Informational Democracy: A Complexity Approach to Epistemic and Institutional Innovation. Working Group │ Max Planck Institute for Political and Social Science │ June – December 2026.
We are living through the second great transformation of democratic information — and to understand this moment, we need to understand the first crisis of what was called the public sphere.
In 1922, Walter Lippmann published Public Opinion, his devastating diagnosis of the gap between the world and the pictures citizens form of it. In The Phantom Public (1925), he pressed the conclusion further: the fully informed citizen was a fiction, he argued; democratic governance as traditionally conceived rested on an assumption that could not survive contact with the complexity of modern industrial society. Citizens could not possibly process the complexity of modern governance. His conclusion was technocratic — experts must govern because citizens cannot.
John Dewey disagreed. In The Public and Its Problems (1927), he accepted Lippmann\'s diagnosis while rejecting his remedy. The problem was real: industrial society had produced a "Great Society" too vast and interconnected for citizens to comprehend. But Dewey\'s answer was institutional rather than technocratic. Democracy required new forms of communication, new structures of association, new ways of making complex interdependence visible and governable. It required redesign, not abandonment.
That debate was never resolved. It was superseded — by postwar prosperity, by the apparent stability of mass democracy, by a world in which broadcast media, mass education, and professional journalism created a functioning if imperfect epistemic infrastructure. Citizens could not know everything, but institutions could channel what mattered through shared channels that most people trusted most of the time. A working informational regime, however fragile, held.
Then the internet arrived. And with it, a transformation so rapid and so total that its full implications are only now becoming visible.
The digital revolution did not simply give us more information. It changed the nature of information itself. For most of human history, producing and distributing information was expensive, slow, and gatekept. Manuscripts, printed books, newspapers, broadcasting — each successive technology widened the circle of information access, but each maintained the fundamental structure of scarcity: a relatively small number of producers, a relatively legible set of channels, a shared informational commons that citizens navigated differently but inhabited together. Democratic institutions were built around this structure. They assumed it. Their entire architecture — the press, the broadcaster, the expert, the official record — depended on information being scarce enough to channel and abundant enough to inform.
The internet inverted that structure. As James Gleick chronicled in The Information: A History, A Theory, A Flood (2011), the logic of information has always moved toward abundance — but the digital turn made that movement catastrophic in its speed and scale. When information is cheap, attention becomes expensive. What was once the problem — too little information reaching too few people — became the solution to a problem that had not yet arrived: too much information, arriving everywhere at once, with no institutional infrastructure to evaluate, sort, or legitimize it.
The consequences unfolded in two waves. The first was structural: the collapse of the gatekeepers. Editors, broadcasters, professional journalists — the institutions that had performed the epistemic triage function of democratic information — lost their privileged position in the information ecosystem. Anyone could publish. Anyone could reach an audience. This was, in important respects, liberating. It opened channels that had been closed, gave voice to those who had been excluded, and created new possibilities for distributed knowledge production and civic coordination that could not exist within hierarchical media systems.
The second wave was algorithmic. Social media platforms, designed not for democratic deliberation but for engagement and advertising revenue, replaced the collapsed gatekeepers with a different kind of epistemic infrastructure — one optimized for attention capture rather than truth-tracking, for emotional resonance rather than verification, for tribal reinforcement rather than shared deliberation. The result was not, as early internet optimists had hoped, the emergence of a richer public sphere. It was the fragmentation of the informational commons into divergent epistemic territories, each with its own sources, its own authorities, and its own version of events.
Yochai Benkler\'s analysis in The Wealth of Networks (2006) had opened a genuinely hopeful line of argument: that distributed digital networks make possible new forms of collective intelligence, peer-validated knowledge, and civic coordination unavailable within hierarchical systems. That potential is real. But it has been largely captured by platform architectures that harness network dynamics for commercial rather than democratic ends — and the question of how to release that potential for democratic purposes remains urgently open.
Now, artificial intelligence has arrived as a third wave — and the implications for democratic information are still being formed. Large language models can generate fluent, convincing text at scale and at negligible cost. Synthetic content is becoming indistinguishable from authentic content. The epistemic infrastructure of democratic life — already under strain from the collapse of gatekeeping and the fragmentation of algorithmic media — faces a challenge that is qualitatively new: not just more information, not just differently mediated information, but information whose origins, authorship, and relationship to truth cannot be assumed.
