Democratic systems worldwide are not just underperforming — they are structurally failing. Declining trust, polarized publics, fragile coalitions, an inability to approach hard questions with long time horizons and the growing influence of disinformation are oftentimes treated as separate problems. However, they are not. Taken together, they point to something deeper: a fundamental crisis in how societies make collective decisions — at the very moment when the stakes have never been higher.
This essay contextualises the current crisis of democracy within the broader history of political experimentation. Building on this perspective, it introduces contemporary deliberative and digital democracy experiments as concrete responses — and argues that our democratic operating systems need no reform, but rather a redesign built on collective intelligence.
On ever-changing social order
Human societies are ultimately highly-complex communication designs, spanning across space and time and evolving along the social technologies that enable them. Social systems distribute the difficult task of understanding complex issues across different people, so the population as a whole can make decisions. But this makes a decisive question unavoidable: who controls the discourse? Who gets to decide how agendas are set, how decisions are informed, structured and conducted? These questions are much older than even the earliest recorded concepts of political philosophy itself.
Throughout history, power relations were often very fluid, as reconstructed by David Graeber and David Wengrow in their new history of humanity. One key example concerns nineteenth‑century Inuit communities, who lived in small, dispersed family bands during the summer, marked by strong paternal authority and strict rules of private property. In winter, those bands merged into large communal houses, where norms of private property were suspended, goods were freely shared, and social life became markedly more egalitarian and collective. Building on earlier ethnographic evidence, the two authors argue that the Inuit were politically self‑conscious, deliberately adopting different organizational modes for different seasons. These modes were not just preset practical adaptations, but conscious political choices: people knew alternative systems existed and chose when and how to implement them.1
This pattern was not unique to the Iniut: based on extensive ethnographic and archaeological evidence, the authors show that seasonal political variation emerged repeatedly across contexts. The record reaches back to paleolithic Europe fifteen to twenty thousand years ago, when hunter-gatherer societies organized along these patterns for long periods during the ice age. Roughly eleven thousand years ago, the same patterns culminated in the construction of Göbekli Tepe in what is now Turkey. A few hundred people gathered seasonally for religious reasons and organized the construction of a small city two or three millennia before the advent of agriculture. Four thousand years ago, the first mega-settlements without monarchy evidence emerged alongside agriculture not only in Mesopotamia, but also in Tripillia, nowadays Ukraine – ten thousand people living egalitarian, likely via public debate and collective decisions.2
Graeber and Wengrow suggest political experimentation is prehistoric. It may be stretching back to language evolution as core social technology.3
Democracy as social experimentation
Fast forward to ancient Athens in the fourth century BC. The so‑called cradle of democracy and home to roughly a quarter of a million inhabitants, only about twenty percent of its population – eligible male citizens – shaped political life, while slaves, women and foreigners were excluded. For the organization of democratic life, a special device was developed: the Kleroterion. Its function was to draw the wooden “IDs” of citizens at random in a transparent way and distribute them across an array of political institutions.
After a long tradition of using lots in religious and cultural contexts, the device emerged after Athens’ defeat in the Peloponnesian War and two oligarchic coups, leading to a political system that made lottery‑based allocation a core tenet to fight corruption and elite capture, while elections were reserved for specialist roles like military leaders. Every male Athenian could reasonably expect to be selected and to participate in multiple public functions throughout his life. For about eighty years, the Kleroterion remained a core social technology in a flourishing society, before vanishing for more than two millennia after the conquest by Macedonia and later Rome.4
The Roman republic, by comparison, developed a different mixed system that primarily relied on elections for officials. Even beyond the powerful senate of wealthy citizens, elections were, for religious reasons, held only in the city of Rome and thus limited to those able to attend. Literacy was high in the upper strata, but remained low for the majority. Under such conditions, declining citizen participation was effectively enshrined in the expanding empire, until the system’s demise.
