PDF Summary:Superbloom, by Nicholas Carr

An endless flood of digital messages, images, and performances is reshaping how we think, relate to others, and function as a democratic society. In Superbloom, technology critic Nicholas Carr argues that we’ve entered “hyperreality,” where simulations have replaced our direct experience of the world, with serious consequences. He writes that every communication technology from the telegraph to social media has promised connection but delivered division—and we keep falling for it because we misunderstand human nature. This shift is fueling a mental health crisis, eroding our capacity for deep thought and empathy, and fragmenting democracy itself.

This guide breaks down Carr’s argument that the digital platforms we interact with everyday exploit our psychological vulnerabilities, that top-down changes to the system will likely fail, and that individual resistance may be our only path forward. We’ll also examine whether Carr’s pessimism is justified, explore the debate over social media’s role in mental health, and connect his call for resistance to a long American tradition of dissent.

(continued)...

Perhaps most troubling, Carr contends that spending most of your time engaged with screens limits what kind of thinking you’re capable of. He explains that our ability to understand the world develops through physical interaction with it: through touch, movement, spatial navigation, and unmediated sensory experience. When we bypass this embodied learning and interact with digital representations instead, we become skilled at recognizing patterns and recombining existing ideas, but we lose the capacity to create genuinely new understanding. Like AI systems that mix and match what they’ve seen in their training data, we risk becoming derivative thinkers who can’t generate original insights.

(Shortform note: Carr’s concern reflects how AI systems actually work—and what happens when we interact with AI-generated content rather than physical reality. AI systems don’t understand the material they work with. Instead, they convert vast databases of human writing and art into numerical representations, then identify patterns in how those numbers relate to each other, and recombine these patterns to produce new outputs. This means AI-generated text and images are mathematical echoes of human thought patterns, or abstractions of the writer or artist’s original experience. Critics like Carr worry that as we increasingly consume AI-generated material, we may begin thinking in similarly pattern-based, recombinatory ways.)

Hyperreality Undermines Human Connection

Beyond reshaping how we think, hyperreality also transforms how we relate to other people, and not for the better. Carr argues that digital communication undermines human connection—ironically, by making one-way connection easier. We intuitively believe that getting to know people better makes us like them more, but research finds the opposite pattern: In studies designed to mimic the kind of one-sided information disclosure that happens as we follow people on social media, participants received varying amounts of trait information about fictional people——and the more they learned, the less they liked the person being described.

This occurs through what researchers call dissimilarity cascades: You tend to like those who seem similar to you and dislike those who appear different. Crucially, the disliking tendency proves stronger. Once you encounter a significant dissimilarity, you interpret subsequent information as further evidence of difference while similarities fade into the background. Social media, where people continuously broadcast their personal information and opinions, creates perfect conditions for these dissimilarity cascades. As you encounter endless streams of information about people’s political views, religious beliefs, parenting philosophies, and consumer preferences, the differences accumulate.

Does Learning More About People Really Make Us Like Them Less?

When he describes dissimilarity cascades, Carr refers to a 2007 study by psychologist Michael Norton, but a 2025 study failed to reproduce its results. While participants still believed they’d like someone more if they knew more about them, showing them more traits actually had no effect on how much they liked the described person, and the researchers found no evidence for the dissimilarity cascades that Norton described. They suggest that learning new facts about someone may not always change your opinion of them, particularly if you already like them.

In fact, the 2025 study found that when people initially perceived the described person as dissimilar to themselves, learning more of their traits increased both how much they liked the person and their perception of similarity—but only up to a point, after which it leveled off. This suggests more information might help overcome negative first impressions, but it doesn’t do much to change your opinion once you already like someone. The failed replication of Norton’s study doesn’t necessarily undermine Carr’s argument, but it does raise questions about whether this specific mechanism operates as reliably as he suggests.

Carr argues the damage might be contained if social media at least fostered empathy to counterbalance this dynamic. But to empathize with other people, we need to pay sustained attention to them and communicate within the close proximity needed to read another person’s physical cues—their facial expressions, tone of voice, and body language. Communication through screens lacks these elements, making it nearly impossible to develop the attention and observational skills that enable us to practice empathy for others.

