Why we choose what we do not want
Mar 2026
Thought Supply
A young scholar surrounded by giants
Why do people keep using products they say they would prefer did not exist? Many young users, for instance, say they would rather live without platforms such as TikTok or Instagram. Yet they continue to use them every day. For Sunstein, this paradox illustrates what he calls a “product trap”, a class of goods that individuals consume if they exist, while collectively preferring a world in which they did not. In a new episode of Thought Supply, legal scholar and behavioral economist Cass Sunstein and development economist David Yanagizawa-Drott reflect on a question that has shaped much of Sunstein’s work: why individuals so often act against their own preferences.
The social media example serves as an entry point into a much broader conversation. Over the course of the discussion, Sunstein traces the intellectual arc of behavioral economics – from the early challenges to the rational-actor model at the University of Chicago to the policy implications of Nudge and, more recently, to the limits of artificial intelligence. The path from an outsider idea to a widely influential framework, however, was anything but straightforward.
A young scholar surrounded by giants
To understand how these ideas emerged, Sunstein takes the conversation back to the beginning. He started his academic career at the University of Chicago, where the prevailing assumption was clear: human beings are rational actors who respond predictably to incentives. Having studied literature rather than economics, he was skeptical. His celebrated colleagues – Nobel laureates and towering figures of the law-and-economics tradition – were, as he recalls, brilliant but hardly immune to the behavioral biases their models ignored: overconfidence, loss aversion, and the occasional poor decision on the tennis court.
A colleague eventually pointed him toward the work of economist Richard Thaler. The discovery proved decisive. What began as an intellectual curiosity soon turned into a close collaboration exploring the limits of human rationality and the role of law in shaping preferences. Their work culminated in Nudge, published in 2008 and soon to become a global bestseller. At the time, however, the ideas were far from universally welcomed. In early seminars at Chicago, senior scholars reacted with open skepticism. One colleague famously remarked that Sunstein’s argument reminded him of John Stuart Mill – and that what he was saying was almost as foolish as what Mill had said. In hindsight, the episode illustrates how controversial behavioral economics once appeared within the rationalist orthodoxy it would later reshape.
Libertarian paternalism and choice architecture
The two central ideas of Nudge – libertarian paternalism and choice architecture – have since entered policy vocabularies around the world. Libertarian paternalism describes interventions that steer behavior in welfare-improving directions while leaving individuals free to choose otherwise: a GPS device suggesting the fastest route, a pension plan that enrolls workers by default, or a food label highlighting nutritional information. Choice architecture – a term Sunstein traces to designer Don Norman – captures the broader insight behind such interventions: every decision environment has a structure, and that structure inevitably influences choices.
The key question, therefore, is not whether choice architecture exists. It always does. The real question is who designs it and for whose benefit. Sunstein is candid about the darker side of this insight. The same behavioral knowledge that can encourage healthier choices or help people save more for retirement can also be used to capture attention and prolong engagement. Silicon Valley firms have long drawn on behavioral insights when designing digital platforms, and the underlying tools are neutral as to purpose.
From individual biases to collective traps
Behavioral economics initially focused on how individuals make decisions. But many of the most important societal outcomes emerge from interactions among large numbers of people. This broader perspective led Sunstein to explore how individual biases can scale into collective dynamics. In his book How Change Happens, he highlights three mechanisms that often drive large-scale social change: group polarization, informational cascades, and preference falsification. Like-minded groups, through discussion, tend to move toward more extreme positions. Informational cascades occur when individuals follow earlier signals rather than relying on their own information. And preference falsification – a concept developed by economist Timur Kuran – describes situations in which people hide their true views because they believe they are in the minority.
Yanagizawa-Drott connects these ideas to his own research in Saudi Arabia. In a large field experiment, he and his co-authors discovered that Saudi men privately held far more liberal views about women working outside the home than they believed their neighbors did. When participants learned that these views were widely shared, behavior shifted rapidly. A simple information intervention dissolved an equilibrium that almost no one privately preferred. Sunstein sees the parallel immediately: when policies or social signals reveal what people actually think, seemingly stable equilibria can change surprisingly fast.