At the same time, AI offers tools for democratic deliberation of unprecedented power: collective sense-making at scale, AI-assisted synthesis of diverse perspectives, new architectures for distributed intelligence that could help democratic institutions handle the complexity they were never designed to manage. Whether these tools will be used to deepen democratic capacity or to hollow it out further is not determined. It is a design question.
The Lippmann-Dewey question has returned, at a scale and velocity that makes the 1920s look like a rehearsal. Lippmann\'s diagnosis — that complexity had outrun the citizen\'s capacity for comprehension — was correct for industrial society. For the society produced by the internet and artificial intelligence, it is even more poignant. And Dewey\'s instinct — that the answer was institutional redesign, not democratic retreat — remains the right one.
What neither Lippmann nor Dewey had was a theoretical and practical framework adequate to the scale of the challenge. Building one begins with a working concept. Informational Democracy is not offered here as a settled theory but as a way to ask how democratic systems depend on the infrastructures, institutions, and practices through which societies produce knowledge, make meaning, and coordinate collective judgment. Understood through its informational conditions, democracy's current crises can be diagnosed more precisely — and institutional responses imagined that go beyond restoring old gatekeepers, regulating content, or educating citizens. That is the inquiry this working group is convened to undertake.
This project is a first act of field-building — this requires both ambition and honesty about what six months can achieve. Field-building is not the same as producing research. Research generates knowledge. Field-building generates the conditions under which knowledge can accumulate, compound, and travel. A field is an infrastructure: shared vocabulary, productive provocations, common methods, and a community of practice capable of sustaining inquiry across institutions and generations. Without that infrastructure, insights remain isolated. Disciplines are not discovered; they are constituted.
The field this project is building is informational democracy as a design science. That ambition will not be completed in six months. What this working group can do — and what it is convened to do — is to take the constitutive steps: to draw the right conversations together, to establish the provocations worth arguing about, and to create the intellectual community within which a new domain can begin to grow. The longer horizon is implementation and policy: concrete proposals for the redesign of democratic institutions as informational regimes, developed in sufficient depth and with sufficient institutional grounding to be picked up by actors in government, technology, and civil society. Those proposals will not emerge fully formed from this first cycle of work. But the agenda-setting work done here, the networks established, and the conceptual foundations laid will make them possible.
That longer horizon begins, necessarily, from the present moment in artificial intelligence. The latest advances in large language models have made it impossible to think about democratic information without thinking about AI — not mainly or only as a future risk to be anticipated, but as a present condition to be reckoned with. More acutely still, the emergence of agentic AI — systems that do not merely generate content but act autonomously in the world, pursue goals across extended time horizons, and interact with other agents at scale — introduces a qualitatively new dimension to the democratic information problem.
The field this project is constituting sits at the intersection of complexity science, democratic theory, AI governance, political economy, and institutional design. These conversations are productive in isolation. What is missing is the act of drawing them together — across disciplines, across sectors, across the divide between theory and practice — and establishing the shared intellectual infrastructure within which a new domain can grow. The networks this working group will establish — among scholars, practitioners, technologists, and policy actors in Germany, across Europe, and in the broader international conversation about democratic renewal — are part of this intellectual project. A field without a community is not a field. Building that community, at the cutting edge of democratic innovation and technological reflection, is one of the working group\'s primary purposes.
There is a ghost haunting democratic theory. Its name is the public sphere.
In 1962, Jürgen Habermas published Strukturwandel der Öffentlichkeit — translated into English three decades later as The Structural Transformation of the Public Sphere — and in doing so established what would become the dominant normative framework for thinking about democracy and information in the twentieth century. The argument was historical and philosophical: that the eighteenth-century bourgeois public sphere had created, for the first time, a social space between the state and the private individual in which citizens could engage in rational-critical debate, hold power accountable through reasoned argument, and constitute themselves as a democratic public. The press was its instrument, coffeehouses its arena, and the force of the better argument its regulative ideal. Democracy, on this account, was fundamentally a communicative achievement — and its health turned on the quality of public deliberation.
