suboptimal satisfaction. Say that I have watched and liked one Fellini film and one Spielberg film. Perhaps I am directed, by choice or algorithm, toward more Spielberg films rather than Fellini films. I continue to like Spielberg films and hence grow my love of Spielberg. I may never know the counterfactual––whether I would have preferred the Fellini films. The best chance of exiting a suboptimal equilibrium is exposure to unchosen speech, or at least less-precisely-chosen speech. Optimizing, therefore, involves reflective equilibrium between chosen and unchosen exposure.
A collective cultural danger lurks here, too, at least insofar as listeners rely on algorithmic services to make these finer speech selections for them.22More strikingly, sometimes we do not actively want to choose at all, but just “lean back and let Spotify choose things.” Pelly, supra note 16, at 25; see also id. (explaining how Spotify started to “optimize . . . for a less engaged user” experience). Because these services operate within a market economy, their corporate makers may face incentives to nudge listeners toward more popular, hence more easily satisfiable, preferences. Forced to rely on the same services to find listeners, speakers—including musicians, writers, filmmakers, journalists, and others—may also feel pressure to create content to win the algorithm’s game, as many have already.23Id. at 33 (describing music streaming as a system that is “not only designed by the major labels but also prioritizes the type of music engineered for and roundtabled for mass-scale success”); see also id. at 115 (“[M]usic that sounds like other music is the most data-blessed.”). The result may be a blander culture and public sphere, with fewer options over which to optimize our preferences.
B. Avoiding Error
Listening primarily to chosen speech may proliferate error. The first problem is that it may skew our available evidence. If, as assumed, we choose to listen to speakers who share our beliefs, including false ones, then they will be unlikely to offer evidence against those beliefs.24See, e.g., Jieun Shin & Kjerstin Thorson, Partisan Selective Sharing: The Biased Diffusion of Fact-Checking Messages on Social Media, 67 J. Commc’n 233, 247–50 (2017). They may even pass on new false beliefs based on old ones. These harms are well-documented within “epistemic bubbles,” that is, discussion spaces in which certain views are systematically excluded.25C. Thi Nguyen, Echo Chambers and Epistemic Bubbles, 17 Episteme 141, 143–44 (2020) (citing sources); Elizabeth Anderson, Political Epistemology 11 (2021). I do not dispute political science findings that echo chambers, those that tend to magnify and insulate extreme political views and consult limited media sources, may not currently be common. Cf. Amy Ross Arguedas, Craig T. Robertson, Richard Fletcher, Rasmus Kleis Nielsen, Echo Chambers, Filter Bubbles, and Polarisation: A Literature Review, Reuters Institute Report, 2022, at 4 (surveying recent literature); Damian Trilling & Claus Schoenbach, Challenging Selective Exposure, 3 Digital Journalism 140, 141 (2014) (explaining how current structural factors and the existence of mainstream news outlets at least tend to give most citizens decently diverse news diets, even if contrary to their preferences).
A second problem is that too much chosen listening may objectionably narrow the range of subjects on which we possess evidence. People may choose to listen to little or no speech concerning decisions they will or must take, including voting. For instance, research suggests that the most promising ways to inform voters may involve incorporating speech that they do not choose alongside speech they do—such as entertainment.26See Markus Prior, Post-Broadcast Democracy: How Media Choice Increases Inequality in Political Involvement and Polarizes Elections 17 (2007).
Yet a third problem is that a steady diet of chosen speech may dull our capacity to think itself. As John Stuart Mill observed 166 years ago, reasoning is better stimulated and optimized by speech that is uncomfortable and challenging.27John Stuart Mill, on Liberty (Rethinking the Western Tradition) 90–91 (David Bromwich & George Kateb eds., Yale Univ. Press 2003) (1859). But chosen speech is, by assumption, typically the opposite. Chosen speech can exacerbate self-favoring and in-group-favoring cognitive biases.28See, e.g., Anderson, supra note 25, at 13. Chosen speech may not even offer reasons, because it does not seek to persuade. Chosen speech can, by repeating or just assuming our beliefs, reinforce and bury them so deeply that we have trouble getting the “mental distance” necessary to scrutinize them—to question them and their justification.
