Towards a new democratic culture

This document grew out of a simple but deep intuition: that the word “democracy” has lost its original meaning, and that, paradoxically, the technology that dominates our lives today could be the tool for getting it back. It is not a political manifesto, a technical paper or a draft law. It is, above all, a story — the story of how a revolutionary idea, born twenty-five centuries ago, was gradually emptied of its content, and of how we might fill it again.

We live in societies that call themselves democratic, yet the vast majority of their citizens have never taken part in a single collective decision beyond dropping a ballot paper into a box every four years. We live in systems where we hand our voice to professional politicians who, all too often, spend their terms preparing the next election campaign instead of governing. We live, in short, in what Benjamin Barber called “thin democracies”: structures that look like popular sovereignty but whose substance is more hollow every year.

But we also live at a unique moment in history. For the first time, we have tools that would let millions of people deliberate, propose, argue and decide together without having to gather on a hillside. The question is no longer whether a more direct, more participatory democracy is technically possible. The question is whether we have the collective will to build it, and whether we can design it so that it avoids the mistakes of the past without repeating the exclusions of the present.

The pages that follow try to answer that question. They do so by way of a journey that starts in Athens in the fifth century BC, passes through the American and French revolutions, crosses the contemporary crisis of political representation, and arrives at a concrete proposal: a collective intelligence aggregator for citizen participation. A platform that does not claim to replace existing institutions or to decide in their name, but to create a continuous layer of participation that gathers scattered knowledge, cuts through the noise and brings clear, representative proposals to the surface.

What you will not find here is the technical part. That will come in a second document. Here we deal with the “why” and the “what”; there, the “how”. Because no tool is any use until we understand which problem it is trying to solve, and which view of the world stands behind it.

The Athenian miracle

In the fifth century BC, in a city-state in the eastern Mediterranean with little more than 250,000 inhabitants, an idea appeared that had no precedent in human history: that decisions affecting a community should be made by the people who have to live with them. The Athenians gave that idea a name that everyone knows today and almost no one understands in its original sense: demokratia, literally “the power of the people”.

That first democracy worked in ways radically different from what we now associate with the word. There were no political parties, no election campaigns, no professional politicians. The system rested on three main institutions, each designed to make sure that power never piled up in the same hands.

The Ekklesia, the popular assembly, was the supreme body. It met at least two or three times a month on the hill of the Pnyx, a natural amphitheatre with room for some six thousand people. Anyone present could ask to speak, put forward a proposal, argue for or against a decision. Votes were taken by show of hands, and the majority won. But that sovereignty was not unlimited: no question was put to the vote unless the Council had prepared it beforehand, and anyone could take to the courts a decree they thought contrary to the laws (the graphe paranomon); if the charge succeeded, the decree was annulled and whoever had proposed it risked a fine. Within those limits, the assembly decided on legislation, taxes, peace treaties, declarations of war and the election of certain public officers.

The Boule, or Council of Five Hundred, was the executive and administrative body. Its members were not elected but drawn by lot from the ten Athenian tribes — fifty from each. The Council prepared the Assembly’s agenda, managed public finances and oversaw the day-to-day running of the administration. No one could serve on it more than twice in a lifetime, a restriction designed explicitly to stop power from becoming a profession.

The courts, or dikasteria, worked on a similar logic. A body of six thousand people, chosen by lot every year, administered justice. To rule out any form of corruption or influence, the Athenians used a mechanical device called the kleroterion — a stone slab with slots for name tablets and a tube down which black and white dice fell, randomly assigning jurors to the day’s cases. Jurors did not know which case they would try until the last moment; the accused did not know who would judge them. For its time, it was a remarkably sophisticated system.

Isegoria: the forgotten right

We talk constantly about demokratia, but in Athens they had a second concept that they thought just as important, or even more so: isegoria — everyone’s equal right to speak in the assembly. Being able to vote was not enough; what mattered was being able to speak and to be listened to. Isegoria was not merely a passive right to be present. It was the active recognition that every citizen had a legitimate perspective on common affairs, and that no perspective was inherently superior to another.

When Herodotus wanted to praise the Athenian system, he did not talk about demokratia but about isegoria. What made Athens different, for him, was not so much that the people voted — other peoples had held assemblies — but that anyone, whatever their origin, wealth or education, could climb onto the bema (the stone from which speakers addressed the assembly) and address everyone else as an equal.

This distinction is crucial to what we will propose later in this document. A citizen participation platform is not, at heart, a voting machine or a parallel authority. It is a tool of isegoria and a collective intelligence aggregator: it makes sure everyone’s voice can be heard, helps distil the useful core of what people contribute and makes consensus visible. The final institutional decision remains with the competent authority, unless that authority has made a specific commitment.

Why Bema

On the Pnyx you can still see, cut into the rock, the platform from which speakers addressed the assembly. In Greek it was called the bema. It was not a throne, nor a rostrum reserved for a few: it was a stone step that any citizen who wished to speak could climb. From there, laws were proposed, peace treaties defended and abuses denounced before thousands of people.

The platform we propose takes its name from here. Bema wants to be that stone, carried into the twenty-first century: a place where any resident of the municipality can make a proposal, where a proposal is judged by what it says and not by who says it, and where the system guarantees what Athens guaranteed at its best — that everyone’s voice has the same right to be heard. With one essential difference: this time the stone must belong to everyone, without the exclusions that shut out most of the Athenian population.

The lottery as a democratic pillar

One of the most distinctive — and most forgotten — features of Athenian democracy was the systematic use of the lottery (kleros) to fill the great majority of public offices. It was neither a minor device nor a stopgap: it was the heart of the system, the piece that Athens considered truly democratic.

The reasoning was relentless. If you choose who governs by election, you inevitably choose the most eloquent, the richest, the best known or the best connected. By its very nature, election discriminates in favour of an elite, whatever the definition of elite at any given moment. The lottery, by contrast, levels: it gives every citizen exactly the same chance of reaching power, regardless of wealth, eloquence or connections. In Athens, election was an aristocratic mechanism (from aristos, “the best”: choosing the best); the lottery was the truly democratic one.

This conviction was not naive. The Athenians knew perfectly well that not everyone is equally competent at every task. For offices that required specific technical skills — above all the military command of the strategoi — they kept election by vote. But for the great majority of government functions, the lottery was considered not just adequate but superior, because it guaranteed something that election never can: that power would circulate.