Habermas knew this model was idealized, and he knew it was already under threat at the moment he described it. The structural transformation of his title was a story of decline: the colonization of the public sphere by commercial media, public relations, and the administered state. His later work — The Theory of Communicative Action, the procedural turn in Between Facts and Norms — sought to reconstruct a normative foundation for democratic legitimacy that could survive this decline. The project was enormously generative. It produced a field: decades of public sphere scholarship, deliberative democracy theory, media studies, political communication research — all organized around Habermas\'s central problematic, all oriented by his normative horizon.
That field is now one of the largest and most institutionally embedded in the social sciences. And it has become, in a precise sense, part of the problem.
Not because Habermas was wrong — within the structural conditions he described, his account was penetrating and its normative aspirations genuine. But because the social and technological conditions that gave the model its purchase have been dissolved, and the academic infrastructure built around it has outlived them. The gatekeepers whose passing Habermas mourned are gone. The bounded, legible public whose deliberative capacity he sought to protect has fragmented beyond recovery. The scarcity of information that made the public sphere both possible and necessary has been replaced by a surfeit that no institutional arrangement oriented toward Habermasian ideals can navigate. Yet the field continues — producing ever more refined accounts of deliberative procedures, communicative rationality, and the conditions of ideal speech, as if the structural conditions for these to operate had not been permanently altered. It mistakes a map for a territory that no longer exists.
This is not merely an academic problem. It has political consequences. When the dominant intellectual framework for thinking about democracy and information is oriented toward recovering a lost model rather than designing a new one, it shapes how institutions respond to crisis — and consistently in the wrong direction. Calls to restore journalistic gatekeeping, to protect the rational public sphere from disinformation, to recover the conditions for authentic deliberation: all are responses calibrated to a world that will not return. The nostalgia is understandable. The fetishization of what once was, as if the arrow of history could be reversed by sufficient institutional will, is a form of political risk. Democracies that look backward for their informational models will not find answers to the challenges forming ahead of them.
The dominant models now emerging — and hardening into infrastructure — are not the Habermasian public sphere but its successors by displacement. In the United States, the platform economy has produced an extractive informational architecture: one that harvests attention, monetizes engagement, and concentrates epistemic power in a small number of private actors whose incentives are commercial rather than civic. In China, a different architecture has emerged: one in which the state deploys digital infrastructure as an instrument of surveillance, social management, and the pre-emptive foreclosure of dissent. These two models are routinely presented as opposites — the free market against the authoritarian state, democratic openness against digital totalitarianism. The reality is less comfortable. Both are systems of concentrated informational power. Both optimize for the extraction of value from citizens' communicative behavior — one for advertising revenue, one for political control. Both erode the conditions for genuine epistemic autonomy.
What artificial intelligence adds to this picture is more than a new dynamic, it is an acceleration of an existing one — and a concentration of a different order. The development and deployment of the most consequential AI systems is controlled by a handful of private actors in the United States and a state-directed ecosystem in China. In both cases, the architecture of AI mirrors the architecture of the informational regimes it is emerging from: optimized for scale, oriented toward control, and structurally indifferent to the epistemic conditions that democratic self-governance requires. The differences between the two models are real, but the similarities are deepening, and the convergence is instructive: in the absence of a deliberate alternative, the default trajectory of digital information infrastructure — platform, algorithmic, and now AI — runs toward concentration, extraction, and control, whatever the political system surrounding it.
Europe stands at a different moment — and the difference is structural. Neither model is fully entrenched. The legal and regulatory architecture of the European Union has created, imperfectly but genuinely, a space in which information can be governed in the public interest rather than extracted for private gain. The GDPR, the Digital Services Act, the AI Act — whatever their limitations — represent an institutional commitment to democratic accountability over digital infrastructure that has no equivalent elsewhere at scale. European cities, from Barcelona to Amsterdam to Helsinki, have pioneered experiments in data sovereignty, digital democracy, and civic technology that offer working prototypes of different arrangements.
This working group proceeds from the conviction that the moment to act is now — in anticipation of the crisis that is forming. We are looking at least five years ahead: not predicting the future, but reasoning from plausible trajectories to the institutional designs that a more complex, more algorithmically mediated, more AI-saturated democratic life will require. The diagnostic work has been done. What is needed now is anticipatory design.