C. Innovation
Unchosen listening can stimulate not just rational thought, but more creative and generative forms of thought that do not merely analyze evidence but propose new ideas or solutions.
Creativity is by its nature unexpected. It often involves making connections among facts, ideas, disciplines, cultures, and so forth that others—including oneself—had never thought to draw. We are thus sometimes likelier to arrive at an innovative solution not by continuing to think directly about the problem—constrained by our previous expectations—but by allowing our mind to wander and randomly associate.29See, e.g., Jonathan Gingerich, Is Spotify Bad for Democracy?: Artificial Intelligence, Cultural Democracy, and Law, 24 Yale J.L. & Tech. 227, 262–63 (2022). The revolutionary inventions, discoveries, insights, and movements that have been prompted by contact with unusual contexts, other disciplines, or other cultures are countless. Consider an eclectic sampling. Clinical psychology has been seriously influenced by Buddhist religious practices. Impressionism in painting was largely inspired by Japanese ukiyo-e prints30Colta Feller Ives, Japonisme, Metro. Museum of Art (Oct. 1 2004), https://www.metmuseum.org/essays/japonisme [https://perma.cc/EZP2-D6YX].; and the painting of Jackson Pollock, darling of the Supreme Court’s own First Amendment cases, was likely inspired by Native American sandpainting.31Jackson Pollock, One: Number 31, 1950, Museum of Mod. Art, https://www.moma. org/collection/works/78386 [https://web.archive.org/web/20250114000836/https://www.moma.org/collection/works/78386]. The QR code was suggested by the Go board,32Justin McCurry, ‘I’m Pleased It Is Being Used for People’s Safety’: QR Code Inventor Relishes Its Role in Tackling Covid, The Guardian (Dec. 10, 2020), https://www.theguardian. com/technology/2020/dec/11/qr-code-inventor-relish-its-role-in-tackling-covid [https://perma.cc/G6AS-3NMG]. while the foldable shape of heart stents was suggested by Japanese origami.33Wei Zhao, Nan Li, Liwu Liu, Jinsong Leng & Yangju Liu, Origami Derived Self-assembly Stents Fabricated via 4D Printing, 293 Composite Structures 1, 1 (2022). Numerous engineering innovations, from aeronautics to robotics, have come from observing animals.34See Sandy B. Primrose, Biomimetics: Nature-inspired Design and Innovation 9, 81 (2020) (describing, among many examples, innovations in bullet trains from observing birds, and in robotics from observing gripping mechanisms on gecko feet). The latest monumental computer science invention, large language models, grew from insights in both linguistics and cognitive science. And this symposium’s own Aziz Huq uses observations in sociology for thinking about digital speech.35Aziz Huq, Islands of Algorithmic Integrity: Imagining a Democratic Digital Public Sphere, 98 S. Cal. L. Rev. 1287, 1289 (2025).
Creativity thus seems likeliest to be prompted by exposure to speech that is unexpected, or at least unguided by our own choices.36See, e.g., Charlan Jeanne Nemeth & Margaret Ormiston, Creative Idea Generation: Harmony Versus Stimulation, 37 Euro. J. Soc. Pysch. 524, 532–33 (2007). Unchosen speech can offer new concepts and frameworks of thinking that supplement or integrate with existing ones. Or it can simply interrupt habitual frameworks—generating another (creative) form of mental distance—and thus open the mind to the potential relevance of unexpected inputs. By contrast, chosen speech can lock in certain expectations of relevance.