Ostracism: the self-defence mechanism

The assembly held an extraordinary power that few later systems have dared to reproduce: the ability to temporarily expel any citizen judged to be a threat to democracy. This mechanism, known as ostracism, worked in two stages. First, the assembly voted on whether one should be held that year; if so, each voter wrote the name of the person they wanted exiled on a fragment of pottery (an ostrakon). If the quorum was reached, usually put at six thousand votes, the person with the most votes had to leave Athens for ten years — without losing their property or their rights as a citizen.

Ostracism was, essentially, a preventive mechanism against tyranny. Nobody had to have committed a crime; it was enough for the community to sense that someone’s power or influence was potentially dangerous to the system. It was an extreme tool, certainly, but it revealed a very clear awareness that democracy is a fragile system that needs active mechanisms of self-protection.

The shadows of the Athenian model

It would be dishonest to idealise Athenian democracy without pointing out its deep shadows. Participation was limited to free adult men who were citizens: women were excluded, as were enslaved people — a very substantial part of the population — and the resident foreign population, the metics. It is estimated that those with the right to take part never exceeded thirty per cent of the total adult population. At its height, Athens may have had as many as sixty thousand people entitled to vote, out of a far larger total population.

That exclusion was neither a minor detail nor a flaw in the system; it was a fundamental contradiction that Athenian democratic rhetoric never resolved. A democracy built on slavery is, at best, a partial democracy. And yet the mechanisms the Athenians invented — the lottery, rotation, the deliberative assembly, ostracism — remain tools of remarkable conceptual power, and their absence from modern systems is due not to obsolescence but to very deliberate political decisions, as we shall see.

What the founders really wanted

One of the most persistent and consequential misunderstandings in modern political history is the belief that the founders of today’s representative systems meant to create democracies. They did not. They knew it, they said so openly, and the historical record is abundant and unambiguous.

James Madison, in Federalist No. 10 (1787), drew an explicit line between a “democracy” — the Greek system of direct participation, which he considered dangerous, unstable and vulnerable to the passions of the crowd — and a “republic”, the representative system he advocated as the superior solution. For Madison, representation was not an imperfect substitute for democracy but a deliberate improvement: a filter that would “refine and enlarge the public views” by passing them “through the medium of a chosen body of citizens”, an enlightened governing class able to make rational decisions on the people’s behalf.

The Abbé Sieyès, one of the architects of the French Revolution, was more explicit still. He argued that citizens should choose representatives because most people “have neither the time nor the education needed to concern themselves directly with public affairs”. Representation was, in his view, a rational division of labour: just as those who till the soil do not make their own tools, citizens should not make their own laws. The parallel is revealing: politics was conceived as a trade for specialists, not as a universal right of participation.

Elections as an aristocratic mechanism

The French political scientist Bernard Manin, in his landmark Principes du gouvernement représentatif (1995), showed with scholarly rigour what many people had sensed: that elections are an aristocratic mechanism by design, not by accident. The argument has a devastating elegance.

Etymologically, “elect” and “elite” share a Latin root: eligere, to pick out, to select the best. When we elect, by definition we select someone we consider superior to ourselves in some respect — eloquence, presence, charisma, communication skills or, simply, the resources to make themselves visible. By its very mechanics, election favours people who stand out; that is, elites. Not necessarily elites of merit, but elites of visibility.

Manin contrasts this with the lottery, which in ancient Greece was regarded as the truly democratic mechanism. The lottery gives every person exactly the same chance of exercising power, with no advantage for eloquence, wealth or connections. In Athens the difference was clear: election was aristocracy (rule by the best), the lottery was democracy (rule by the people). Aristotle put it without ambiguity: “it is considered democratic for offices to be filled by lot, and oligarchic for them to be elected” (Politics, IV.9, 1294b). Twenty centuries later, Montesquieu took up the point: “Suffrage by lot is natural to democracy, as that by choice is to aristocracy” (The Spirit of the Laws, II.2).

The founders of the modern systems — Madison, Hamilton, Sieyès, Montesquieu — knew Aristotle perfectly well. They chose elections in full awareness that this was a filter against direct popular participation. They did not do so out of ignorance or technical limitations; they did so because they distrusted the people and wanted a system in which power lay in the hands of a selected governing class, even if it was nominally legitimised by the popular vote.

The appropriation of the word

If the founders did not want to create democracies, why do we call the systems they created “democracies” today? The answer is fascinating, and deeply political.

It was Alexander Hamilton who first used the term “representative democracy”, in 1777, in a letter about the new constitution of the state of New York. The combination was new and, in a way, oxymoronic: to the ancient Greeks a “representative democracy” would have been a contradiction in terms, like speaking of a “square circle”. But the term turned out to be a rhetorical masterstroke. It allowed the new governments to inherit the symbolic prestige of the Greek concept — the power of the people — without taking on its substance — the direct participation of the people.

After the American and French revolutions of the eighteenth century, and especially as the franchise widened in the nineteenth and twentieth, the word “democracy” took on a positive connotation so powerful that every political system wanted to be associated with it. Representative governments adopted the term to give themselves popular legitimacy. They were states too large and too complex for direct democracy — or so they argued — but they wanted the prestige of the Greek concept all the same.

Jean-Jacques Rousseau, with prophetic clarity, had already warned that a system in which the people delegate power to representatives ceases to be a true democracy: “The people of England regards itself as free; but it is grossly mistaken; it is free only during the election of members of parliament. As soon as they are elected, slavery overtakes it, and it is nothing.” Two and a half centuries later, Rousseau’s words ring with an uncomfortable relevance.

The Roman Republic, not Athens

Strictly speaking, what we call “democracy” today resembles the Roman Republic far more than Athenian democracy. Rome had elected magistrates, a senate with real power, popular assemblies with limited functions and a complex system of checks and balances designed to prevent the concentration of power. But effective power always lay in the hands of a ruling class — first the patricians, then the optimates — who governed in the people’s name without the people actually governing.

The fundamental difference is the direction of power. In Athens, power flowed from the assembly and was delegated temporarily — by lot — to executive office-holders. In Rome, and in modern systems, power flows nominally from the people but is permanently concentrated in a class of representatives who, once elected, act with almost total autonomy until the next election. The people have the power to choose who governs, but not the power to govern.