The informational crisis of democracy is best described as three interconnected problems, operating simultaneously on the demand side, the supply side, and the systemic level. Understanding their interaction is the first task of any serious response.
The demand side of democratic information — the receiving end, the audiences, the publics — has undergone a structural transformation. Lippmann and Dewey both assumed, however critically, the existence of something like a public: a body of citizens whose informational needs and capacities, however diverse, were navigable within a shared framework. That assumption can no longer be sustained.
Contemporary democratic societies are characterized by what Steven Vertovec has called superdiversity: a complexification of social life that goes beyond ethnic or cultural plurality to encompass the multiplication and intersection of social categories, identities, legal statuses, and life trajectories in ways that produce emergent configurations that supersede prior forms of diversity. Superdiversity is not simply more diversity. It is a qualitatively different condition — one in which the social categories through which democratic institutions once addressed their publics have become fluid, porous, and internally stratified to the point where no singular public can be assumed. The social complexity that Dewey confronted in industrial America was, by this measure, comparatively simple.
For democracy as an informational system, this matters profoundly. There is no longer a singular public to inform, deliberate with, or address. There is a radical multiplicity of audiences, each positioned differently in relation to information, expertise, and institutional authority. Democratic institutions designed to communicate with a relatively legible public are now operating in conditions of social complexity they were never designed to handle. The result is not merely a communication problem. It is a structural mismatch between the social reality institutions must govern and the social reality they were built to engage.
But the demand-side crisis runs deeper than fragmentation. Platform architectures optimized for engagement have not merely divided the informational commons — they have actively produced a new social condition: the atomized citizen. Connected to everything, anchored to nothing, algorithmically sorted into micro-communities of reinforcement rather than genuine association, individuals in contemporary democracies experience isolation and connection as simultaneous conditions. The social fabric through which shared reality was once maintained — through local institutions, professional communities, civic associations, shared media — has been thinned to the point where many citizens navigate their informational environment without the anchoring relationships that once made epistemic trust possible. Conspiratorial thinking is not a symptom of irrationality. It is the rational response of individuals stripped of those anchors: narratives that offer coherence, belonging, and explanatory power when legitimate epistemic institutions have ceased to provide them. The demand side of the democratic information problem is not simply a public that has lost its gatekeepers. It is a citizenry that has, in significant measure, lost the social conditions for epistemic trust.
On the supply side, the nature of information itself has changed. The digital revolution did not simply multiply sources; it dissolved the institutional arrangements through which democratic societies once maintained shared epistemic standards. The scarcity that structured the original informational problem of democracy — getting reliable knowledge to citizens — has been replaced by a surfeit in which the challenge is no longer access but navigation, not distribution but evaluation. The problem has inverted.
Technology is the central driver of this inversion, but it is not a neutral one. Algorithmic mediation shapes what reaches whom; synthetic content blurs the boundary between the authentic and the fabricated; platform architectures reward engagement over accuracy. The informational environment is no longer a shared landscape that citizens navigate differently. It has fractured into divergent epistemic territories. And with AI, the conditions for producing and distributing information have changed again — in ways that are still unfolding and whose democratic implications remain deeply uncertain.
Yet informational complexity is also a condition of possibility. Yochai Benkler\'s account of networked information environments in The Wealth of Networks showed that distributed digital networks make possible genuinely new forms of knowledge production, collective sense-making, and civic coordination that are unavailable within hierarchical information systems. The same structural conditions that have fractured the epistemic commons have also enabled new architectures for distributed intelligence, peer validation, and participatory knowledge-making. The question is whether democratic institutions can be redesigned to harness these possibilities rather than be overwhelmed by the pathologies they have so far produced.
Beneath both the social and informational transformations lies a deeper shift in the nature of the systems democratic institutions must govern. Contemporary societies are complex systems in the full technical sense: they are characterized by emergence, feedback loops, non-linear dynamics, and threshold effects that cannot be captured by the linear models of cause and effect on which democratic institutions still depend. Small perturbations produce large, unpredictable consequences. Interventions generate unintended effects. Systems shift discontinuously rather than gradually. Stability, where it exists, is dynamic rather than static — maintained by continuous adaptation, not equilibrium.