One might expect collective, rather than individual, innovation to be fueled through highly personalized listening. It might segregate most listeners into speech subcultures, within which they could refine their own insular ideas—like members of a species separated among islands and continuing to evolve. But then a few innovators sampling from those radically diverse subcultures might produce innovations of even greater enormity. Yes, it seems just as plausible that, so long as enough listening is chosen, occasional sampling by a greater number and diversity of listeners would provide offsetting benefits of more sophisticated—if slightly less radically distinct—subcultures and more frequent innovation.37See generally Tyler Cowen, Creative Destruction: How Globalization Is Changing the World’s Cultures (2004) (arguing that globalization typically increases diversity and innovation within each society, even if cultures grow closer together).
D. Self-Evolution
Persons arguably have a fundamental autonomy interest in being able not just to make up their minds, but to change them, too. This ability to change applies to their beliefs, as described above, but also their ends—their important goals, values, commitments, and so forth—and their tastes.38See, e.g., John Rawls, Political Liberalism 186 (1993) (assuming, as part of citizens’ “moral powers,” that they “can regulate and revise their ends and preferences”); Gerald Dworkin, The Nature of Autonomy, 2 Nordic J. Stud. Educ. Pol’y 7, 12 (2015) (“Autonomy should have some relationship to the ability of individuals, not only to scrutinize critically their first-order motivations but also to change them if they so desire.”). I do not mean the mere freedom to zig when one previously zagged. Certain factors beyond our control inevitably change over time: our external circumstances vary, and our set of experiences and beliefs expands. In response, it is often appropriate or even necessary to change ourselves in order to cohere, adapt, and grow. Otherwise, we become frozen into a course determined not by us, but by the “dead hand” of our past self.39Joel Feinberg, The Concept of Autonomy, in The Inner Citadel: Essays on Individual Autonomy 34 (John Christman ed., 1989) (“Always the self that contributes to the making of the newer self is the product of both of outside influences and an earlier self that was not quite as fully formed.”); see also Gingerich, Spotify, supra note 29, at 276 (2022); Jonathan Gingerich, Spontaneous Freedom, 133 Ethics 38, 42 (2022). Indeed, one might say that autonomous self-development just is the continual evolution of our self in response to new, external inputs.
It is almost always possible to change our ends, at least insofar as nothing physically prevents us. But pursuing our ends is often habitual, or even inertial. For us to have a meaningful opportunity to change,40The freedom to change cannot mean that one simply could change. That would be either impossible or true in nearly every circumstance, depending on one’s metaphysics. We anyway do not fully choose our ends. Rather, a meaningful opportunity is all that we can hope for. we must occasionally gain mental distance—of yet another sort—sufficient to reflect directly on our existing ends and their continued desirability, with adequate understanding of alternatives.41Cf. Gingerich, Spotify, supra note 29, at 277.
Listening only to chosen speech impedes this mental distance. Because we as listeners choose speech based on who we are, that speech is unlikely either to prompt critical self-reflection or to present an adequate range of alternatives for whom we could become. It may even passively reenforce our present ends and hinder our evolution.42Id. at 276; see also Gingerich, Spontaneous Freedom, supra note 39 , at 42. By contrast, consuming speech that is unchosen and hence more unfamiliar or even genuinely surprising is likelier to enable self-transformation.
A society that lacks such dynamism among its members, too, seems destined to remain trapped in outdated ways of life. As its membership grows and shifts, and the world changes, it may not evolve—culturally, politically, or legally—to meet new needs. As a staunch advocate of “[a] constitutional regime fearful of political entrenchment and dedicated to continual adaptation,”43incent Blasi, Holmes and the Marketplace of Ideas, 2004 Sup. Ct. Rev. 1, 45 (2004). Justice Oliver Wendell Holmes saw protection of specifically “the expression of opinions that we loathe” as indispensable.44Abrams v. United States, 250 U.S. 616, 630 (1919) (Holmes, J., dissenting).
