The Digital Republic and Platform Democracy

The Book

The Digital Republic: On Freedom and Democracy in the 21st Century was published in the United States by Pegasus Books on July 5, 2022. The official Simon & Schuster publisher page lists the hardcover as distributed by Simon & Schuster, 304 pages, with ISBN13 9781643139012. Susskind is a barrister and the author of Future Politics; his own site describes his work as sitting at the crossroads of technology, politics, and law.

The book's central claim is not that technology replaces politics. It is that digital technology has become political infrastructure. Interfaces sort attention. Platforms set terms of association. Data practices decide what counts as a person, a risk, a market, a target, or a pattern. Susskind's republican language matters because it shifts the focus from consumer preference to domination: the problem is not only that users dislike some platform outcomes, but that private systems can structure options before public argument begins.

Digital Republic, Defined

A digital republic is a political order in which digital systems that structure visibility, association, trade, identity, ranking, and delegated action are subject to public law, evidence access, institutional duties, and democratic contest. It is not a demand that every service become state-owned. It is a demand that private ordering at public scale stop hiding behind the language of mere product design.

The working unit is the control point. A feed, app store rule, identity requirement, search ranking, ad auction, model API, workplace dashboard, public-service portal, payment rail, or agent permission can become a political object when it shapes what people can realistically do. A digital republic asks whether that control point is reviewable, proportionate, appealable, and capable of being changed by institutions other than the firm that benefits from it.

In that sense, platform democracy is procedural before it is participatory: inventory the control point, state its authority, preserve its evidence, give affected people recourse, and keep an independent institution able to require change.

The key word is domination. A person is dominated when another actor can shape the field of action arbitrarily: by changing rules, burying alternatives, imposing opaque scores, extracting data, limiting appeal, or making exit technically possible but socially ruinous. In that sense, platform democracy begins before elections and speech rules. It begins with the app store, the recommender, the ad system, the identity checkpoint, the payment rail, the search interface, the school platform, the work dashboard, and the agent that takes action on someone else's behalf.

Current Context

As of this review on June 23, 2026, Susskind's argument is no longer only a theory of future reform. The European Union's Digital Services Act, Regulation (EU) 2022/2065, is operational: its general date of applicability was February 17, 2024, and the European Commission directly supervises designated very large online platforms and very large online search engines. Commission materials now describe live enforcement tools, including requests for information, orders for access to data and algorithms, interviews, inspections, interim measures, preliminary findings, binding commitments, non-compliance decisions, and fines.

The DSA also turns platform power into records. It creates duties around notice and action, statements of reasons, complaint handling, advertising transparency, recommender-system information, dark-pattern restrictions, systemic-risk assessment, independent audits, and data access. Commission transparency materials say the DSA transparency-reporting format has been standardised and that the first harmonised reports were published in February 2026. On July 2, 2025, the Commission adopted a delegated act on data access for qualified researchers and launched a DSA data access portal. That is the practical version of one of the book's deepest concerns: public governance needs evidence from systems whose effects cannot be judged from the user interface alone.

The Digital Markets Act addresses a different layer of the same problem. The Commission describes the DMA as a law for fairer and more contestable digital-sector markets, aimed at gatekeepers that operate core platform services such as search engines, app stores, operating systems, social networks, messaging services, online intermediation, and advertising. The gatekeeper portal currently lists 23 designated core platform services. The DMA is not a speech or safety law, but it matters to platform democracy because bottleneck control can make private design choices feel like the only available world.

The EU AI Act, Regulation (EU) 2024/1689, adds an AI-system layer. The Commission's current implementation page says the Act entered into force on August 1, 2024; prohibited AI practices and AI literacy obligations began applying on February 2, 2025; governance rules and obligations for general-purpose AI models began applying on August 2, 2025; transparency rules are due in August 2026; and, following the 2026 political agreement on the AI omnibus proposal, the current high-risk timeline points to December 2, 2027 for systems in specified high-risk areas and August 2, 2028 for systems embedded in regulated products. This does not turn every platform problem into an AI Act problem. It shows that public law is being forced to distinguish platforms, markets, model providers, deployers, and high-risk uses instead of treating "technology" as one policy object.