“Democratic fatigue” is the wrong diagnosis

One of the most repeated commonplaces of contemporary political analysis is that citizens are “tired of politics” or “fatigued by participation”. Rising abstention, disaffection with parties and widespread distrust of institutions are presented as proof of an apathetic society that prefers to delegate and switch off.

Empirical research says something very different. Citizens are not tired of participating; they are tired of participating without impact. When participation is real and binding — as in Ireland’s citizens’ assemblies, Swiss referendums or the participatory budgets of Porto Alegre — involvement rises significantly. The problem is not that people do not want to take part; it is that they have learned, through repeated experience, that their participation makes no difference.

Ireland offers one of the most telling examples. The country tackled two of its most divisive social questions — the legalisation of same-sex marriage and the reform of abortion law — through citizens’ assemblies made up largely of ordinary people picked at random. Those assemblies deliberated for months, heard from experts, argued among themselves and arrived at recommendations that were later put to a popular referendum. The result was overwhelming: proposals that no professional politician would have dared put on the table won massive public support.

The Irish example shows two things. First, that ordinary people are perfectly capable of making informed, nuanced decisions on complex questions when they are given the information, the time and the space to deliberate. Second, that “democratic fatigue” is not a disease of the citizenry but a symptom of a disease in the system: people disconnect because the system disconnected them first.

Barber’s “thin democracy”

The political philosopher Benjamin Barber proposed a distinction that is extraordinarily useful for understanding the present moment. He set “thin” democracy against “strong” democracy, which in this document we call “thick”, as the two poles of a spectrum.

Thin democracy is the one we live in: voting every four years, delegating to representatives and getting on with our private affairs until the next election. Political participation shrinks to a single act — the vote — and for the rest of the time citizens watch the political game from the stands. Thin democracy runs on the assumption that politics is a specialised field, like medicine or engineering, and that citizens neither need nor want to get involved beyond the bare minimum.

Thick democracy, by contrast, means continuous participation, regular deliberation and shared responsibility for collective decisions. It does not ask every citizen to be a political expert — that would be absurd and undesirable — but for spaces, mechanisms and habits of participation to exist so that anyone who wants to get involved can do so in a meaningful way. Thick democracy does not replace representative institutions; it complements them with a layer of participation that feeds them collective intelligence and holds them to account continuously, not only every four years.

What we propose in the following parts of this document is, explicitly, a transition from thin democracy to thick democracy. Not an instant revolution — which is neither possible nor desirable — but the gradual incorporation of elements of direct participation that, little by little, build a new democratic culture. A culture in which participation is a habit, not an exception; a right exercised, not a right on paper.

The party system: from representation to electoral machine

Political parties were born as instruments for aggregating social interests. Their original function was to identify the needs and aspirations of broad sections of the population, turn them into coherent political programmes and fight to implement them from within the institutions. For decades, and for all their imperfections, they performed that function with reasonable effectiveness.

But at some point — hard to date precisely, though the literature places it in the last decades of the twentieth century — parties stopped being instruments of representation and became electoral machines. Their primary function shifted from “governing well” to “winning elections”. Internal organisation turned towards vote-winning, strategic communication and the building of electoral narratives. Terms of office became periods of preparation for the next campaign. Politics stopped being an exercise in government and became a permanent contest for power.

The result is what we live with: four years spent plotting ways to win elections, no years of real government. Parties that define themselves more by what they reject in the other camp than by what they propose. Public debates that look more like spectacles of confrontation than deliberations about the common good. And citizens who, understandably, feel more and more like strangers to a game that they sense — rightly — is not being played for them.

There is, however, an important nuance that it would be irresponsible to ignore. Parties still perform a function that nobody has managed to replace: organising political conflict peacefully. Societies have opposing interests, incompatible world views and real conflicts that need institutional channels. Without some mechanism for aggregating interests, a purely direct democracy could splinter into a thousand micro-conflicts with no possibility of synthesis or resolution. Any proposal for democratic reform has to take this function into account and offer alternative mechanisms that replace it or complement it.

Depolarisation as an urgent need

One of the most worrying phenomena in contemporary politics is affective polarisation: the growing tendency to define ourselves politically not by what we believe but by what we reject on the other side. Research in political psychology shows that in recent decades, and very markedly in the United States, hostility towards the “other side” has grown much faster than attachment to one’s own. We do not love our own ideas more; we hate other people’s more. In Europe the trend varies more from country to country, but the incentives that feed it are just as present.

This polarisation is neither natural nor inevitable; it is, in large part, a product of the design of today’s political and media systems. Parties have incentives to polarise because confrontation mobilises more than collaboration. The media and social networks amplify conflict because it draws more attention than consensus. And citizens, immersed in this ecosystem, end up identifying their political positions with their personal identity, which makes rational disagreement almost impossible.

There is, though, an empirically proven antidote: “blind” proposals. Studies in political science have shown consistently that when people assess concrete policy proposals without knowing which party they come from, cross-party support rises significantly. Proposals that right-wing voters would instinctively reject if they knew they came from a left-wing party — or vice versa — win broad support when judged purely on their content. The party label acts as a cognitive filter that prevents rational evaluation.

This finding is no small matter: it suggests that a very significant share of political polarisation is artificial, built on tribal identities more than on real, substantive disagreement. If we can design spaces where ideas are judged on their content and not on their colour, the consensus that emerges could be far wider than party politics would have us believe.

The incremental strategy

The vision we set out in this document is not that of a revolution in the political system. Revolutions, by definition, mean a break with the existing order, and breaks provoke proportionate resistance. History teaches that the most lasting systemic changes do not come from frontal confrontation but from gradual erosion from within: building habits, generating evidence and growing a new culture until the old order becomes unsustainable.

The strategy is therefore incremental: to introduce, bit by bit and starting at the local level, elements of direct democracy into the everyday life of communities. Not to ask the system’s permission to change it, but to create practices that show another model is possible and, above all, that it is better. When people experience for the first time that their opinion has shaped a real decision that affects them — a street, a budget, a public service — they do not forget it. Participation creates a participatory identity, and a participatory identity creates demand for more participation. Once that circle is under way, it is very hard to stop.