E. Tolerance and Persuadability
In a pluralistic society, we must cooperate across irresolvable differences and disagreements. We must work with and take into account the interests of those with values, beliefs, and tastes significantly different from our own (let us call them our counterparts). Sometimes we must reach compromises with our counterparts, even at partial expense of our own ends. This demands tolerance toward our counterparts, that is, engaging and respecting them as rational agents and persons as we engage with them. We must, so far as possible, attempt to understand and make sense of their claims, identify issues of agreement, and persuade them on issues of disagreement.
Highly personalized listening can be expected to decrease tolerance. Humans have a well-documented proclivity to exhibit negative personal attitudes toward those who hold views—especially moral and political views—with which they strongly disagree.45See, e.g., Cass R. Sunstein, Going to Extremes: How Like Minds United and Divide, 1–20 (2009) (detailing tendencies within likeminded groups toward both more extremism and more negative perceptions toward outsiders); Lee C. Bollinger, The Tolerant Society 111, 120 (1986). We may also interpret the speech of our counterparts in uncharitable ways, thereby missing any merit in it.46See, e.g., Julia A. Minson, Frances S. Chen & Catherine H. Tinsley, Why Won’t You Listen to Me? Measuring Receptiveness to Opposing Views, 66 Mgmt. Sci. 2801 (2019). But these effects can be countered, the more often we hear counterparts speak.47See, e.g., Diana C. Mutz, Cross-Cutting Social Networks: Testing Democratic Theory in Practice, 96 Amer. Pol. Sci. Rev. 111, 122–23 (2002). The more we so listen, the likelier we become to succeed in the social cooperative tasks mentioned above, and to empathize with our counterparts.
Tolerance is useful within any social group, but its usefulness is heightened within a democracy. For in a democracy, we exercise coercive power mutually over each other. As Jonathan Gingerich has argued, all participants in a democratic system, and especially potential losers, must have a genuine opportunity to persuade their fellow citizens on political issues.48Gingerich, Spotify, supra note 29, at 264; see also Robert Post, Religion and Freedom of Speech: Portraits of Muhammad, 14 Constellations 72, 75–75 (2007). Gingerich focuses on the importance of being able to persuade fellow citizens on cultural issues; but the point holds even if one centers political issues. See also Chen, supra note 10, at 105–06, 108–11. This logically entails that citizens must open themselves, even ever so slightly, to the possibility of being persuaded49For an enlightening discussion of persuadability as a personal virtue, see generally Joseph Blocher, “The Road I can’t Help Travelling”: Holmes on Truth and Persuadability, 51 Seton Hall L. Rev. 105 (2020)., in light of their own fallibility and the basic rationality of other citizens—at least enough to actively listen.50Teresa Bejan has cogently criticized public speakers today for talking to ourselves (those who agree with us), and thereby primarily seeking not to persuade but to acquire and bestow recognition. Teresa Bejan, A People’s History of Free Speech, Persuasion (Oct. 9, 2024), http://www.persuasion.community/p/a-peoples-history-of-free-speech [https://perma.cc/DD2Q-PRDW]. For any opportunity for a speaker to persuade, however slight, dies if all listeners tune out.51Cf. Chen, supra note 10, at 108, 111.
One might go further. Elected officials within a democracy owe a duty to their constituents to hear out their political concerns, no matter how much the officials would prefer not to. One might by extension argue that voters, as joint rulers over one another, have an obligation to hear out the political concerns of at least some of their counterparts.52Of course, choosing across many axes of preference may incidentally expose a listener to speech that is unchosen, and mitigate some of these harms. The socialist may encounter anti-socialist speech on the Yankees-fans’ or parents’ forum. However, insofar as viewpoints tend to cluster, and conversations tend to be more focused online than in-person, the mitigating effects may be limited. As Justice Louis Brandeis once declared, “public discussion is a political duty.”53Whitney v. California, 274 U.S. 357, 375 (1927) (Brandeis, J., concurring).
II. Limitations and Implications
Having made the general case for unchosen listening, I offer a few brief remarks to situate it within contemporary free speech doctrine.
