A shorter, evolving statement of the working group's premise — the moment, the frame, the ambition. The fuller argument is developed in the Concept Note.
Democratic theory has a history of reckoning with moments when the informational conditions of self-government change faster than the institutions built to rely on them. In 1922, Walter Lippmann diagnosed the gap between the world and the pictures citizens form of it, and concluded that the fully informed citizen was a fiction democratic governance could no longer assume. His remedy was technocratic: experts must govern because citizens cannot. John Dewey took the same diagnosis and rejected the remedy. The problem was real, but the answer was institutional redesign — new structures of association and communication adequate to a "Great Society" too vast for any citizen to comprehend directly. That debate was never resolved. It was superseded, for a few decades, by a working — if imperfect — epistemic infrastructure: broadcast media, mass education, professional journalism, channels that most people trusted most of the time.
The internet dissolved that infrastructure, and did so in two waves. The first was structural: gatekeepers collapsed, and anyone could publish. The second was algorithmic: platforms built for engagement rather than deliberation replaced the collapsed gatekeepers with an epistemic infrastructure optimized for attention capture, emotional resonance, and tribal reinforcement. What emerged was not a richer public sphere but a fragmented one — divergent epistemic territories, each with its own sources and its own version of events. Artificial intelligence now arrives as a third wave, and a qualitatively different one: not just more information, or differently mediated information, but information whose origin, authorship, and relationship to truth can no longer be assumed. At the same time, AI is the first genuinely new tool democratic institutions have had in a generation for handling complexity at scale — collective sense-making, synthesis across diverse perspectives, distributed intelligence. Whether it deepens democratic capacity or hollows it out further is not determined. It is a design question, and one this working group takes as its starting point.
For sixty years, the dominant framework for thinking about democracy and information has been Jürgen Habermas's public sphere: citizens engaging in rational-critical debate within a space between the state and the private individual, holding power accountable through the force of the better argument. It was a genuine and generative account of the conditions Habermas described. But those conditions — a relatively legible public, a shared informational commons, scarcity that made channelling both possible and necessary — no longer hold, and an academic infrastructure built around recovering them has outlived the world it was built for. This working group does not propose to refute Habermas. It proceeds from a different starting point entirely: rather than ask how to restore the conditions for rational-critical deliberation, it asks what informational and infrastructural conditions make any form of democratic self-governance possible at all. That is a shift from communication to infrastructure as the primary object of inquiry — not a claim that communication does not matter, but a wager that the more urgent and more neglected question, in an age of algorithmic mediation and artificial intelligence, is what lies beneath communication and makes it possible or forecloses it in the first place.
The working group's organizing move is to treat democracy as a social technology: a designed system for collective sense-making, the distribution of power, and the resolution of conflict, built and maintained through infrastructure rather than sustained by argument alone. On this account, elections, parliaments, courts, and the press are not democracy itself but a historically particular set of institutional answers to a more basic question: how does a plural society generate the shared judgments it needs to govern itself? That question can be asked again, under conditions those institutions were never designed for.
The metaphor is deliberately provocative, and its limitations are worth stating plainly rather than assuming away. Treating democracy as a technology risks a technocratic reduction: a framing so easily captured by engineers and platform designers that the political and economic questions — who owns the infrastructure, whose interests it serves, what forms of power it entrenches — get treated as implementation details rather than the heart of the matter. It risks technological determinism: the assumption that better systems produce better politics, when the history of the last two decades argues the opposite just as often. And it risks losing sight of what a purely infrastructural lens leaves out — democracy as a way of life, a set of civic dispositions and relationships that no architecture, however well designed, can substitute for.