(Shortform note: Psychologist Daniel Goleman argues in Focus that empathy requires attuning yourself to cues like the catch in someone’s voice that reveals they’re not really “fine” or the tension that reveals their anxiety. Communication via screens undermines this attention not only by stripping away many of these cues, thus disrupting your processing of body language and vocal cues, but also by fragmenting your attention. Sherry Turkle (Alone Together) reports that having a phone present during face-to-face conversation, even turned off and face-down, makes both people feel less connected. The phone reminds you of all the other places you could be, dividing your attention when empathy requires you to focus on the person in front of you.)

Hyperreality Threatens Democracy

The effects of hyperreality extend beyond individual psychology and personal relationships to threaten democratic governance itself. Carr argues that democracy requires some degree of shared reality—agreement on basic facts, common information sources, and the ability to distinguish truth from falsehood. These foundations crumble in hyperreality.

On digital platforms, all information competes for attention on the same terms, with algorithms delivering whatever captures attention most effectively without regard for its importance or accuracy. Serious policy discussions must compete with outrage bait and conspiracy theories. Because your mind uses repetition as a proxy for truth—what you encounter repeatedly feels familiar, and familiarity signals reliability—repeated exposure to false stories makes them feel true. The result is a fragmented information environment where different groups inhabit different realities and reach incompatible conclusions about basic questions.

The Rise of the “Post-Fact” Society

Tech journalists have been particularly well-positioned to predict the crisis Carr describes. In their 2008 book True Enough, Farhad Manjoo predicted how digital media would fragment the shared reality on which democracy depends—years before Facebook introduced its News Feed, before Twitter became a major news source, and before the term “fake news” entered our lexicon. Manjoo predicted that the internet’s core feature, its ability to let people choose their information sources, would interact dangerously with how our brains process information. Research shows we don’t weigh information rationally or objectively: We interpret new evidence through the lens of our existing beliefs, a phenomenon called “biased assimilation.”

Psychologists observe that effect at play in studies where people who see identical evidence about controversial topics consistently interpret that new evidence as supporting their pre-existing views. Interpreting the evidence in this biased way, their opinions just become more entrenched. Repetition of a claim can also create false memories of where you heard it, making it feel even more like established knowledge. This means repeated exposure to claims doesn’t just make them feel familiar: Each exposure gives us another chance to interpret the information in ways that support what we already believe.

Carr warns that artificial intelligence threatens to make this fragmentation even more severe. He notes that generative AI systems can now create deepfakes: images, videos, and audio nearly indistinguishable from authentic recordings. When you can no longer trust visual or audio evidence, you may begin doubting all information presented through any form of media. This shift would undermine democratic accountability, and whoever controls the technology for producing synthetic media will gain unprecedented power to shape our collective understanding of what’s real.

(Shortform note: Carr doesn’t mention that as deepfakes advance, our ability to detect them improves too. Deepfakes are often created using generative adversarial networks (GANs), which consist of two AI systems competing against each other. One system, the generator, tries to create convincing fake content, while the other, the discriminator, tries to spot whether each photo or video is real or fake. Through thousands of rounds, the generator gets better and better at fooling the discriminator, eventually producing fakes that look real. But the same process that makes deepfakes possible also leaves subtle traces, like inconsistencies in lighting or unnatural patterns in pixels, that both people and new tech tools can be taught to detect.)

Why We Built a System That Harms Us

The hyperreality Carr describes wasn’t an inevitable result of technological progress. It stemmed from a specific mistake we’ve made repeatedly: believing that more communication would produce greater understanding. This section explores why we keep making this mistake and how our psychology leaves us vulnerable to technologies designed to exploit our cognitive limitations.

We Believed Communication Would Unite Us

Carr shows that misplaced optimism about communication technology stretches back over 150 years. The telegraph was expected to abolish war, radio was to usher in tolerance, television was to unite all peoples, the internet would democratize information, and social media was to build a global community. This hope persisted because it seemed intuitively obvious that if we could just communicate more effectively with each other, we would understand each other better.

(Shortform note: Communication requires effort—the cognitive work of interpreting nonverbal cues, responding to unspoken feelings, and sustaining meaningful exchange—effort that new technologies help us avoid. In Bowling Alone, Robert Putnam notes that television played a role in this shift: Though TV promised to help us understand each other, it fractured social bonds by replacing communal activities with passive consumption. Digital communication has accelerated this pattern. Email, texting, and social media enable increasingly shallow interactions, devoid of the thoughtfulness, emotionality, and empathy we experience in face-to-face conversation. When communication becomes effortless, we lose the very friction that builds understanding.)