This is the condition that complexity science has developed the tools to understand. And it is the condition that democratic institutions were least designed to handle. They were built for a world of manageable interdependencies and predictable causal chains. The operating assumptions embedded in their design — that problems can be bounded, that solutions can be implemented and evaluated, that expertise can be concentrated and applied — were not unreasonable in the world that produced them. In conditions of systemic complexity, they become sources of institutional failure.
Together, these three complexities — social, informational, systemic — constitute the condition this working group calls informational democracy's defining crisis. They are not independent phenomena. Social complexity multiplies the difficulty of informational complexity; systemic complexity amplifies the consequences of both. What has emerged is a mutually reinforcing set of pressures that no single intervention can resolve. What is required is a rethinking of democracy itself as an informational regime.
The deepest contribution Lippmann and Dewey made was not their competing prescriptions. It was the recognition, shared between them, that democracy is fundamentally an informational project — that its success or failure turns on whether citizens can form warranted beliefs about the world, deliberate collectively on the basis of those beliefs, and hold power accountable through them. Democracy does not merely require information as an input. It depends on a functioning epistemic infrastructure: shared mechanisms for producing, evaluating, and distributing knowledge across a polity.
For most of the history of modern democracy, the central challenge of this infrastructure was access. Information was expensive to produce, hard to distribute, and controlled by a relatively small number of actors. Democratic institutions were designed around that problem — to widen access, to enable informed participation, to create the conditions for public deliberation in a world of scarcity. That problem has been solved. And in being solved, it has given way to a harder one.
The problem is no longer access. It is trust. But trust, in conditions of informational abundance, is not only an epistemic problem. It is a social one — and its social dimensions have been systematically underestimated. The collapse of gatekeepers matters, but it does not fully explain the depth of the democratic information crisis. What equally matters is what has happened to the individuals on the receiving end. Platform architectures optimized for engagement have not merely fragmented the informational commons. They have actively produced the atomized citizen — connected to everything, anchored to nothing — and with that atomization, the social substrate in which conspiratorial thinking flourishes. Democratic renewal requires addressing the social conditions of epistemic trust, not only its informational architecture.
This is what it means to treat democracy as an informational regime. The question is not which policies democracies adopt toward information but what informational architecture makes democracy possible at all — what structures of epistemic trust, knowledge production, and collective sense-making enable a polity of radically diverse citizens to govern themselves in conditions of permanent complexity.
The question before us is not how to fix democracy\'s information problem. It is how to redesign democratic institutions as informational regimes capable of sustaining epistemic trust under conditions of permanent social, informational, and systemic complexity.
Artificial intelligence has arrived into this condition as an accelerant and a transformer. The implications extend well beyond content generation. Agentic AI — systems that act autonomously, coordinate with other agents, and pursue goals across extended sequences without human involvement at each step — introduces challenges and possibilities for which democratic institutions have no existing framework. Where information abundance challenged citizens' capacity to evaluate what they received, agentic AI challenges the prior question: whether the informational environment citizens inhabit is the product of human agents at all. Fabricated consensus, synthetic participation, automated influence at population scale — these are not hypothetical futures. They are the conditions democratic institutions must now be designed to withstand.
Yet artificial intelligence is also the most significant opportunity for democratic renewal since the printing press — if its capacities are redirected toward democratic rather than extractive ends. Andrew Hall's concept of political superintelligence points toward one dimension of this possibility: AI that makes citizens genuinely smarter, representatives more faithful to those they represent, and institutions more responsive than anything two millennia of democratic experimentation have yet produced. The challenge Hall identifies is structural — not whether such tools can be built, but whether their infrastructure can be owned by citizens rather than controlled by a small number of private companies. This is not a technical question. It is a constitutional one.
Audrey Tang and Caroline Green's Civic AI project offers a complementary and more architecturally specific vision. Where Hall asks what AI could do for democracy, Tang and Green ask how AI can be governed democratically — from the ground up, through what they call the Kami: bounded, local AI stewards, community-governed rather than platform-controlled, designed to help neighbourhoods, schools, unions, and civic groups do what collective self-government has always promised but rarely delivered at scale. The Civic AI framework insists that alignment is not solved once by engineers at the top. It is maintained continuously, in public, through institutions that show their work, repair harm openly, and carry civic memory across generations. This is Dewey's instinct — that the answer to democratic complexity is institutional redesign, not technocratic substitution — made concrete and operational for the age of AI.
