Key Takeaways
Fix the internet's ownership model, not just its worst behaviors
The problem is structural. The book argues that the internet's basic infrastructure remains decentralized, but the services most people use became concentrated in platforms that collect personal data and control what users see. Their systems are optimized for engagement and profit, not for public trust or individual agency. The result, the authors say, is a kind of digital feudalism: people depend on powerful companies that hold their data and set the terms of participation.
Reform the plumbing. Privacy laws, content moderation, and antitrust action may reduce particular harms, but the authors argue that they leave the underlying data monopoly intact. Their proposed alternative is a more human-centered internet where people control access to their information and can move between competing services.
The book's strongest move is to shift attention from harmful posts to the architecture that determines which posts spread and who profits. That resembles a systems approach in public health: treating individual incidents matters, but prevention also means changing the conditions that repeatedly produce them. The distinction between internet infrastructure and dominant applications is useful because it avoids blaming the network itself. Still, structural reform is not a single cure. Different harms, including scams, harassment, and misinformation, have distinct causes, so changing data control would need to complement enforcement, safety design, and media literacy.
Information technology distributes power according to who controls it
Tools are not neutral. The book contrasts Gutenberg's printing press with the propaganda system Joseph Goebbels built in Nazi Germany. Printing helped circulate translated texts, expand literacy, and weaken the Church's control over knowledge. Radio, film, photography, and print were also used by the Nazi regime to spread hateful ideology. The same broad lesson applies to the internet: a technology's consequences depend partly on who controls access and what the system rewards.
Design shapes outcomes. The authors do not call for abandoning digital technology. They want to preserve its reach and usefulness while changing the power structure that governs information. They imagine an internet that supports connection and shared knowledge without giving a few companies covert control over people's data and attention.
The historical comparison is effective because it treats communication systems as political infrastructure, not merely as channels for content. The printing press did not automatically create enlightenment; literacy, institutions, and contest over authority helped determine its effects. That qualification strengthens the book's case: technology can widen participation, but it does not decide who benefits. The comparison also warns against assuming that decentralization alone guarantees good outcomes. A distributed network can still circulate propaganda. The relevant question is whether institutions and incentives make manipulation more difficult and constructive participation more rewarding.
Treat behavioral data as part of personhood, not merely a privacy setting
Data reflects a life. The authors argue that online records of relationships, interests, movements, and creations increasingly represent a person's identity and experience. They connect this concern to genetic information: the cost of sequencing a human genome fell from about $3 billion in 2003 toward $100, raising questions about who stores that information and who can use it. The book's central claim is that control over personal data is not just a matter of keeping secrets. It affects a person's ability to define and direct their life.
Convenience has a price. Platforms often present services as free, while gathering information that can be used to predict and influence behavior. The authors argue that users should be able to decide what is shared, with whom, and for what purpose.
Framing data as personhood makes an abstract policy problem legible: information can shape a person's opportunities and how others perceive them. It also helps explain why a privacy setting alone may be inadequate when data is copied, combined, and used to make consequential inferences. The framing needs care, however. Data is often incomplete, ambiguous, or about several people at once, so treating it as property owned by one individual cannot settle every dispute. Strong rights over access and use may be more practical than assuming every trace has a single owner.
Measure social media's cohort effects, not only individual screen time
The social setting matters. The book cites CDC data showing that from 2007 to 2021, suicides among people ages 10 to 24 rose 62 percent. It also discusses social psychologist Jonathan Haidt's distinction between an individual's direct exposure to social media and a broader cohort effect. If nearly all a teenager's friends use a platform, leaving it may mean social isolation, even if the platform is harmful overall. In that situation, one person's decision cannot solve a problem created by the whole peer group.
Think collectively. The authors argue that feed-ranking systems reward attention-grabbing content, including material that provokes anger or comparison. They therefore see youth well-being as a problem of shared digital conditions, not simply a matter of telling each family to limit screen time.
The collective-action lens explains why advice aimed at individual users can fail: a teenager may pay a social cost for opting out while the wider environment remains unchanged. This resembles coordination problems in public policy, where people can prefer a safer shared outcome but struggle to achieve it alone. The statistics are alarming, but correlation across years does not, by itself, establish that social media caused the increase. A sound response can take the risk seriously while distinguishing evidence of association, plausible mechanisms, and proof of causation.
Put your social graph under portable, permission-based control
Connections create leverage. A social graph is the evolving record of a person's relationships and interactions, from friends and follows to messages and shared activity. Platforms can use this information in combination to infer interests and target content. The book proposes the Decentralized Social Networking Protocol, or DSNP, as a new layer that would let people control a digital identity and grant selective access through a digital wallet. A user might prove they are old enough to buy a drink without revealing an address or full birth date.
Make access revocable. Under this model, people could carry contacts and other social data between compatible services, rather than rebuilding their online lives each time they switch. The protocol is intended to support privacy, interoperability, and proof that an account belongs to a unique person.