Hopefully it is clear that I do not mean to reject listener choice as a critical First Amendment concept. Listener choice is rightly viewed by courts as definitive in certain limited domains, such as the home.54Frisby v. Schultz, 487 U.S. 474, 484–85 (1988). The doctrine largely limits those domains to one, the home, see Bhagwat, and it may do so rightly. The home is a place of private intimacy, where one can retreat in order to avoid the social cooperative pressures that pervade our lives in so many other spaces. Arguably courts should, even most of the time, defer to the listener’s choice, as the least objectionable, content-neutral proxy for their interests. After all, listeners’ choices usually converge with their personal interests. Respecting listeners’ choices also tends, through market mechanisms, to generate decently diverse media options, which are good for their democratic interests.55Grimmelmann is thus right that listener choice should be a significant factor guiding regulation of some intermediaries. Grimmelmann, supra note 3, at 1231–32. He also elegantly weighs the diverse interests of speakers, listeners, and intermediaries.
As mentioned at the outset, I am also inclined to believe that the First Amendment does not protect a right to coerce listeners.56See Bhagwat, supra note 3, at 1157–58; Norton, supra note 3, at 1366. The final scene of A Clockwork Orange is not a good look for the Constitution. Some speech, such as hundreds of creepy Facebook messages,57Counterman v. Colorado, 600 U.S. 66, 70 (2023). See also Norton, supra note 3, at 1365, 1367. may be constructively coercive, because of its repetition, intimidation, or substantial disruption of normal activities. Other speech may be coercive because of time-constrained psychological vulnerabilities of a listener.58See generally, e.g., McCullen v. Coakley, 573 U.S. 464 (2014) (abortion clinic patients); Snyder v. Phelps, 562 U.S. 443 (2011) (funeral attendees); Ohralik v. Ohio State Bar Ass’n, 436 US 447 (1978) (accident victims). The first two cases upheld free speech claims against listeners, perhaps rightly, but all three involve listeners who might be considered (temporarily) psychologically vulnerable. But listening is not coerced simply because unrequested, unaccepted, or even undesired. I am inclined to believe that even speech that ignores a listener’s affirmative assertion of unwillingness is not coerced, absent such exacerbating factors.59I do appreciate, however, that Bhagwat contemplates limitations on assertions of the right not to listen, such as across time, and perhaps even a requirement of one-time minimal exposure. Bhagwat, supra note 3, at 1129. Speech is not sex. Nor is listening coerced if the asserter had some minimal exposure to the speech on a prior occasion—the glimpse before we “avert our eyes” or click to delete.60The power of government to compel listening, because of its power, may be uniquely limited.
One of the main venues for speech of this sort is the public forum. Of course, speech in these spaces works better—to spark correction, creativity, transformation, or tolerance—when the listener is game. Indeed, the more actively a listener resists, the less likely speech will achieve any such benefits. But even if most listeners avert their eyes against unwanted speech in public, they at least gain minimal exposure to the message and the fact that someone wants to speak it. Even that regular exposure can heighten tolerance and reduce resistance toward other, perhaps more articulate, unchosen speech.
Virtual forums can prevent even such minimal exposure to speech. With digital listener’s-choice technology, you need not avert your eyes whenever you see an unwelcome message, but can often effectively swipe right to pre-screen all similar messages. Indeed, some messages can be silenced not because you affirmatively resisted them, but because messages you liked better were prioritized over them. Yet media platforms may increasingly turn to listener’s choice as a means of filtering principle that both satisfies users and evades responsibility.