This is precisely what political science calls “incremental paradigm change”: not waiting for the system to change from above, but building from below a parallel reality that shows, with facts, that citizens are capable of far more than the current system allows them. When that reality is visible enough and successful enough, the pressure to bring it into the formal system will be irresistible.

The deliberative turn

While parties were turning into electoral machines, one strand of political philosophy began looking for another way out. In the late twentieth century, what has come to be called the “deliberative turn” shifted the central question of democracy: it was no longer only a matter of who votes, but of how opinions are formed before they are cast.

Jürgen Habermas argued that a political decision is legitimate when it arises from public deliberation among free and equal people, where the only thing that counts is what he called “the unforced force of the better argument”, not a mere tally of the preferences we already held before we started talking. Carole Pateman had added a complementary idea in Participation and Democratic Theory (1970): participation educates. The more people decide on what affects them, the better they get at it.

James Fishkin turned these ideas into an experiment. In 1988 he proposed the deliberative poll, and in 1994 the first one was held, in the United Kingdom: gather a sample of randomly chosen people, give them balanced information and time to debate, and measure how their opinions change. The result has been repeated since in dozens of countries: people do not become more polarised; they become better informed, more nuanced, and better at understanding the other side’s reasons.

Hélène Landemore went further. Drawing on studies of cognitive diversity, she argued that a diverse group of ordinary people often decides better than a small group of experts — not because each person knows more, but because together they see more angles. It is, in large part, Aristotle’s intuition: the many, when they come together, can be better than the few.

This is the theoretical frame of the platform we propose. The lottery is not just a tool of equality; it is also the simplest way of bringing together the diversity that makes a group think well. And deliberation is not a luxury: it is what turns a tally of preferences into a decision that can be justified to everyone.

Local first, global later

The “local first” approach is not only a question of scale; it is a question of knowledge. Residents know their immediate surroundings far better than any political or technical officeholder. They know where a pedestrian crossing is dangerous, which square needs shade, which municipal service is not working. This “distributed intelligence” is an enormously valuable resource that no centralised system can reproduce.

The most successful experiences in citizen participation bear this out. Tolosa (Gipuzkoa), for example, has run citizens’ assemblies that the OECD recognises as exemplary practice in deliberative democracy, and is now working on a permanent citizens’ assembly with digital participation tools. Porto Alegre showed for years that participatory budgets are not only viable but can steer investment towards the neighbourhoods that need it most. It is also a lesson in how fragile such gains are: the process weakened with changes of government until it was practically suspended around 2017. And dozens of municipalities around the world have found that when people take part in decisions that directly affect them, the quality of decisions improves and trust in institutions grows.

Starting local also has a fundamental strategic advantage: it lets you make mistakes in a setting where the consequences are manageable. A participatory platform that does not work well in a neighbourhood of five thousand people can be corrected quickly; one that fails at national scale can discredit the idea for an entire generation.

Elinor Ostrom, the first woman to receive the Nobel Prize in Economics (2009), showed this through decades of fieldwork. Fishing, irrigation and grazing communities around the world — including the irrigation systems of the Valencian huerta — have managed common resources for centuries, often better than designs imposed from above, because those close to a problem know it better and live with its consequences. Her conclusion makes a good compass for this project: “a core goal of public policy should be to facilitate the development of institutions that bring out the best in humans.”

That is why the first pilot begins not in a country or a big city but in a town: Cardedeu.

Technology as infrastructure, not as the solution

At this point we need an honest reflection on the limits of technology. Evgeny Morozov has spent years warning about what he calls “technological solutionism”: the belief that any social problem can be solved with an app. A digital platform does not, on its own, create democratic culture. It can make it easier, amplify it and make it scalable, but it cannot replace it.

Democratic culture is built from real social practices: face-to-face debates, places to meet, civic education and shared experiences of deliberation. A purely digital platform risks reproducing the dynamics of social media, which pile up polarisation, superficiality and the viral spread of conflict. Bema’s proposal is deliberately the opposite: to aggregate collective intelligence, separate signal from noise and keep the core of the ideas that add value.

So the vision we set out here treats technology as infrastructure — a channel, a tool, an enabler — but not as the solution itself. The platform has to come with in-person activities, with training, with spaces for face-to-face debate. Technology makes what is local scalable and what is fleeting lasting; but the content, the substance, the culture have to come from the human community that uses it.

Put another way: a good platform in a community with no culture of participation will achieve little. But a community with a culture of participation and no good platform will be limited to the scale of what can be done face to face. It is the combination of the two that can be genuinely transformative.

The pages that follow describe, in functional rather than technical terms, the design of a citizen participation platform that builds in the principles discussed in the earlier parts. Every element of the design answers a specific problem identified in the preceding analysis; none is arbitrary or decorative.

The proposal system

Monthly quota and enforced reflection

Everyone registered on the platform has one proposal per month. This quota is the system’s first quality filter and does several jobs at once: it forces people to think before proposing (if you can only make one, you think it through), it disarms troll behaviour by design, with no need for censorship or prior moderation, and it balances the weight of a very active person against that of a less active one.

The quota is partially cumulative: if you do not use your proposal one month, you can have two the next, but never more than two banked. This stops anyone from stockpiling proposing power indefinitely, yet does not penalise someone who was travelling, ill, or simply had nothing to propose that month.

Quota transfer: the “proposal proxy”

One of the system’s most innovative mechanisms is the possibility of handing your monthly quota to someone else. This lets a person with a good idea but no quota left receive the backing of others who trust them. Transfer creates an organic form of real social reputation — not an algorithm that scores people, but people who willingly give up their turn to support someone else’s proposal.

Transfer also produces a valuable signal for moderators: a proposal with quotas from five people behind it has already passed an informal social filter. It is not the same for someone to put forward a proposal alone as to do so with the explicit backing of five people who have given up their own quota to support it.

To stop influencer figures from piling up quotas out of all proportion, there is a limit on the quotas a proposal can receive — a maximum of five, say — and a transfer is irrevocable once made. Nobody can accumulate disproportionate power within the system.

Duplicate detection and smart merging

In a system with many users, similar or identical proposals will inevitably arise independently. Without a way of handling them, hundreds of near-identical proposals could split the support and never reach the threshold needed to escalate. The platform therefore includes a duplicate detection tool that works in three steps.