In 1922, journalist Walter Lippmann challenged the assumption that better communication leads to better understanding by questioning whether communication could ever produce unity when people perceive reality through different lenses. He argued that modern society had become too complex for individuals to grasp directly, and that we necessarily construct what he called “pseudo-environments”—simplified mental models filtered through our limited information and biases. If everyone operates from their own distorted understanding of reality, Lippmann reasoned, more communication won’t create unity—people will just talk past each other.

(Shortform note: The gap between our mental models and reality means we navigate the world with a map that doesn’t match the territory. This creates a troubling implication for democracy: If voters necessarily operate from distorted understandings of complex issues, can we ever achieve the democratic ideal of informed citizens making reasoned judgments? Lippmann didn’t believe that people are stupid—just that our abilities are finite. He argued we need simplified models because processing reality’s full complexity would paralyze us. The pseudo-environment enables us to make decisions and take action, reducing the complexity of the world to a more manageable scope.)

We Misunderstood Human Nature

Modern psychological research has vindicated Lippmann’s skepticism by identifying the specific mechanisms behind our cognitive limitations. Carr notes that we have bounded rationality—we can’t process unlimited information, so we rely on mental shortcuts to navigate complexity despite our finite attention and processing capacity. Psychologists distinguish between two modes of thinking: fast, intuitive judgments and slow, deliberate analysis. Social media’s speed and volume favor the fast mode, forcing you to rely on gut reactions instead of careful reasoning.

Carr’s argument is that these cognitive limitations aren’t bugs that better technology can fix—they’re fundamental features of human psychology that any communication system must account for. The problem is that we designed digital systems as if these limitations don’t exist, or as if exposing people to more information faster would somehow overcome them. We misunderstood ourselves, and that misunderstanding allowed us to build systems that exploit our weaknesses.

The Philosophical Debate Over Rationality

Carr draws on research in psychology to argue we must design communication systems to account for bounded human rationality. But this recommendation sits atop a deeper philosophical debate: Should we have a concept of “ideal rationality” that exists outside of our cognitive limitations, or should rationality be understood entirely in terms of what’s humanly possible? In this view, an “ideal” rationality that ignores cognitive limits fails to describe rationality at all. Just as we don’t fault someone for failing to lift a five-ton boulder, we shouldn’t fault someone for falling short of a perfect rationality that humans don’t have the cognitive capacity to achieve.

Whether we accept bounded rationality as sufficient or believe ideal rationality should be a goal determines how we evaluate Carr’s recommendations. If bounded rationality is all that matters, then his call to design systems around human cognitive limitations is (relatively) straightforward: We should optimize digital systems for what we need to make rational decisions in the real world. But if ideal rationality also matters, it might be worth asking not just “What systems work for humans as we are?” but also “What systems would help humans get closer to perfect rationality?”

Personal Resistance to Hyperreality Is Our Best Hope

Given the depth of the problem Carr sees—a technological system that was built on misplaced optimism that now exploits our psychological vulnerabilities at a massive scale—what can we do about it? This section examines why Carr believes top-down solutions like regulation and antitrust enforcement will likely fail, and why he argues that individual acts of resistance grounded in physical reality offer the most promising, if modest, path forward.

Top-Down Solutions Will Likely Fail

Many critics of social media advocate regulatory changes to fix the system’s worst harms. Some propose what they call “frictional design,” which would encourage more thoughtful behavior by deliberately reintroducing inefficiencies into platforms. This might include adding delays before new posts appear, limiting how many times messages can be forwarded, adding extra clicks to like or reply, or even banning infinite scrolls, autoplay functions, and personalized feeds. The logic is appealing: If removing friction created the problem, adding it back might solve it.

(Shortform note: Frictional design isn’t entirely new. For centuries, societies have regulated when, where, and how people can communicate publicly. Modern laws continue this tradition by prohibiting megaphones at night, requiring permits for marches, and restricting protests near certain locations. These aren’t about what you say but about how you say it. The frictional approaches Carr describes would work similarly: not by censoring online speech, but by regulating the conditions under which we communicate, just as speed bumps slow traffic without restricting who can drive or where they go.)