The proposal combines two ideas often discussed separately: identity verification and data minimization. A person can establish a relevant fact without exposing every detail attached to an official document, reducing unnecessary disclosure. Portability also addresses a real source of platform power: the cost of leaving when one's relationships are tied to a service. The difficult work is implementation. Wallets must be secure and usable, credentials must be trusted across institutions, and people need recovery options if they lose access. A protocol can enable user control, but it cannot guarantee it without careful governance.
Make rights, responsibilities, rewards, and rules work as one system
A four-part social contract. The authors organize their proposal around the Four Rs: rights, responsibilities, rewards, and rules. Rights protect individual agency and control over data. Responsibilities concern how people treat one another and sustain shared resources. Rewards shape what businesses and users are encouraged to do. Rules establish and enforce the boundaries of participation. The argument is that none works well alone: rights without social obligations can neglect the common good, while rules without rights can become coercive.
Trust needs practice. The chapter on responsibilities uses examples such as volunteer-run youth soccer leagues and communities that care for shared spaces. These everyday acts depend on people believing others will contribute, rather than exploiting the commons. The authors argue that online systems should likewise reward constructive participation instead of amplifying divisive behavior.
The Four Rs provide a helpful way to test whether a technology proposal is socially complete. A service may protect privacy while creating incentives for abuse, or encourage cooperation while giving moderators unchecked power. Considering rights, duties, incentives, and enforcement together makes those trade-offs visible. The framework also raises a practical question: who defines the common good when a community disagrees? Local self-governance can fit a group's needs, but it may exclude minorities or entrench dominant voices. Durable systems need both collective choice and safeguards for members with less power.
Give data owners bargaining power, not just a download button
Access is not control. Downloading a copy of one's data may be of little use if it cannot connect to other people's information or operate across services. The authors want people to set terms for access and, where appropriate, pool data with others. They compare this to farmers forming cooperatives: individuals can combine resources and negotiate more effectively without surrendering all control to a single intermediary.
Value can take many forms. In one example, Sharon Terry and her husband organized families affected by a rare genetic disorder to share medical histories and samples with researchers, who agreed to share findings. The work helped identify a gene associated with the disease. The book imagines similar opt-in arrangements for research, services, or commercial offers, with compensation that might be financial, practical, or social.
The cooperative model makes a stronger case than a simplistic promise that every person will be paid for every data point. It emphasizes choice, pooled bargaining, and benefits beyond cash. It also draws on a central problem in economics: people often lack negotiating power when information and market access are controlled by an intermediary. Yet personal data can lose value when separated from context, and collective pools require rules about representation, consent, and benefit-sharing. The design challenge is to prevent new brokers from becoming familiar gatekeepers under a more appealing name.
Govern online communities through accountable choice, not hidden feeds
Make governance visible. The book imagines communities whose members set content standards, choose moderators, vote on proposals, and appeal decisions. Open-source software and independent audits could help members check whether a service follows its stated policies. People could also leave for another provider while taking their social data with them, making exit a form of accountability.
Identity protects voting. Anonymous accounts can make it easy to distort a vote by creating many fake identities, a problem known as a Sybil attack. The authors argue that digital credentials and privacy-protecting proofs could help establish that each participant is a unique person without revealing unnecessary personal information. They also acknowledge that government may be needed to require platforms to release historical data in a usable form.
The proposal treats governance as more than content moderation. It asks who sets standards, who can challenge decisions, and whether members can leave without losing their relationships. That is a meaningful democratic improvement over opaque, unilateral platform rules. Still, openness does not eliminate power: technical complexity can make an ostensibly public system legible only to experts, and majority votes can threaten vulnerable groups. Good governance needs transparent procedures, accessible explanations, appeal rights, and protections that do not depend on winning a popularity contest.
Build AI on human-controlled data before scaling its power
The same incentives scale up. The book sees generative AI as both a source of major benefits and an extension of the existing data economy. It describes a newborn whose seizures were traced to a genetic condition through rapid genome analysis, illustrating AI's potential in medicine. It also warns that opaque models trained on vast collections of personal and creative work can concentrate power, reproduce bias, and generate convincing falsehoods.
Change the conditions. The authors argue for human control over training data, transparent or open-source code, and accountability for those deploying AI. Their concern is not that machines independently choose to harm people. It is that systems built to serve narrow commercial goals can produce harmful outcomes at enormous scale, while the people whose data fuels them have little say.
This framing usefully shifts the safety debate from machine intentions to human incentives and institutional control. That is often where accountability can be assigned: developers, deployers, and organizations decide which objectives to optimize and which risks to accept. Open code can support scrutiny, but it is not a complete safeguard. Models may be hard to audit even when some components are visible, and publishing powerful systems can increase misuse. Governance therefore needs to match the risk, combining data rights with testing, access controls, and clear responsibility for harms.