First, as someone writes their proposal, the system shows similar existing proposals in real time. Second, it offers the option of joining an existing proposal instead of creating a new one — in which case they get their monthly quota back, which encourages merging. Third, if the person decides to create a new proposal anyway, the system flags it to the moderators as a possible duplicate, so they can judge whether it is genuinely different or whether it would be better merged.

In the platform’s first versions this tool can start with a simple text search and evolve into an artificial intelligence system for clustering similar opinions, like the one used by the vTaiwan platform, which has proved very effective at bringing real consensus to the surface among apparently scattered positions.

The moderation system

Moderation by lot

Moderation on the platform — that is, deciding which proposals are fit for public debate and which are not — rests with moderation panels chosen by lot from among everyone registered. Everyone is on the moderation list by default, but if someone is picked and does not wish to take part, they can decline without any penalty. This is what we call a “passive inclusive lottery”: everyone in by default, with the right to opt out.

This model avoids a common bias of voluntary systems: that the same profiles always end up taking part — people with free time, a high level of education, already politicised. The passive lottery guarantees real statistical representation of society — retired people, young people, shift workers, families with small children — and the right to decline respects individual freedom.

A soft diversity quota could also be applied: if the people selected at random all turn out to be men aged 40 to 60, say, the system partially redraws until the sample is reasonably representative of the real population. Ireland applies a similar mechanism: participants in its citizens’ assemblies are selected with quotas for age, gender, region and social class.

Majority voting and overlapping panels

A moderation panel votes by majority on whether a proposal is fit for public debate. So that the system works robustly and does not collapse at times of high activity, it uses a design of parallel, overlapping moderation panels: groups of seven people serving two-week terms, with a third renewed every week. That way there are always people with moderation experience alongside newcomers; the panel never starts from scratch, and knowledge passes on organically.

This design solves three problems at once. First, it eliminates troll behaviour and spam without creating a caste of professional moderators with interests of their own. Second, it spreads responsibility among many rotating people, which makes corruption or capture of the system very difficult. Third, it builds civic involvement, because many people go through the experience of moderating, not just of proposing.

Hidden reputation

The platform keeps a reputation system tied to each account: how many proposals someone has made, how many passed moderation, how many received support, how many were implemented. But this reputation is hidden during the moderation process. Moderators do not see the reputation of whoever proposes while they assess the proposal; the information appears only in later records, for analysis of the system as a whole.

The reason is that showing reputation during moderation creates an unavoidable bias: if moderators see that the proposer has a high reputation, they will tend to approve with less attention; if it is low, with more suspicion. Informational blindness at the critical moment of decision is a principle that judicial systems already apply in various contexts, and applying it here guarantees that each proposal is assessed on its content alone.

For the same reason, moderation is also blind to authorship: whoever reviews a proposal does not know who wrote it. The proposer’s name, reputation and affinities stay outside the room. Only the proposal goes in.

Conviction support

A proposal that passes moderation needs support to move forward. But not all support says the same thing. An impulsive click on the day a proposal goes viral does not express the same as the sustained commitment of someone who believes in it week after week. If the two count equally, the system rewards quick mobilisation campaigns, not real conviction.

That is why supports are scarce too, and mature over time. Each person has a limited monthly budget of supports — ten, in the pilot. A new support counts one point; if it is kept for a week it counts two, and if it is kept for two weeks, three. It can be withdrawn at any time, and if that is done within the same month, the support comes back to be given to another proposal.

The effect is twofold. On one hand, last-minute campaigns carry little weight: anyone hoping to inflate a proposal all at once will not manage it with an avalanche of clicks. On the other, giving support becomes a considered decision, because every support given to one proposal is a support that cannot be given to another. The mechanism draws on what digital governance calls conviction voting, and follows the same logic as the proposal quota: when the resources for taking part are limited, every contribution counts.

The threshold escalation system

One of the most powerful and genuinely innovative elements of the design is the mechanism of escalation by support thresholds. Proposals are not born and left to die at a fixed level; they move up a level according to the support they win, gaining visibility and influence as they escalate.

At the first level — the neighbourhood — a proposal might need, say, fifty supports from residents. If it gets them, the platform would make it visible and ask for a public response. At municipal, regional or national level, progressive thresholds could give access to formal spaces for debate or registration. These effects are not automatically binding: they depend on the channels and commitments that each competent administration has approved and formalised.

The effect of this mechanism is profound. A subject that troubles a large part of the population surfaces organically and reaches levels where it is already politically impossible to ignore, without any party having decided to put it on the agenda. That breaks the monopoly parties hold today over what gets talked about, which is perhaps the most invisible power they have, and the most real. In political science this phenomenon is called a “deliberative cascade”: a subject that starts out local gains support and visibility until it becomes a matter of general interest that demands an institutional response.

Every proposal, at every level, has public real-time tracking: “under debate”, “with sufficient support”, “referred to the town council”, “in progress”, “implemented”, “dismissed, and why”. Absolute transparency of the process is the key to keeping trust. Whoever made the proposal does not just see that it has been heard; they see exactly where it stands at any moment, and why.

Blind proposals: depolarisation by design

One of the most original elements of the design is the possibility of looking at proposals from political parties without knowing which party they come from. In a special browsing mode on the platform, party proposals are stripped of every identifying mark — no logo, no colour, no party name. Whoever browses reads the content, judges it and expresses their degree of agreement or disagreement on the substance alone.

Empirical research shows that removing the party label significantly increases cross-party support for concrete proposals. Ideas that right-wing voters would instinctively reject if they knew they came from a left-wing party — or vice versa — are well received when presented “naked”. This is no academic exercise: it is a practical tool for bringing out the real consensus that exists in society and that party dynamics hide.

The purpose of blind proposals is not to abolish parties or to deny the legitimacy of political identities. It is to show citizens — and the parties themselves — that consensus is wider than they think, and that many of the barriers they see as insurmountable are, in reality, cardboard walls painted in party colours.

Proposals and complaints are different things

The platform draws a clear line between two kinds of contribution: proposals (ideas for improvement, new projects, innovative policies) and complaints (existing problems, damage, services that are not working). The distinction is not cosmetic; it is functional. A complaint about a broken street needs a very different handling process from a proposal to create a new cycle lane. Mixing them in a single channel would breed confusion, frustrate users and swamp moderation.