A century of democratic theory has developed the conceptual vocabulary to think more precisely about what such a redesign requires. Hélène Landemore's work on democratic reason and open democracy argues that democratic institutions have an epistemic edge precisely when they are genuinely inclusive — that cognitive diversity across a polity is not a problem to be managed but a resource to be activated. Cristina Lafont's "Democracy Without Shortcuts" develops a complementary argument: against conceptions of democracy that delegate epistemic authority to experts or algorithmic systems, Lafont insists that democratic legitimacy requires citizens to own and endorse the decisions that govern them. There are no shortcuts. What Landemore, Lafont, Tang, and Hall together suggest is that the informational conditions of democracy are not peripheral to its theory. They are central to it. The crisis of democratic epistemics is not a crisis of communication or media policy. It is a crisis of democratic self-government.
This working group begins from a premise that distinguishes it from most existing initiatives on democracy and information: complexity is not a problem to be overcome. It is the permanent condition within which democratic institutions must learn to operate — and, approached with the right conceptual tools, a resource for institutional design.
Complexity science has shown that adaptive, resilient systems share structural features: they maintain diversity rather than enforcing uniformity; they enable distributed sensing rather than centralizing information; they embed feedback mechanisms that allow learning rather than optimizing for a stability that may no longer be achievable; they generate trust through repeated interaction and demonstrated reliability rather than through authority or hierarchy. These are not abstract principles. They are design specifications — for institutions, for governance architectures, for the informational infrastructure of democratic life.
The central resource that democratic institutions need to cultivate in conditions of complexity is epistemic trust: the capacity of citizens, institutions, and communities to form warranted beliefs about the world, about each other, and about the information they receive. Epistemic trust is not the same as agreement. It does not require consensus on values or identity. It requires functional confidence in shared mechanisms for producing and evaluating knowledge — and in the institutions that maintain those mechanisms. Under conditions of informational abundance and social superdiversity, that trust cannot be produced by access alone. It must be built through institutional architecture.
This is precisely the question that the most generative practice-oriented thinking on institutional design is beginning to address. Indy Johar and Dark Matter Labs have argued that the institutions we can see — governments, parliaments, agencies — are only the visible surface of a deeper institutional infrastructure: the dark matter of monetary systems, governance protocols, regulatory frameworks, and civic architectures that shapes what any visible institution can do. Redesigning democratic institutions for an age of complexity requires making this dark matter visible — and reimagining it. This is a design challenge as much as a theoretical one: it requires working simultaneously at the level of concepts and the level of concrete institutional architectures.
Understanding how to design those architectures — and how complexity thinking, cognitive science, democratic theory, and the emerging practice of collective intelligence can inform and improve that design — is the working group's core intellectual task.
The Provocations are not assignments. The term is deliberate: a Provocation is not a hypothesis and not a research question. It is an assertive, directional claim — about what is failing, what is possible, and what democratic institutions must become — staked in order to be argued with. Where a hypothesis invites testing within an established framework, a Provocation contests the framework itself. It is designed to generate productive disagreement, to sharpen thinking, and to advance an agenda rather than to be confirmed or refuted.
The Provocations are not assigned. They are developed. Each working group member enters into an initial dialogue with the co-leads to develop their own Provocation — one that reflects the distinctive perspective they bring to the project. That Provocation then guides three connected essays across the arc of the cycle: grounding the claim in evidence and theory, visioning what it implies for democratic institutions, and demonstrating what it would require in practice. The process is collaborative by design: the working group does not assemble experts to endorse predetermined conclusions. It convenes an intellectual community to work on genuinely difficult questions and to produce, from that work, insights that none of its members could have reached alone.
The working group brings together scholars and practitioner-intellectuals whose work crosses the boundaries between complexity science, democratic theory, cognitive science, computational social science, institutional design, and non-Western democratic experience. The composition reflects a deliberate commitment: to genuine interdisciplinarity rather than multidisciplinarity, and to intellectual diversity as a structural feature of the inquiry rather than an ornament.