But Carr argues that measures to reintroduce friction will likely fail for several reasons. First, history shows that once people adapt to greater efficiency, reductions in efficiency feel intolerable. In a culture used to instant gratification, frictional design would be nearly impossible to sell to users. Second, political obstacles make effective regulation unlikely, and many people would see such interventions as government overreach. Third, complex technological systems become nearly impossible to change once they’re deeply embedded in society because changing them causes too many disruptions for too many people. Carr contends that the moment to shape the internet’s development was in the 1990s, and that moment has passed.

The GDPR as a Test Case

The European Union’s General Data Protection Regulation (GDPR), which took effect in 2018, suggests Carr may be too pessimistic about top-down solutions. The GDPR requires websites to obtain explicit user consent before collecting data, forces companies to report breaches within 72 hours, and enables users to request the deletion of their personal data. Companies that violate the regulation face fines of up to 4% of their global revenue. These are precisely the kinds of “frictional” measures Carr describes, and the effect of the law suggests that meaningful regulation is possible, but it requires sustained enforcement, which depends more on political will than technology.

Do people resist efficiency reductions? Yes and no. Survey data shows EU residents experienced more friction using the internet for daily tasks between 2017 and 2019, while non-EU residents reported easier experiences. But users in the EU didn’t abandon digital services—they simply adapted to clicking through consent forms as they became ubiquitous, even if the experience of browsing the internet became more annoying. They accepted the tradeoff: EU citizens gained legal rights over their data at the cost of interruption by privacy notices that many users click through without reading.

Do political obstacles prevent effective regulation? Partially. The regulation passed and took effect across all EU member states, but enforcement has been inconsistent. Critics note that Ireland, which regulates most US tech giants, has been slow to impose penalties, suggesting that politics and resource constraints affect how regulations are applied. Therefore, while changing regulations isn’t impossible, meaningful oversight requires ongoing commitment to enforcement, which may wane over time. When enforcement falters, users gain protections on paper without meaningful consequences for companies that violate them.

Are embedded systems impossible to change? Not quite. Major platforms adapted their systems to comply, and the regulation influenced privacy laws in jurisdictions worldwide, suggesting that large-scale regulatory change becomes feasible when a major economy commits to it. But some companies chose different paths: Hundreds of American news outlets blocked European visitors rather than comply—a news blackout that continues years later. This geographic blocking raises a question: Does protecting data privacy justify restricting access to information based on where someone lives? Some observers argue this exposes a tension between privacy protection and the free flow of information.

Carr thinks that other proposed solutions face similar limitations. Breaking up tech companies through antitrust actions might increase competition, but it won’t stop the next generation of companies from facing the same incentives and reproducing similar problems. Similarly, content moderation addresses symptoms rather than causes: The issue isn’t just harmful content but that algorithms promote whatever generates engagement, and users themselves create the demand for divisive material.

There’s also what Carr calls the media absorption effect: Even resistance to the system gets absorbed by it. When people criticize social media on social media, the system neutralizes their opposition by turning it into another form of engagement.

Why Regulation Can’t Fix What We Want

Carr argues that antitrust action, content moderation, and even resistance all fail because they assume the problem is what the system does—exercising monopoly power, hosting harmful content, or manipulating unwilling users. But the real problem may be what we want: the engagement, stimulation, and validation that hyperreality provides. For example, antitrust legislation treats hyperreality as a monopoly problem, assuming that breaking up large tech companies (like Meta, Instagram, and WhatsApp) will restore healthy competition. But this doesn’t address user behavior. The next generation of platforms would likely reproduce the same dynamics because the demand hasn’t changed.

Similarly, content moderation treats hyperreality as a harmful content problem. But a 2025 study tracking 400 million tweets over three years found that Twitter’s sustained efforts to reduce Covid-19 vaccine misinformation had the opposite effect: Vaccine misinformation increased 140%, and when Twitter removed 70,000 accounts, related communities became more active and more viral, not less. The researchers concluded that moderation failed because users actively sought divisive content, and Twitter’s network architecture let them easily route around any restrictions.

Lastly, resistance to digital platforms gets absorbed because it treats hyperreality as an imposition. But philosopher Frank Mulder argues that we learn what we want from others, and technology doesn’t just satisfy pre-existing desires—it actively shapes them. This explains why solutions addressing supply (what companies provide) can’t work when the real issue is demand (what we desire). We’re not passive victims of manipulation, but active participants who’ve become people who want what hyperreality offers.