Complaints can have a lower-friction channel — for example, one that does not use up the monthly quota — and a faster response flow, tied directly to the municipal maintenance services. Proposals, by contrast, follow the full circuit of moderation, debate and escalation described above. Both flows live side by side on the platform, but on clearly separate tracks.

Artificial intelligence for synthesis

As the platform grows, the volume of opinions, comments and debates can become hard to analyse by hand. Adding a synthesis AI — one that groups similar opinions, identifies points of consensus and brings real tensions to the surface — can be a very valuable complement.

The reference model is vTaiwan, Taiwan’s digital democracy platform, which uses an artificial-intelligence clustering system to analyse thousands of opinions and automatically identify where most people agree, whatever their political affiliation. The result is striking: on almost every topic, the consensus is far wider than party debate would suggest. The AI does not make decisions; it lights up the common ground so that people can make better ones.

Switzerland: the semi-direct democracy that works

Switzerland is probably the modern state closest to a functioning direct democracy. Swiss citizens vote in referendums several times a year on questions ranging from tax policy to immigration rules. Any citizen can launch a popular initiative which, if it gathers enough signatures, goes to a binding popular vote. The system is not perfect — turnout is often modest and campaigns can be dominated by well-resourced groups — but it shows that direct participation at national scale is not a theoretical utopia.

Estonia: digital governance

Estonia has built one of the most advanced digital governance systems in the world. Electronic voting has been a reality for years, and digital identity lets citizens deal with the public administration quickly and securely. The Estonian experience shows that technology can make participation significantly easier, provided the system’s security, privacy and accessibility are guaranteed.

Decidim: the platform born in Barcelona

Decidim is an open-source platform created in Barcelona in 2016 that has become an international reference in digital participation. Hundreds of institutions and organisations around the world use it, and it has generated thousands of proposals, some of which public administrations have implemented. Decidim supports proposals, votes, online assemblies and participatory budgets.

Even so, Decidim has significant limitations compared with what we have described in this document. It lacks moderation by lot, the transferable quota, smart duplicate detection, threshold escalation and blind proposals. Decidim could serve as a technological base to build on, but the genuinely innovative pieces of the system we propose do not exist in any platform today.

g0v and vTaiwan: forking the government

Taiwan offers one of the most inspiring and least known precedents of what a digital civic culture can achieve. The g0v movement — pronounced “gov-zero” — is a community of programmers, designers and activists that literally forks government services (in software terms): it builds parallel versions of public portals and services that are more transparent, accessible and participatory.

g0v’s philosophy is identical to the one we set out in this document: not to replace government but to create a layer of participation that government cannot ignore. The movement’s most visible result was vTaiwan, the digital deliberation platform that has let Taiwanese citizens take a direct part in designing legislation on complex issues such as the regulation of ride-sharing, e-commerce and the online sale of alcohol.

Taiwan’s story has an especially significant epilogue: Audrey Tang, one of the central figures of the g0v movement, ended up being appointed the government’s digital minister. The person who had built tools to hold government to account from the outside was called on to reform it from within. It is the most tangible proof that the incremental strategy — building from outside until the inside cannot ignore it — can go a very long way.

Ireland’s citizens’ assemblies

Ireland has been, as noted earlier, an extraordinary laboratory for deliberative democracy. Its citizens’ assemblies — made up of people selected at random, with quotas for age, gender, region and social class to ensure representativeness — successfully tackled questions that parliamentary politics had been unable to resolve for decades.

The process is revealing. The Constitutional Convention (2012–2014) brought together 66 randomly chosen citizens and 33 politicians; the Citizens’ Assembly (2016–2018), 99 ordinary people with no office. In both cases, participants met over months, heard from experts across the spectrum, debated among themselves under rules of respectful deliberation, and arrived at recommendations that were later put to a popular referendum. On same-sex marriage (2015), support was 62%; on reform of abortion law (2018), 66%. Results that no politician would have dared predict.

The Irish experience shows that the lottery works, that deliberation works, and that ordinary people — with information, time and a suitable framework — are perfectly capable of making nuanced decisions on enormously complex and emotionally charged questions.

No proposal for democratic reform deserves to be taken seriously unless it is honest about its risks and limitations. To present a utopian vision without nuance is an exercise in self-deception that ends up discrediting the very ideas it sets out to defend. So we devote a few pages to the real tensions and dangers in what we propose.

The tyranny of the majority

One of the classic arguments against direct democracy is the risk of the “tyranny of the majority”: the possibility that a majority votes against the rights of minorities with no institutional filters to stop it. The argument is not mere rhetoric; it has real historical examples. At various times and in various places, majorities have voted against the rights of women, LGBT people, ethnic minorities and immigrants.

That risk exists and must not be played down. The platform we propose does not replace constitutional guarantees or fundamental rights; it operates within the existing legal framework. Proposals that violate fundamental rights must be filtered out by the moderation system, and institutional thresholds must be respected. Direct democracy is not — cannot be — unlimited; it needs to coexist with the rule of law.

Disinformation and manipulation

In an environment saturated with false information, direct popular decisions can be manipulated by the media, social networks or economic interests. The informed debate that took place on the Athenian Pnyx is hard to reproduce on a mass scale in a society where social media algorithms put outrage ahead of reflection.

The platform includes several mitigation mechanisms — moderation by lot, conviction support, duplicate detection, synthesis AI — but it would be naive to believe they are enough to neutralise informational manipulation completely. Civic education, the promotion of critical thinking and the regulation of digital information ecosystems are indispensable complements that lie outside the scope of a technology platform but do much to determine its success.

The digital divide and new exclusions

A digital participation system excludes, by definition, people without access to technology or with limited digital skills. If it is not handled with care, the platform could reproduce — or even amplify — existing inequalities, creating a “digital democracy” in which the usual people take part with a new tool.

To prevent this, the platform has to be complemented with non-digital channels of participation — physical access points in town halls or civic centres, in-person help for people who struggle with technology, hybrid formats that combine the face-to-face and the digital. The aim is for technology to add without subtracting: to make participation easier for those who already have access without making it harder for those who need another channel.

Technological solutionism, revisited

We warned earlier about the risk of technological solutionism, but it is worth returning to in more detail because it is probably the subtlest and hardest danger to manage. The temptation to believe that “with a good app, everything gets fixed” is enormous, especially in a generation used to solving problems with digital tools.