The product trap: why quitting is not enough
This is where the social media paradox comes into sharp focus. Working with economist Leonardo Bursztyn, Sunstein and colleagues have identified what they call a “product trap”: a class of goods that people will buy if they exist, but would prefer did not exist. Instagram and TikTok fit this description precisely. Users demand substantial compensation to leave these platforms for a month. Yet when asked how much they would demand to leave if everyone in their community left simultaneously, the answer reverses. They would pay to be off. Thus, the problem is not individual weakness. It is a collective action failure: exit is only attractive if others exit too.
Sunstein connects this to preference falsification. Just as Eastern Europeans kept silent about their dislike of the Soviet regime because they assumed they were alone in thinking it, young people may privately dislike social media but assume, wrongly, that everyone else loves it. Making the true distribution of sentiment visible could itself shift the equilibrium – without a ban, without a fine, and without any change in the underlying platforms. On Australia’s outright ban for under-16s, Sunstein is measured: whether a mandate is warranted depends on welfare data, and until the evidence is clear, the presumption should favor freedom.
Artificial intelligence: promise and hard limits
The conversation eventually turns to Sunstein’s most recent book, Imperfect Oracle: What AI Can and Cannot Do. In some ways, the topic grows directly out of behavioral economics. If human decision-making is systematically biased and noisy, could AI help improve it? Sunstein begins with an optimistic view. AI systems, if well designed, are largely free from the cognitive biases and “noise” – unwanted variability in judgment – that affect human decisions. In fields such as medicine and law, growing evidence shows that algorithms can sometimes outperform human experts.
But the optimism has limits. Sunstein invokes a distinction that echoes the work of economist and Nobel laureate Friedrich Hayek, who emphasized that many social outcomes depend on dispersed and unknowable information. Some questions, Sunstein argues, are fundamentally unpredictable because the necessary data do not exist in advance. Which song will become a hit, whether a revolution will erupt, or whether two people will fall in love.
He illustrates the point with a personal story. While working on Barack Obama’s presidential campaign, Sunstein accidentally sent a sharply worded internal message to the entire organization. The resulting embarrassment led a colleague to suggest that he collaborate with Samantha Power on a speech. When they met, he fell in love within minutes. No algorithm, he notes, would have predicted that chain of events.
Takeaway
The social media paradox turns out to be a window onto something much larger, namely the gap between what individuals prefer and what markets deliver when collective action fails. What connects nudge theory, group polarization, product traps, and the limits of AI is a single underlying question: how much can any outside actor, whether state, company, or algorithm, improve on what people choose for themselves? Sunstein’s answer remains empirical rather than ideological. The appropriate response depends on welfare evidence. But until the evidence is clear, the presumption, he argues, should favor freedom.
Why do people keep using products they say they would prefer did not exist? Many young users, for instance, say they would rather live without platforms such as TikTok or Instagram. Yet they continue to use them every day. For Sunstein, this paradox illustrates what he calls a “product trap”, a class of goods that individuals consume if they exist, while collectively preferring a world in which they did not. In a new episode of Thought Supply, legal scholar and behavioral economist Cass Sunstein and development economist David Yanagizawa-Drott reflect on a question that has shaped much of Sunstein’s work: why individuals so often act against their own preferences.
The social media example serves as an entry point into a much broader conversation. Over the course of the discussion, Sunstein traces the intellectual arc of behavioral economics – from the early challenges to the rational-actor model at the University of Chicago to the policy implications of Nudge and, more recently, to the limits of artificial intelligence. The path from an outsider idea to a widely influential framework, however, was anything but straightforward.
A young scholar surrounded by giants
To understand how these ideas emerged, Sunstein takes the conversation back to the beginning. He started his academic career at the University of Chicago, where the prevailing assumption was clear: human beings are rational actors who respond predictably to incentives. Having studied literature rather than economics, he was skeptical. His celebrated colleagues – Nobel laureates and towering figures of the law-and-economics tradition – were, as he recalls, brilliant but hardly immune to the behavioral biases their models ignored: overconfidence, loss aversion, and the occasional poor decision on the tennis court.