A broader public-law layer is emerging as well. The Council of Europe describes its Framework Convention on Artificial Intelligence and human rights, democracy, and the rule of law as the first international legally binding treaty in this field; it opened for signature on September 5, 2024. Its public materials emphasize documentation, procedural rights, effective complaint routes, risk and impact assessment, and possible bans or moratoria for some AI uses. That matters for Susskind's argument because the object of governance is not only consumer harm or market power. It is whether AI and platform systems remain compatible with democratic institutions.

The United Kingdom is moving on a parallel safety track. Ofcom's Online Safety Act materials say illegal-harms duties required providers to assess risk by March 16, 2025 and, from March 17, 2025, to take the safety measures in the Codes or use other effective measures. Ofcom's child-safety statement required covered services to complete and record children's risk assessments by July 24, 2025, with safety measures from July 25, 2025. Ofcom's November 2024 open letter also explains that some generative-AI chatbot services can fall within the user-to-user regime when users can share AI-generated content, interact in group settings, or create user chatbots made available to others.

Power After the Interface

This makes The Digital Republic a useful companion to Code, The Black Box Society, and The Platform Society. Susskind is less interested in whether software is neutral than in what kind of power software makes routine. The interface is not just a window. It is a small constitution: it allocates choices, makes some acts easy, makes others invisible, and turns institutional preference into everyday friction.

That constitution is usually written in operational parts: ranking weights, review queues, API limits, default settings, fee schedules, suspension rules, fraud thresholds, age checks, permission scopes, and support scripts. None of those parts looks like a legislature. Together they can decide who speaks, earns, appears, buys, learns, travels, appeals, or receives service.

That is the right frame for this site's recurring concern with mediated reality. Belief and behavior do not spread only through doctrine. They spread through defaults, dashboards, prompts, notifications, metrics, and the ordinary pressure to comply with whatever the system makes legible. A recommendation feed can create a public without calling it one. A workplace platform can impose discipline without naming it command. A ranking system can create an attention economy before anyone gets to vote on its principles.

The interface also shifts accountability. When a feed amplifies, a marketplace ranks, a search engine summarizes, or a chatbot proposes the next action, the decision can be distributed across product teams, model providers, ad incentives, moderation rules, training data, user feedback, and business metrics. Susskind's contribution is to insist that distributed power is still power. If a system reliably shapes action, the fact that no single human editor pressed the final button is not enough to remove it from public scrutiny.

The democratic failure is often evidentiary before it is ideological. The people affected by a control point may be unable to see the rule, the data category, the threshold, the vendor change, the experiment, the appeal outcome, or the reason the system treated them differently. That is why public registers, transparency reports, audit trails, and appeal statistics matter: they make power inspectable enough for argument to begin.

Freedom Needs Institutions

The book is strongest when it refuses the old comfort that digital life can be governed by individual exit. Exit is weak when public conversation, work, identity, trade, and administrative access are mediated by a small number of platforms. A person may technically leave a service and still live inside the environment the service helped build. Susskind's answer is institutional: public standards, duties, rights, regulators, professional norms, and democratic control over infrastructures that affect civic life.

Exit is especially weak where the platform is also a credential, a workplace, a school portal, an app distribution layer, a public-service front door, or the only practical route to customers. The user can leave the account and still face the scoring system, the ranking logic, the identity requirement, the marketplace rule, or the API dependency. Susskind's point is not that exit never matters. It is that exit cannot substitute for governable institutions when the infrastructure is shared.

Those institutions need more than broad values. They need powers to inspect evidence, require records, test risk claims, protect users and workers from arbitrary treatment, demand interoperability or portability where lock-in becomes civic dependence, and set boundaries around surveillance and automated discipline. The lesson is not "regulate technology" in the abstract. It is to locate the control point, name the duty, preserve the evidence, create an appeal path, and give an accountable public body enough capacity to act.