But democracy is not an optimisation problem; it is a human process of living together. Deliberation requires empathy, listening, the ability to change your mind — qualities that a screen can help with but never replace. The platform is a necessary but not sufficient condition. The sufficient condition is a community that uses it with a genuine will to collaborate, and that is built with education, shared experience and time.

Low and unequal participation

The experience of digital participation platforms shows a pattern we must keep in mind: participation tends to be low and concentrated in a few profiles. In many processes, a small share of the people registered makes most of the contributions, and those people tend to have more time, more education and more of a habit of taking part. A platform open to everyone does not, by itself, guarantee that everyone is there.

The design eases this in part — the quota evens out each person’s weight, and the lottery brings into moderation people who would never have put themselves forward — but it does not solve it. We will need to measure who takes part and who does not, and actively seek out the sectors that are missing, above all outside the digital environment.

Capture by organised groups

Any participation tool can be captured by the best organised: associations with great mobilising capacity, interest groups or the parties themselves, who can turn the platform into one more battlefield. The risk is not that they organise — collective organisation is legitimate and necessary — but that the ability to mobilise comes to outweigh real consensus.

Several mechanisms act as counterweights: the quota limits each person’s weight, conviction support takes the force out of quick campaigns, the cap on transfers prevents the stockpiling of quotas, and blind moderation prevents a proposal from being rewarded or punished because of who is pushing it. None of these mechanisms is infallible, and we will need to watch for patterns of coordinated behaviour.

The privacy of political opinions

A platform like this inevitably accumulates sensitive information: what each person proposes, what they support, how they moderate. The European General Data Protection Regulation treats political opinions as a special category of data, with reinforced protection. This is no formality: in the wrong hands, knowing what each person in a town thinks is a tool of pressure.

That is why privacy must be a design principle and not an add-on: collect only the data that is essential, do not publicly expose who supports what, separate verified identity from civic activity wherever possible, and guarantee the right to be forgotten without erasing the collective memory of the process.

The system’s resistance

Finally, we have to be honest about an uncomfortable reality: elected officials have no interest in giving up power. A system that gives citizens a direct voice inevitably erodes the parties’ monopoly over the political agenda, and that will generate resistance. Some will be explicit — political opposition, hostile regulation, media delegitimisation. Others will be subtle — co-optation of the system, bureaucratisation of participatory processes, formal mechanisms that look like participation but lack its substance.

The incremental strategy we propose is, in part, a response to this risk: if the platform starts with a small community, shows results and scales organically, it is far harder to stop than if it starts with a legislative proposal that the parties can block. But the system’s resistance is a real factor that will shape the whole process, and one to keep in mind at every step.

Twenty-five centuries ago, on the hill of the Pnyx, any citizen of Athens could climb onto a stone and address everyone else to propose a law, debate a peace treaty or denounce an abuse of power. They needed no party’s permission, no lobby’s backing, none of the visibility that wealth or fame brings. They needed only one thing: the will to speak. And the system guaranteed them another: the right to be heard.

That right — isegoria — is what we have lost. Not the right to vote every four years, which we formally keep; but the right to be an active part of a continuing conversation about how we want to live together. The right to propose, debate, decide and see our decisions become reality. The right, in short, to govern ourselves — not in the abstract and by delegation, but concretely and directly.

The preceding pages have tried to show that the loss of this right was not accidental but deliberate; that modern representative systems were expressly designed to filter popular participation; and that the word “democracy” was appropriated to lend legitimacy to a system that its own founders did not consider democratic. But they have also tried to show that recovering it is possible — not totally and instantly, but gradually, locally and cumulatively.

The platform we have described in functional terms is not a theoretical exercise. It is a concrete proposal, buildable with today’s technology, testable in a small pilot and scalable with no theoretical limit. It brings together lessons from twenty-five centuries of democratic experiment: the Athenian lottery, the rotation of power, collective moderation, conviction support, threshold escalation, depolarisation by design and absolute transparency of process.

It does not set out to replace existing institutions. It sets out to create a layer of participation beneath them — a democratic infrastructure for the twenty-first century and a collective intelligence aggregator — that turns scattered contributions into clear, tested and representative proposals. A flow that starts in a neighbourhood, spreads to a municipality, escalates to a region and, eventually, reaches the state. Not by imposition or automatic binding force, but through the strength of evidence, consensus and traceability.

The architecture of the system can be summed up in three principles: low friction to get in, high quality to rise, maximum transparency to earn trust. Anyone can propose with a single click; but for a proposal to go far it must pass progressive quality filters — moderation by lot, conviction support, threshold escalation — that ensure what reaches the institutions is genuinely representative of a real collective concern. And the whole process is public, traceable and auditable.

The challenge is not technical. The tools exist. The challenge is cultural and political: to build a culture of participation in a society that has learned not to participate, and to do so in a context where the existing system has every incentive to resist change.

But the precedents exist — Switzerland, Estonia, Ireland, Taiwan, Barcelona — and conditions are favourable. We have never had so much technology within reach, so much disaffection with the current system or so much academic evidence that another model is possible.

Perhaps the time has come to climb back onto that stone.

This document has been deliberately non-technical. But the story it tells needs, if it is to become real, a second document defining the platform’s technical architecture. The notes below are the starting points for that second document, and each will have to be developed in depth.

Identity and registration

The system needs a registration mechanism that verifies that each account belongs to a real and unique person, without demanding a level of verification so high that it deters participation. In a first pilot with a small community, light verification may be enough — for example, an email address tied to a verifiable postal address. In later versions, integration with existing digital identity systems, such as the digital certificate or electronic identification, can be explored. The key is to find the balance between security and accessibility at each stage of deployment.

Quota and transaction management

The system of monthly quotas, partial accumulation and transfer requires a reliable transaction engine that guarantees no quota is duplicated, lost or wrongly reassigned. The business rules will need to be defined precisely — what happens if someone transfers a quota and the recipient does not use it, what the expiry period is, how edge cases are handled — and implemented with robust transactional logic.

Lottery algorithm

The drawing of moderation panels must be verifiably random and auditable. We will need to define the selection algorithm, the diversity quota mechanism, the rotation logic for overlapping panels and the replacement procedure when a selected person declines. Transparency in the drawing process is essential to trust in the system.