Aggressive use of listener-chosen filtering may even portend the demise of the traditional public forum. The area of publicly owned space available for talking appears to have been shrinking for decades, as the Court has recognized.61See Food Emps. v. Logan Valley Plaza, Inc., 391 U.S. 308, 324 (1968) (describing movement from urban to suburban spaces and hence the growth of shopping malls as places of public congregation); see also Sarah Schindler, The Publicization of Private Space, 103 Iowa L. Rev. 1093, 1106 (2018) (“Fewer cities are investing in the direct creation of new publicly owned public space, and there has been an increase in privatized public space.”). Listeners, willing and unwilling, have fled those spaces even faster since the digital era. Research shows that pedestrians now walk faster, and interact less often, in public spaces.62Arianna Salazar-Miranda, Zhuangyuan Fan, Michael B. Baick, Keith N. Hampton, Fabio Duarte, Becky P.Y. Loo, Edward L. Glaeser & Carlo Ratti, Shifting Patterns of Social Interaction: Exploring the Social Life of Urban Spaces Through A.I. (Nat’l Bureau Econ. Rsch., Working Paper No. 33185, 2024) (analyzing pedestrian behavior using video evidence from 1979–2010). Many factors seem to exacerbate this trend: ever more densely developed urban areas,63Zenovia Toloudi, Are We in the Midst of a Public Space Crisis?, The Conversation (Jun. 7, 2016) (describing urban planning choices that restrict access to public space). constrained public budgets,64Nina Lakhani, Millions of Americans Lack Access to Quality Parks, Report Reveals, The Guardian (May 20, 2020), https://www.theguardian.com/environment/2020/may/20/park-inequality-access-coronavirus-wellbeing [https://perma.cc/KHT4-S5JH] (describing budget constraints as one reason why 100 million Americans lack access to public open and green spaces). perceived rises in crime, architectural features designed to make spaces less welcoming (likely targeted at the homeless),65See, e.g., Faith Ruetas, Anti Homeless Architecture, Rethinking the Future, https://www.re-thinkingthefuture.com/designing-for-typologies/hostile-architecture-anti-homeless-architecture/ [https://perma.cc/XY2M-4GUP] (depicting various forms of architecture that seem designed to exclude lingering). and increasing stretches of everyday life spent online. When we talk to strangers, it is thus mostly in private, digitally moderated spaces, to the willing. Even when we do find ourselves conversing in physical spaces, severe geographical political segregation means that our interlocutors will often still be likeminded.66See, e.g., Yongjun Zhang, Siwei Cheng, Zhi Li, Wenhao Jiang, Human Mobility Patterns Are Associated with Experienced Partisan Segregation in U.S. Metropolitan Areas, (Nat’l Econ. Bureau Rsch., Working Paper, 2025) (observing partisan segregation not just across geographical regions but within residential neighborhoods of cities). All of this stymies the serious engagement across difference that is critical to the democratic process.
Admittedly, the public forums of old were never ideal sites for reasoned discourse. Soapbox speakers are typically better at stirring passions than parsing arguments. Yet the solution should not be doctrinal abandonment of the public forum, and the commitment to unchosen listening that it represents.67See Ronald J. Krotoszynski, Jr., Our Shrinking First Amendment: On the Growing Problem of Reduced Access to Public Property for Speech Activity and Some Suggestions for a Better Way Forward, 78 Ohio St. L.J. 779, 784 n.15 (2017) (collecting cases in which courts permitted the government to exclude citizens from public spaces). Rather, courts should stand ready to approve the expansion of spaces in which we encounter at least some unbidden speech—if necessary, privately owned ones68See generally Erin L. Miller, The Private Abridgment of Free Speech, 32 Wm. & Mary Bill Rts. J. 615 (2024) (arguing for the validity of First Amendment claims against certain powerful private agents, including potentially the largest social media platforms).—and encourage reason-giving within them.69See Rebecca L. Brown, Remarks on Academic Freedom and Free Speech: Reflections on Blocher, 98 S. Cal. L. Rev. 1379, 1385–86 (2025). That would be the First Amendment paradise.
98 S. Cal. L. Rev. 1399
Download
* Associate Professor of Law & Philosophy, University of Southern California Gould School of Law. I am grateful to the participants in this symposium for inspiring this comment, to the student editors of the Southern California Law Review for their diligence at every step of the production process, and to Ashutosh Bhagwat, Rebecca Brown, James Grimmelmann, and Felipe Jiménez for detailed feedback.
Like this article?