The Agent Reading

Read in 2026, the book points beyond social-media governance toward AI agents. An agent that books travel, filters applicants, drafts correspondence, updates records, buys services, or routes requests does not need inner life to exercise institutional force. It needs permission, integration, memory, objectives, and an organization willing to treat its output as action. The political question becomes sharper: when automated systems do things, who is entitled to inspect the grounds, contest the result, limit the mandate, and assign responsibility?

Susskind's republican vocabulary helps because agentic AI is often sold through convenience. Convenience is real, but it is also how power becomes ambient. When a system suggests the next step, pre-fills the justification, selects the recipient, and scores the priority, it narrows action while feeling helpful. Freedom in that environment is not restored by a checkbox saying the user remains responsible. It requires logs, appeal, procurement rules, domain limits, human review with authority, and public standards for when delegation is unacceptable.

That connects the book to the site's work on AI agents, agent identity, AI browsers as control surfaces, agent stores, and computer-use systems. An agent governance program should ask who authorized the action, what account or credential the agent used, which tools were available, what data was read, what record was changed, what human could stop it, and what receipt proves what happened.

The agent version needs a chain of mandate: task, account, tool, data source, approval gate, log, rollback path, and liable owner. Without that chain, delegated convenience becomes unassigned public power.

Governance and Safety

A digital-republic reading turns governance into a series of concrete design and enforcement questions:

• Purpose and inventory. Institutions should know which platform, model, recommender, identity, payment, moderation, and agent systems they rely on, why each exists, who owns it, and what public function it affects.
• Public mandate. High-impact systems should have a plain statement of public purpose, affected population, lawful authority, nontechnical alternatives considered, proportionality limits, and the office or body that can stop the use.
• Terms and interface governance. Rules, defaults, nudges, recommender settings, ad parameters, age gates, and dark-pattern controls should be treated as governance instruments, not cosmetic product choices.
• Evidence and audit. Risk assessments, transparency reports, ad libraries, system cards, audit trails, incident reports, researcher access, and regulator access matter because they turn private power into contestable evidence.
• Contestability. Notice, appeal, human review, explanation, and recourse must attach to consequential decisions, including content removal, account restriction, ranking demotion, benefit denial, employment screening, school discipline, and automated service refusal.
• Exit and interoperability. Where a platform becomes shared infrastructure, governance should test portability, data export, switching costs, API dependence, record preservation, and whether exit is practical for the people or institutions asked to rely on it.
• Delegation limits. Agents need mandate boundaries, permission maps, revocation controls, action logs, identity bindings, approval gates, procurement terms, and clear rules for domains where automation should assist rather than decide.
• Independent enforcement. Public bodies need technical capacity, legal authority, funding, and procedural safeguards. Otherwise "governance" becomes a documentation ritual for choices already made elsewhere.

For high-impact platform or agent deployments, the evidence file should preserve the system owner, vendor, model or ruleset version, policy basis, data categories, ranking or routing logic at the right level of abstraction, human-review authority, appeal outcomes, incident triggers, rollback conditions, and post-deployment monitoring plan. Without that file, a public institution may be able to say that a system helped, but not prove what happened when it harmed.

A democratic control file should join the public mandate, system inventory, procurement terms, data-flow map, risk assessment, audit log, appeal metrics, incident record, and retirement rule. It is the answer to the platform-era habit of treating power as an interface feature.

The safety implication is sober. A platform or agent system can harm people without being conscious, malicious, or autonomous in any grand sense. Harm can come from scale, opacity, incentive alignment, weak appeal, bad procurement, misleading design, fragile identity controls, overbroad surveillance, or organizational willingness to launder decisions through software. That is why safety belongs with governance, rights, labor, and evidence rather than only with model benchmarks.

Where the Book Needs Care

The risk in The Digital Republic is that republican repair can sound cleaner than implementation. Regulators need money, expertise, jurisdiction, political backing, and the courage to say no to profitable systems. Professional duties matter, but they do not by themselves overcome cloud concentration, advertising incentives, lobbying, procurement lock-in, or labor markets where workers have little power to refuse surveillance and automation. The book is best read as a theory of public control, not as a complete account of political economy.