Members have been selected on the criterion that their work cannot be fully located within a single discipline, and that their presence in the same room will generate productive friction rather than comfortable agreement. The working group is not a panel of experts assembled to endorse a predetermined conclusion. It is an intellectual community convened to work on a genuinely difficult set of questions — and to produce, from that work, insights that none of its members could have reached alone.
The working group runs from June to December 2026 and is organized around three workstreams: Provocations, the conceptual work carried by the core members; Synthesis, led by the organizers, drawing concepts and tensions across the group; and Field Conditions, an organizer-led inquiry into what it would take to carry the agenda forward.
The core group consists of up to eight scholars, practitioners, and writers who commit fully to the work. Each core member receives a stipend of €10,000 and takes on a substantive intellectual obligation: three connected essays developing one Provocation — a directional claim developed through the arc of Grounding, Visioning, and Demonstrating. These are not conference papers or preliminary sketches. They are arguments developed in sustained engagement with a shared question — refined through dialogue with fellow core members and the collective discussion of each other's work. Core members participate in fortnightly webinars at which work in progress is presented, discussed, and pushed further. The essays accumulate; they build on each other and, taken together, constitute a contribution larger than any individual piece.
This work feeds directly into a broader conceptual project: a collaboratively developed, GitHub-based theory of Informational Democracy — open, versioned, and designed for iterative refinement, in the spirit of Audrey Tang and Glen Weyl's "Plurality". The working group's Substack, Informational Democracy, is the public knowledge layer of the project, organized around four formats: Provocations (the core members' essays), Synthesis Notes (organizer reflections after each phase), Field Notes (shorter pieces on ecosystem, funding, policy, and continuation), and Framework Pieces (the integrative writing that consolidates the emerging framework). Together they build a public intellectual presence for the questions the group is working on as the inquiry unfolds.
Synthesis is the organizers' central responsibility. After each essay round they draw together the concepts, tensions, and design principles emerging across the group into a synthesis note, building toward a provisional framework for Informational Democracy at the close of the cycle. The framework is not imposed at the outset; it emerges from the collective work. In parallel, the Field Conditions workstream asks what it would take to carry the agenda forward beyond December — ecosystem, funding models, policy awareness, and institutional vehicles — pursued by the organizers through a dedicated session and bilateral exchanges rather than added to the core members' load. The cycle closes with a two-day in-person synthesis convening in Berlin and a public event.
The working group is not convened to produce scholarship for scholars. The intellectual work is real and the academic ambition genuine — but the test of success is not citation counts or journal placements. It is whether the ideas developed here find their way into the hands of people positioned to act on them, and whether they seed the conditions for a field that will outlast this first cycle of work.
This is a theory of change as much as a research program — and the theory has a longer arc than six months. The working group's immediate contribution is agenda-setting: establishing the Provocations worth arguing about, building the intellectual community capable of arguing about them rigorously, and producing a body of work — essays, a collaboratively developed open theory, a public Substack presence — that can travel across the boundaries between disciplines and between sectors. Its longer ambition is policy: concrete proposals for the redesign of democratic institutions as informational regimes, developed in sufficient depth and with sufficient institutional grounding to be picked up by actors in government, technology, and civil society. That ambition will not be fulfilled in six months. But the agenda-setting work done here — the conceptual foundations laid, the networks established, the vocabulary developed — is what makes it possible.
What is needed, and what this working group is designed to seed, is a community of practice spanning research, policy, technology, and civil society that shares a common analytical framework and a common orientation toward institutional design. That community does not yet fully exist in the form this project requires. Part of the working group's purpose is to bring it into being — in Germany, across Europe, and in the broader international conversation about democratic renewal in the age of AI.
Three audiences matter in particular, and they require different things from this work.
Policy makers are navigating the informational crisis largely without adequate conceptual tools. The frameworks available to them — media regulation, competition policy, platform governance — were designed for an earlier informational moment and fit the present badly. What the working group offers is not policy prescriptions but something prior and more durable: a reframing of the problem itself, and the beginnings of a design vocabulary for institutions that understand the crisis as one of epistemic trust and informational architecture rather than of content and misinformation. The essays and the collaborative theory developed through this project are conceived to be legible to serious policy practitioners, not only to academic specialists — and to prepare the ground for the policy proposals that a sustained second phase of the work will develop.