The working group takes these risks seriously rather than resolving them in advance. Democracy is also, irreducibly, a set of normative commitments about legitimacy, freedom, and collective self-rule — and the group includes members who hold different, sometimes conflicting, views on what a good democratic order looks like: consensus-oriented and agonistic, liberal and communitarian, technologically optimistic and structurally skeptical. The working group does not adjudicate between these positions. It asks a prior question that cuts across all of them: whatever form of democracy one is committed to, what informational and infrastructural conditions does it require in order to function at all? Political economy is inseparable from this question — ownership and control of the infrastructure are not secondary to the inquiry but central to it. The social-technology frame is offered as a genuinely productive lens for asking that question at the infrastructural level others have left unexamined, not as a theory that displaces the normative disagreements democratic theory has always had to hold.
This is a first act of field-building, and field-building is not the same undertaking as producing research. Research generates knowledge; field-building generates the conditions under which knowledge can accumulate, compound, and travel — a shared vocabulary, a set of provocations worth arguing about, a community capable of sustaining the inquiry across institutions and years. Six months is enough to take the constitutive steps. It is not enough to complete them, and this working group does not claim otherwise.
The group brings together scholars and practitioners working at the intersection of democratic theory, AI governance, political economy, complexity science, and institutional design — a deliberately cross-disciplinary composition, because the question the group is organized around does not belong to any one of those fields alone. Each member is developing an independent Provocation: not a hypothesis to be tested and not a research question to be explored, but an assertive, directional claim about what is failing in democratic infrastructure, what is possible, and what institutions must become — staked in order to be argued with, not hedged in order to be defensible. The range of what members are arguing is wide by design, from the redesign of specific democratic institutions and the architecture of civic AI agents, to the political economy of attention, to the infrastructural conditions of professional and epistemic trust. The working group does not ask its members to converge. It asks them to develop their sharpest position and hold it in productive tension with the others.
That work unfolds across three connected essay phases — Grounding, Visioning, Demonstrating — moving from diagnosis to institutional imagination to concrete demonstration, and across a fortnightly public webinar series that alternates between Project Clinics, where members' work is put under sustained scrutiny, and Synthesis Workshops, where the group works across its own disagreements toward shared formulations. The Provocations, and the essays and public writing that grow out of them, are published as they develop on the Informational Democracy Substack — a public knowledge layer built in the open, not released only at the end of the process. The working group's cycle concludes with an in-person synthesis convening in Berlin in December 2026, where the group's individual lines of argument are worked, together, into whatever shared formulations the process has actually earned — not a predetermined consensus, but the state of the argument the group has arrived at.
The purpose of this first cycle is agenda-setting: to draw the right conversations together, to establish the provocations worth arguing about across the fields this project sits between, and to build the intellectual community within which a new domain of inquiry can begin to grow. The networks this working group establishes — among scholars, practitioners, technologists, and policy actors in Germany, across Europe, and in the wider international conversation about democratic renewal — are as much a part of this project as the essays are. A field without a community sustaining it is not a field, and building that community is one of this working group's primary purposes.
The longer horizon is implementation: institutional and policy proposals concrete enough, and grounded enough, to be picked up by actors in government, technology, and civil society who are already confronting these questions without the conceptual infrastructure to address them systematically. Those proposals will not emerge fully formed from six months of work. But the agenda set here, the networks established, and the conceptual foundations laid are what make them possible in the cycles that follow — and the working group's ambition for what comes after this first cycle is to remain a live, self-sustaining node for this inquiry, not a single fellowship that concludes when its stipends do.
This project is also deliberately positioned. The dominant architectures now governing digital information — an extractive, engagement-optimized platform economy in the United States, a state-directed infrastructure of surveillance and control in China — are routinely presented as opposites, but both concentrate informational power and are structurally indifferent to the epistemic conditions democratic self-governance requires. Europe's legal and regulatory architecture, imperfect as it is, has created an institutional commitment to governing information in the public interest that has no real equivalent elsewhere at scale. This working group works from that position, independent of direct affiliation with any AI laboratory, and treats the development of a distinctly European perspective on informational democracy — neither extractive nor authoritarian — as part of its task, not an incidental feature of where it happens to sit.
Read the full Concept Note →