It also needs a sharper state-power warning. Public control is not automatically democratic control. Age assurance can protect minors while expanding identity checkpoints. Safety mandates can become pressure for over-removal. Researcher access can expose sensitive data if governance is weak. Lawful oversight can slide into surveillance if proportionality, due process, privacy, and judicial review are not built into the institution.

That warning applies to republican language itself. A digital republic should not mean state access to every private platform record, mandatory identity for every interaction, or automated enforcement of every contested norm. It should mean bounded public power: enough access to inspect and remedy public-scale harms, constrained by privacy, minimization, independent review, and appeal.

Democratic control also needs a privacy floor. Researcher access, regulator inspection, ad archives, and public registers should reveal system behavior without turning users, workers, children, or complainants into permanent evidence subjects. The record has to be useful enough to test power and limited enough not to become another extraction system.

The labor reading remains essential. A democratic digital order cannot only protect the user-citizen facing the screen. It must also account for moderators, data labelers, warehouse workers, drivers, outsourced annotators, public servants, teachers, nurses, clerks, and office workers whose tasks are reorganized by metrics and automation. Digital domination often arrives as management. A republic that governs platforms while ignoring workplaces would leave too much power intact.

What This Changes

The Digital Republic gives this site a disciplined language for a recurring concern: automated systems shape belief and action before they become explicit ideology. It is not enough to ask whether a tool is accurate, innovative, or popular. Ask what authority it accumulates, what alternatives it hides, what forms of contest it permits, and what kind of person it assumes the user should become.

The practical test is simple. For any consequential platform or AI system, ask who sets the defaults, who benefits from friction, who can inspect the evidence, who can appeal, who can build an alternative, who bears the error, and who is accountable when private infrastructure performs a public function. The answer should appear in procurement files, public registers, audits, system cards, safety cases, transparency reports, and appeal procedures, not only in brand language.

This is the same pattern behind public registers, system cards, safety cases, agent receipts, and humane friction: a system that governs conduct must leave a record that survives the product story.

Its final value is restraint. Susskind does not require mystical claims about machines. The problem is worldly: institutions are delegating public functions to private technical systems faster than publics can inspect them. A digital republic would treat that as a constitutional problem, not a customer-service complaint. For an archive concerned with AI, cyberculture, surveillance, labor, and belief, that is the useful lesson: govern the loop before the loop becomes the normal way reality is administered.

Related Pages

• Digital Services Act, EU AI Act, Platform Governance, Content Moderation, Recommender Systems, and Algorithmic Transparency cover the platform-law layer.
• Platform Monopoly Power, Election Integrity and AI, Cognitive Sovereignty, and The Age of Surveillance Capitalism connect democratic freedom to chokepoints, attention, data capture, and belief formation.
• Notice and Appeal, Right to Explanation, Algorithmic Recourse, Human Oversight of AI Systems, and AI Audits and Third-Party Assurance turn the republican idea into checks on decisions.
• Vendor and Platform Governance, Transparency and Public Registers, AI System Inventory, AI Procurement, AI Incident Reporting, Digital Public Infrastructure, Public Interest Technology, The AI Register Becomes Public Memory, The System Card Becomes a Release Ritual, and The Safety Case Becomes the Release Gate collect the evidence and procurement side.
• AI Agents, AI Browsers and Computer Use, AI Agent Identity, Agent Tool Permission Protocol, Agent Audit and Incident Review, The Agent Log Becomes the Receipt, The Agent Identity Becomes the Service Account, The Agent Store Becomes the App Store, and The AI Browser Becomes the Control Surface extend the book's platform argument into delegated action.
• Data Minimization, Privacy and Data, Agentic Commerce, AI Agent Observability, and Humane Friction Standard cover the privacy, commerce, logging, and friction tradeoffs that follow once governance becomes operational.