Duplicate detection engine

The system for detecting similar proposals can start with basic semantic search based on text embeddings and evolve towards more sophisticated clustering models. We will need to define the similarity thresholds, the user interface for presenting similar proposals, and the interaction flow when a person decides to merge with an existing proposal or create a new one.

Escalation architecture

The threshold escalation mechanism requires a precise definition of the levels, the thresholds, the scopes of visibility and the associated institutional commitments. The pilot starts from a clear rule: escalating is not the same as obliging. We need to define how proposals are handled where the receiving institutions have no legal obligation to respond, how inaction is documented, and how an administration can voluntarily take on commitments to respond, to provide resources or to deliver.

Blind proposals interface

The blind proposals browsing mode requires a content depersonalisation system that removes not only the obvious marks (logo, party name) but also the indirect clues that could reveal the origin — habitual terminology, writing style, implicit references. We will need to define the degree of neutralisation required and the procedure for achieving it.

Synthesis and clustering AI

Implementing a synthesis AI similar to vTaiwan’s requires a natural language processing model able to analyse large volumes of opinions, group them by semantic affinity and generate visual maps of consensus and dissent. We will need to assess whether to use a proprietary model, an open-source model or a bespoke development, and to define the grouping criteria, the visualisation formats and the mechanisms for human validation of the results.

Infrastructure and security

The platform must guarantee availability, data integrity and the protection of its users’ privacy. We will need to define the server architecture, the backup policy, the protections against attack, compliance with data protection regulations and the security audit procedures. Where a platform handles sensitive information — political opinions, identities, voting patterns — security is not an add-on but a founding requirement.

Pilot strategy

The first pilot is in Cardedeu, a municipality in the Vallès Oriental, together with the Fòrum Obert de Cardedeu: the conclusions of its in-person roundtables have been brought into the platform as diagnoses and proposals, and residents can prioritise them with conviction support. It deliberately starts with a small core — proposals, diagnoses, topics and support; moderation by lot and escalation to the town council will be switched on in later phases. Other municipalities can ask for participation to be opened up there, and they are opened one at a time. We will need to define the criteria for opening new municipalities, the success metrics, the data collection plan, the iteration procedure and the plan for communicating results. The aim of the pilot is not to prove that the platform works perfectly but to learn everything that works badly, so as to fix it before scaling.

Decidim as a base, or in-house development

A fundamental technical decision is whether to build on the Decidim platform — drawing on its open-source ecosystem and functional base — or to develop from scratch with an architecture of our own. Both options have advantages and drawbacks. Decidim offers a solid base and an active community, but the innovations we propose might call for deep architectural changes that are hard to implement as extensions of an existing platform. The technical document will have to weigh both options with judgement.

———

These starting points do not exhaust what the technical document will need to address, but they identify its main axes. Each will require careful thought, an assessment of alternatives and a concrete implementation proposal. What I hope this first document has achieved is to establish the why — the conceptual and philosophical framework — clearly enough for the technical how to be built on it with sense and coherence.

Works cited in this document or on which its arguments are based.

Aristotle (4th c. BC). Politics, books III and IV.

Baiocchi, G. (2005). Militants and Citizens: The Politics of Participatory Democracy in Porto Alegre. Stanford: Stanford University Press.

Barandiaran, X. E., Calleja-López, A., Monterde, A. and Romero, C. (2024). Decidim, a Technopolitical Network for Participatory Democracy. Cham: Springer.

Barber, B. R. (1984). Strong Democracy: Participatory Politics for a New Age. Berkeley: University of California Press.

Boxell, L., Gentzkow, M. and Shapiro, J. M. (2024). “Cross-Country Trends in Affective Polarization”. The Review of Economics and Statistics, 106(2), 557–565.

Cohen, G. L. (2003). “Party over Policy: The Dominating Impact of Group Influence on Political Beliefs”. Journal of Personality and Social Psychology, 85(5), 808–822.

Farrell, D. M., Suiter, J. and Harris, C. (2019). “‘Systematizing’ constitutional deliberation: the 2016–18 citizens' assembly in Ireland”. Irish Political Studies, 34(1), 113–123.

Fishkin, J. S. (2009). When the People Speak: Deliberative Democracy and Public Consultation. Oxford: Oxford University Press.

Habermas, J. (1962). Strukturwandel der Öffentlichkeit. Neuwied: Luchterhand.

Habermas, J. (1992). Faktizität und Geltung. Frankfurt: Suhrkamp.

Hamilton, A. (1777). Letter to Gouverneur Morris, 19 May 1777.

Hansen, M. H. (1991). The Athenian Democracy in the Age of Demosthenes. Oxford: Blackwell.

Herodotus (5th c. BC). Histories, V, 78.

Iyengar, S., Lelkes, Y., Levendusky, M., Malhotra, N. and Westwood, S. J. (2019). “The Origins and Consequences of Affective Polarization in the United States”. Annual Review of Political Science, 22, 129–146.

Landemore, H. (2013). Democratic Reason: Politics, Collective Intelligence, and the Rule of the Many. Princeton: Princeton University Press.

Landemore, H. (2020). Open Democracy: Reinventing Popular Rule for the Twenty-First Century. Princeton: Princeton University Press.

Madison, J. (1787). “The Federalist No. 10”.

Manin, B. (1995). Principes du gouvernement représentatif. Paris: Calmann-Lévy.

Montesquieu (1748). De l'esprit des lois, II, 2.

Morozov, E. (2013). To Save Everything, Click Here: The Folly of Technological Solutionism. New York: PublicAffairs.

OECD (2020). Innovative Citizen Participation and New Democratic Institutions: Catching the Deliberative Wave. Paris: OECD Publishing.

Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge: Cambridge University Press.

Ostrom, E. (2010). “Beyond Markets and States: Polycentric Governance of Complex Economic Systems”. American Economic Review, 100(3), 641–672.

Pateman, C. (1970). Participation and Democratic Theory. Cambridge: Cambridge University Press.

Regulation (EU) 2016/679 of the European Parliament and of the Council (General Data Protection Regulation), Article 9.

Rousseau, J.-J. (1762). Du contrat social, III, 15.

Sieyès, E. J. (1789). Speech to the National Assembly on the royal veto, 7 September 1789.

Weyl, E. G., Tang, A. and the Plurality community (2024). Plurality: The Future of Collaborative Technology and Democracy.