Cass R. Sunstein on Google Scholarbrowse
UBS Center Opinion
Researchers
Cass R. Sunstein is the Robert Walmsley University Professor at Harvard University and founder of the Program on Behavioral Economics and Public Policy at Harvard Law School. He is also co-founder of the Initiative on Artificial Intelligence and the Law. From 2009 to 2012 he served as Administrator of the White House Office of Information and Regulatory Affairs under President Obama and later advised Presidents Obama and Biden on issues of law and public policy. One of the world’s most influential legal scholars, he has contributed fundamentally to the understanding of behavioral economics, regulation, and democratic governance. In 2018, he received the prestigious Holberg Prize for his groundbreaking work at the intersection of law and the humanities. Among his many publications are Nudge (with Nobel laureate Richard Thaler), How Change Happens, Sludge, and The Cost-Benefit Revolution. His latest book, On Liberalism, offers a timely and powerful defense of liberalism as the foundation of freedom and self-government.
Professor of Development and Emerging Markets, Affiliated Professor at the UBS Center
David Yanagizawa-Drott received his PhD from IIES at Stockholm University in 2010. At that point, he was hired as Assistant Professor at John F. Kennedy School of Government, Harvard University. He was then promoted to Associate Professor in 2014. In 2016, he was hired as a full professor at University of Zürich. His research has shown that propaganda can cause violent conflict, studying the impact of hate media during the Rwanda Genocide. David has also examined the role of political protests in shaping policy outcomes and elections, establishing evidence that they can be highly effective in moving public opinion. In developing countries, a lot of his work focuses on the how to improve health outcomes and economic outcomes for poor households. In this line of work, for example, David implemented a randomized field experiment that showed that a simple Community Health Worker intervention in Uganda, based on a social entrepreneurship model, reduced child mortality by more than twenty percent. David is a member of several research networks, such as Poverty Action Lab (J-PAL), The Bureau for Research and Economic Analysis of Development (BREAD), European Development Research Network (EUDN) and Center for Economic Policy Research (CEPR). His work has been highlighted in various international media outlets, such as the New York Times, Washington Post, The Guardian, The Economist and various national TV news broadcasts in the U.S.
Cass R. Sunstein is the Robert Walmsley University Professor at Harvard University and founder of the Program on Behavioral Economics and Public Policy at Harvard Law School. He is also co-founder of the Initiative on Artificial Intelligence and the Law. From 2009 to 2012 he served as Administrator of the White House Office of Information and Regulatory Affairs under President Obama and later advised Presidents Obama and Biden on issues of law and public policy. One of the world’s most influential legal scholars, he has contributed fundamentally to the understanding of behavioral economics, regulation, and democratic governance. In 2018, he received the prestigious Holberg Prize for his groundbreaking work at the intersection of law and the humanities. Among his many publications are Nudge (with Nobel laureate Richard Thaler), How Change Happens, Sludge, and The Cost-Benefit Revolution. His latest book, On Liberalism, offers a timely and powerful defense of liberalism as the foundation of freedom and self-government.
Professor of Development and Emerging Markets, Affiliated Professor at the UBS Center
David Yanagizawa-Drott received his PhD from IIES at Stockholm University in 2010. At that point, he was hired as Assistant Professor at John F. Kennedy School of Government, Harvard University. He was then promoted to Associate Professor in 2014. In 2016, he was hired as a full professor at University of Zürich. His research has shown that propaganda can cause violent conflict, studying the impact of hate media during the Rwanda Genocide. David has also examined the role of political protests in shaping policy outcomes and elections, establishing evidence that they can be highly effective in moving public opinion. In developing countries, a lot of his work focuses on the how to improve health outcomes and economic outcomes for poor households. In this line of work, for example, David implemented a randomized field experiment that showed that a simple Community Health Worker intervention in Uganda, based on a social entrepreneurship model, reduced child mortality by more than twenty percent. David is a member of several research networks, such as Poverty Action Lab (J-PAL), The Bureau for Research and Economic Analysis of Development (BREAD), European Development Research Network (EUDN) and Center for Economic Policy Research (CEPR). His work has been highlighted in various international media outlets, such as the New York Times, Washington Post, The Guardian, The Economist and various national TV news broadcasts in the U.S.