Source Discipline

Book facts here come from the official publisher page and the author's site. Legal and current-context claims use official legal texts, European Commission implementation pages, and Ofcom regulatory materials. Secondary reviews are useful for reception, but they should not carry claims about legal status or current enforcement.

Regulatory dates require care. Commission and Ofcom implementation pages change as designations, guidance, codes, enforcement actions, and timelines move. The AI Act paragraph above follows the Commission's current implementation page as reviewed on June 23, 2026, including the 2026 simplification-package timeline; article-level legal obligations still require the official text, subsequent amendments, guidance, and jurisdiction-specific advice.

International instruments require the same care. The Council of Europe Framework Convention is relevant because it frames AI through human rights, democracy, and rule of law, but implementation depends on parties, reservations, national measures, monitoring, and later practice. It should not be used as shorthand for a finished global AI regime.

Transparency data also need humility. A platform risk assessment, ad library, statement-of-reasons database, audit, or system card is evidence about a reporting system, not a complete picture of social harm. The stronger claim is narrower: without such records, public institutions cannot even begin to test the stories that platforms and AI vendors tell about their own power.

Do not treat designation, investigation, preliminary finding, commitment, non-compliance decision, and fine as interchangeable. Do not treat the existence of a portal, register, audit right, or public database as proof that a system is safe or democratic; it is a route to evidence.

Sources

• Simon & Schuster, The Digital Republic, official publisher page for Pegasus Books hardcover, publication date, length, ISBN13, and author biography, reviewed June 23, 2026.
• Jamie Susskind, official author site, author biography and listing for The Digital Republic: On Freedom and Democracy in the 21st Century, reviewed June 23, 2026.
• EUR-Lex, Regulation (EU) 2022/2065, Digital Services Act, official legal text, reviewed June 23, 2026.
• European Commission, The Digital Services Act, official policy page for rights, provider scope, platform duties, and Commission/national supervision, reviewed June 23, 2026.
• European Commission, The enforcement framework under the Digital Services Act, official overview of investigatory and sanctioning tools, reviewed June 23, 2026.
• European Commission, Supervision of the designated very large online platforms and search engines under DSA, information updated May 28, 2026; reviewed June 23, 2026.
• European Commission, How the Digital Services Act enhances transparency online, official overview of transparency reports, statements of reasons, researcher access, risk reports, audits, and terms-and-conditions records, reviewed June 23, 2026.
• European Commission, DSA Transparency Database, official statements-of-reasons database, reviewed June 23, 2026.
• European Commission, Delegated act on data access under the Digital Services Act, July 2, 2025; reviewed June 23, 2026.
• European Commission, DSA data access portal, reviewed June 23, 2026.
• European Commission, Digital Markets Act, official DMA overview, reviewed June 23, 2026.
• European Commission, DMA designated gatekeepers, official gatekeeper and core-platform-service portal, reviewed June 23, 2026.
• European Commission, AI Act overview, official policy page for Regulation (EU) 2024/1689, governance, GPAI rules, transparency obligations, and implementation timeline, reviewed June 23, 2026.
• EUR-Lex, Regulation (EU) 2024/1689, Artificial Intelligence Act, official legal text, reviewed June 23, 2026.
• Council of Europe, Framework Convention on Artificial Intelligence and human rights, democracy and the rule of law, official convention page, opened for signature September 5, 2024; reviewed June 23, 2026.
• UK legislation.gov.uk, Online Safety Act 2023, official revised legislation, reviewed June 23, 2026.
• Ofcom, Statement: Protecting people from illegal harms online, published December 16, 2024 and last updated June 9, 2026; reviewed June 23, 2026.
• Ofcom, Statement: Protecting children from harms online, published April 24, 2025 and last updated July 30, 2025; reviewed June 23, 2026.
• Ofcom, Open letter to UK online service providers regarding Generative AI and chatbots, November 8, 2024; reviewed June 23, 2026.

Book links are paid affiliate links. As an Amazon Associate I earn from qualifying purchases.

• Amazon, The Digital Republic by Jamie Susskind, reviewed June 23, 2026.

Return to Blog · Return to Books