Migrate toward open networks by making the alternative useful together
Adoption is the hard part. The book's proposed third-generation internet is called NewNet: a people-centered system with personal identity, data control, portability, interoperability, and a public network effect. Its authors compare the change to migration, drawing on the Irish families who crossed the Atlantic in search of a different life. The contrast is deliberate: moving to another digital service should not require the dangers or sacrifice of those historical journeys.
Start with collective action. A new service is less useful when friends and contacts remain on established platforms. The authors argue that interoperable connections, trusted institutions, and businesses adopting shared standards could help overcome this network-effect barrier. Their practical invitation is to explore emerging tools, spread the idea, and press organizations to adopt systems that give people more control.
The migration metaphor gives an abstract technical transition emotional force, while the emphasis on interoperability addresses the classic coordination problem of network effects. People do not switch merely because an alternative is more principled; it must work with their relationships and daily needs. That makes adoption depend on convenience, trust, and credible safeguards, not only persuasion. The book's optimism is energizing, but a new network can reproduce old concentrations of power if governance and funding are weak. The decisive test is whether users retain meaningful choice after the alternative becomes popular.
Analysis
The book is a reform manifesto built around a diagnosis and a proposed institutional redesign. Its central claim is that today's internet problems do not arise chiefly from bad users or individual pieces of content. They arise from a business and technical structure that concentrates personal data, attention, and influence in a small number of platforms. The Four Rs translate this diagnosis into a civic framework: protect individual rights, sustain shared responsibilities, align economic rewards with social value, and make rules accountable.
Its most persuasive insight is that portability and user control could weaken platform lock-in. A person who can take meaningful relationships and records elsewhere has more bargaining power than someone who can only download an unusable archive. The book also rightly treats AI as an institutional question. Powerful models inherit the incentives, data practices, and accountability structures of the organizations that build and deploy them.
The proposed solution is ambitious, but its claims exceed what technical architecture alone can establish. User-controlled identity does not automatically create informed consent, fair markets, or accurate information. Privacy-preserving proofs and open protocols can reduce certain forms of dependence, yet they introduce questions of security, accessibility, recovery, and unequal technical literacy. Decentralized systems can also create new intermediaries, and public access to a shared social graph requires careful limits to avoid exposing people through others' connections.
The book's historical analogies are compelling as invitations to imagine different arrangements, not as guarantees that decentralization produces democracy. Gutenberg's press enabled both learning and propaganda; similarly, an open internet can support cooperation or organized abuse. A durable reform agenda therefore needs technological redesign alongside competition policy, child protections, independent evaluation, and meaningful remedies for harm. The strongest practical lesson is not that DSNP alone resolves the crisis. It is that people need enforceable choices over identity and data, while institutions remain answerable for the consequences of using them.
People Also Read
Glossary
Four Rs
A civic design frameworkThe book's framework for evaluating an internet that serves people and society. Rights protect individual agency, responsibilities describe obligations to others and shared resources, rewards shape the incentives facing users and businesses, and rules establish governance and enforcement. The authors argue that these elements must work together rather than be treated as separate fixes.
digital feudalism
Platform control over digital livesThe authors' term for a system in which people depend on dominant technology companies that collect their data, control access to online services, and set the terms of participation. The analogy casts platforms as powerful landlords and users as people with little control over the information and relationships they generate.
social graph
A map of digital relationshipsA record of a person's connections and interactions across online services. It can include friends, accounts followed, messages, purchases, and shared activity. The book argues that platforms combine social graph information to infer interests and behavior, and proposes that people should control access to their own parts of this network.
Decentralized Social Networking Protocol
A user-controlled data protocolDSNP is the book's proposed protocol for giving people control over digital identity, data, and social connections across services. It is intended to support selective and revocable permission, portable information, and interoperable applications. The authors describe it as an additional layer on existing internet protocols, not a replacement for the internet's underlying network.
Three Ps
Personhood, proofs, and privacyThe authors' shorthand for qualities they want a human-centered internet to support. Personhood means an account can be linked to a unique human being. Proofs allow someone to verify a relevant fact without exposing unnecessary personal details. Privacy means individuals retain control over which information they share and with whom.
self-actualization
Developing and expressing one's identityThe book uses this term for a person's ability to shape a life and identity according to their own aims, rather than having them imposed by a government, company, or algorithm. The authors connect this capacity to control over personal data, arguing that digital systems should not covertly constrain how people understand or express themselves.
NewNet
A proposed people-centered internetThe authors' provisional name for a third-generation internet designed around people rather than machines or centralized platforms. Its envisioned features include user-controlled identity and data, permission-based sharing, portability, interoperability, decentralized storage, and a network effect available to many services rather than monopolized by one company.
Download PDF
Download EPUB
.epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.