We Need to Reconnect With Physical Reality

If society-level solutions won’t work, what’s left? Individual resistance can’t fix a fragmented democracy, an epidemic of loneliness, or the society-wide erosion of deep thinking—but Carr argues it’s the only way to preserve these capacities in yourself. Carr explains that we’re complicit in creating and maintaining hyperreality because we actively choose the simulation, and companies profit by providing it. This means the problem can’t be solved by making companies behave better. The system is too embedded, our habituation too complete, and our desires too aligned with what it provides. Society-level change would require most people to want something different—and Carr sees little evidence that’s happening.

(Shortform note: Carr’s argument that we’re complicit in maintaining hyperreality, yet must resist, echoes a much older tension in American thought about dissent. Henry David Thoreau faced this when he retreated to Walden Pond and refused to pay taxes funding slavery and the Mexican-American War. He writes in Walden that he didn’t withdraw from society—he wanted the community it afforded him—but he resisted by breaking laws and accepting the consequences. Thoreau alone couldn’t dismantle slavery or end the war, so he had to accept that his dissent was insufficient to effect systemic change while insisting it remains morally necessary because the alternative—full complicity through inaction—is worse.)

Since collective action won’t work, Carr argues each person must decide whether to accept the terms of hyperreality or to position themselves at its margins. This means choosing to engage with the physical world frequently and carefully enough that you’re reminded reality exists independent of your perceptions and preferences, and that the material world pushes back against your desires. In practical terms, this means prioritizing physical experiences over digital representations: having conversations where you look at people’s faces and read their expressions, taking walks without your phone so you notice your surroundings, reading books that require sustained attention, and allowing yourself to experience boredom and solitude.

(Shortform note: Carr’s emphasis on physical engagement aligns with Buddhist perspectives on how our thoughts mediate reality. Writer George Saunders argues that we narrate our lives from our own perspective, which creates a gap between how we think things are and how they really are. Screens worsen this problem by feeding these mental narratives constant stimulation and validation rather than interrupting them. Carr’s practices work by forcing direct encounter with physical reality: When you’re not mediating experience through a screen, you must be present with what’s actually in front of you rather than your mental story about it—a form of engagement Saunders sees as one of the highest uses of human consciousness.)

Carr doesn’t pretend that resisting hyperreality is easy. Opting out means losing connection to the social networks that organize much of modern life—you’ll miss references, feel excluded from conversations, and sometimes lose touch with what’s happening in society. Market forces, peer pressure, and your own instincts will pull you back toward hyperreality. Even if you succeed in positioning yourself at the margins, it won’t solve the broader social problems, but Carr insists it matters anyway because of what’s at stake for you as an individual. The qualities that make us most human—empathy, depth of thought, and the ability to create meaning—require us to be grounded in physical reality.

Carr concludes that you can preserve your capacity for deep, embodied thought even if society as a whole continues down a different path. The choice is whether to accept the rules embedded in systems designed to exploit your weaknesses, or to construct a life deliberately oriented toward the physical world that exists beyond the screen.

The Costs of Opting Out—and Who Bears Them

Carr insists that resistance matters because of what’s at stake, but opting out isn’t equally accessible to everyone, and the costs may fall heaviest on the most vulnerable. M. Night Shyamalan’s 2004 film The Village illustrates this problem. In the film, adults traumatized by violence create an isolated community cut off from modern society, raising the next generation in a fabricated 19th-century village to protect them from the dangers of the contemporary world. This protection comes at cost: Children die of preventable illnesses because they lack modern medicine, and those with developmental differences lack access to tools that might improve their lives.

The film illustrates a broader problem with opting out: When one person decides to disconnect, the costs of that decision may fall on those who have no choice in the matter. Some parents want to limit their children’s exposure to hyperreality, but experts argue that shielding children from social media may make them more vulnerable by keeping them from building the critical thinking skills needed to navigate those spaces safely: recognizing manipulation, understanding commercial motives, and questioning assumptions.

The alternative is to ensure that each person has the opportunity to make informed choices for themselves. Research with young children shows they can learn to distinguish between reality and the constructed nature of digital content, especially when adults help them connect what they watch to their existing knowledge of the world. Rather than complete withdrawal, this suggests the value of guided engagement—helping people of all ages develop their own strategies for managing their relationship with technology thoughtfully and critically.