Designers, engineers, platform architects, and AI developers possess the technical capacity to build differently but frequently lack the normative and institutional frameworks to know what building differently would require. The working group does not offer technical solutions. It offers something the technology community urgently needs: a principled account of what democratic informational infrastructure should do, what properties it should have, and what trade-offs it should refuse. Design without normative grounding reproduces the pathologies it intends to solve. The working group's output — including the open, versioned theory of Informational Democracy developed through its GitHub-based collaborative project — is conceived as a resource for practitioners who want their work to be answerable to democratic values rather than merely to commercial or technical imperatives.
The organizations, advocates, journalists, educators, and civic technologists working to sustain democratic life in conditions of informational complexity need legitimation as much as they need analysis. The arguments civil society makes are often dismissed as insufficiently rigorous, parochial, or merely reactive. The working group offers a theoretical foundation for claims that civil society actors are already making in practice: that information is a public good, that epistemic trust is an infrastructure, that complexity requires institutional redesign rather than institutional restoration. By publishing its developing arguments through the Substack Informational Democracy as the work unfolds, the working group builds a public intellectual presence that civil society actors can draw on — not as authority to defer to, but as intellectual resources to deploy.
These three audiences are not separate. The most generative conversations this project can produce are those that cross between them — between the policy practitioner who understands institutional constraints and the technologist who understands design possibilities; between the civil society advocate who understands what democratic life requires on the ground and the researcher who can translate that into theoretical form. Bringing those conversations into being, and giving them a shared conceptual infrastructure to work with, is one of the working group's most important contributions.
The working group proceeds from a conviction: that the complexity of contemporary democratic life is not a temporary condition to be managed until stability returns. It is the permanent condition within which democratic institutions must learn to operate. The question is whether those institutions are designed to fail gracefully and learn — or to become increasingly brittle as the gap between their operating assumptions and the world they must govern continues to widen.
That gap is widening faster than democratic theory has acknowledged. The arrival of agentic AI is not the next disruption after platform media. It is a qualitative shift in the conditions of democratic life — one that makes the informational crises of the past decade look, in retrospect, like a warning that was not adequately heeded. Democratic institutions that were not redesigned to handle algorithmic fragmentation are now being asked to navigate autonomous agents fabricating consensus, synthetic participation at scale, and epistemic infrastructure owned by a vanishingly small number of private actors with no democratic accountability. The window for deliberate design is open. It will not remain so indefinitely.
Lippmann was right that the problem of complexity and democratic governance was the central political problem of modern life. He was wrong that it was unsolvable within democratic terms. Dewey was right that institutional redesign was the answer. He could not yet say what that redesign required. We are now, for the first time, in a position to begin saying — not because the answers are clear, but because the theoretical tools, the empirical knowledge, and the practical examples of democratic innovation are available in a way they were not before. What has been missing is the sustained, interdisciplinary, design-oriented intellectual community capable of drawing them together. That is what this working group is convened to build.
The horizon of this work is a new generation of democratic institutions: institutions that treat complexity as a design condition rather than a threat; that build epistemic trust as infrastructure rather than assuming it as a given; that embed adaptive capacity rather than optimizing for a stability that is no longer achievable; and that are designed from the outset to govern the informational architectures of AI-saturated democratic life rather than to be governed by them. Institutions, in other words, that are designed for the world as it actually is — and for the world it is becoming.
This working group is the first act. The field it builds will need to sustain itself beyond any single year, any single institution, and any single set of Provocations. The essays written here, the networks established, the vocabulary developed, the open theory published — these are not endpoints. They are foundations. The work we are convening to do is larger than what six months can contain. That is not a limitation. It is the point.
The institutions of tomorrow will not be found by protecting what we have inherited. They will be designed — by people willing to think carefully, act ambitiously, and build patiently across the boundaries that have kept the necessary conversations apart.
This is the work we are convening to do.\ \ \ Georg Diez, Lead\ Steve Vertovec, Lead\ Felix Beer, Strategist
Document prepared: April 2026 \| Program: Informational Democracy Working Group \| Max Planck Institute for Political and Social Science, Göttingen
Georg Diez, Lead · Steven Vertovec, Lead · Felix Beer, Strategist