The story is usually continued as nearly two millennia of hierarchical European societies until the Enlightenment laid the foundations for modern representative democracy and voting rights. Yet experimentation with bottom‑up organization without kings or entrenched hierarchies also appeared in places like Teotihuacan, near today’s Mexico City. Inhabited by up to two hundred thousand people, evidence points to a city governed by local assemblies and neighborhood councils in an egalitarian fashion for several centuries. The civilization predated the Aztecs by one thousand years and reflected traditions that shaped parts of Central- and North America long before European settlers arrived – and continued to do so long afterward.5
Seen from that perspective, it is no coincidence that representative democracy evolved in the former British colonies at a time when the Enlightenment philosophers, especially Montesquieu, became influenced by discussions with their Native American counterparts. They visited continental Europe and found societies with low living standards, high inequality and populations with much lower political education. They challenged the European philosophers with this observation in the early enlightenment salons, thereby influencing the political philosophy of the time by no small means.6
By the eighteenth century, printing and literacy in the colonies were already well established: religious texts had long underpinned reading culture, while newspapers and pamphlets enabled the rapid spread of political ideas. Literacy in some colonial regions, particularly New England, reached unusually high levels for the time, creating a broad audience for political argument.
The developments in political philosophy, the accelerated spread of ideas as well as the house of commons in Britain as reference point ultimately set the foundation towards political representation by elections. Lottery based principles to shape political organization between the ruling families were successfully practiced for hundreds of years in the Italian city states Florence and Venice. But while being discussed, they didn’t leave a strong impact on the leading philosophers and revolutionary factions.7 James Madison himself argued on the subject: ”In a democracy the people meet and exercise the government in person; in a republic they assemble and administer it by their representatives and agents. A democracy, consequently, will be confined to a small spot. A republic may be extended over a large region.”8 His views became hegemonic.
The U.S. constitutional order was created by elites who thrived for a representative republic that significantly limited direct popular power and strongly protected property interests. They made the system quite exclusive from the start while knowing in parts that alternative ways existed. Male suffrage and the abolition of slavery in the nineteenth century, as well as female suffrage and black enfranchisement during the twentieth century were by no means historically predestined; they emerged from intense social struggles — and are once again under threat. The US order became nonetheless an inspiration for countries around the world, despite the wide-ranging flaws of its blueprint.
Democracy crisis is a feature, not a bug.
A different logic prevailed for the French Revolution, yet it was marked by similar tensions and contradictions. Once the Ancien régime had been dismantled, no institutional framework existed to mediate competing claims by rival factions to speak “for the people.” The Jacobins filled this vacuum by collapsing the distinction between revolutionary vanguard and general will – dissent became treason.
High literacy and the spread of printing technologies didn’t safeguard against tyranny; instead they accelerated the Reign of Terror, accepted by a majority of Parisians. In the countryside, capacity for collective organization against authoritarian backlash remained limited. Coordinating inter-regional resistance and formulating constitutional alternatives proved a near-insurmountable bottleneck.
The US- and French pattern repeated – and keep repeating. Many countries transitioning towards representative democracy faced heavy backlashes and only partial successes. Old elites lost power to new elites who harnessed popular struggle, then implemented systems limiting citizen participation from the start. While a handful of democracies remained stable from their founding, many faced backslidings, as representative governments always rested on a combination of democratic and oligarchic elements. The democratic ingredient was never guaranteed to prevail.
Elections require candidates to stand out — wealth, education, rhetorical skill, name recognition, charisma. Representatives may listen to constituents but retain the right to act according to their own judgment. This draws power to the more affluent by default and invites influencing attempts from the start. The development from pure representation to party democracies towards audience democracies, with ever-increasing media impact on a declining public discourse, has not altered that logic. It has deepened it.9 The second Trump administration is the wealthiest democratically elected government in history.
The iron law of oligarchy, formulated by Robert Michels over a hundred years ago, reinforces this point: “When you say “organization,” you mean a tendency toward oligarchy.” As organisations grow in size and complexity, they require specialised leadership and bureaucratic structures that tend to concentrate power in the hands of a few, regardless of how democratic their founding principles may be. As democratic theory of elitism, his notion holds value until today.10
Given the current state of the world, the question is whether representation in its current form can remain the default mode. The answer may lie in emerging experiments reconnecting democratic innovation to older traditions while harnessing new technologies – rebuilding collective sense-making from the bottom up.
Basic configurations for digital democracy
By the late nineteen‑sixties, American political scientist James C. Miller argued that changing technology increased opportunities for voter participation in the legislative process. He proposed a mixture of direct and proxy voting, letting people delegate their vote to someone they trusted while retaining the right to reclaim it at any time. This became the founding idea for Liquid Democracy, formulated in the early 2000s around six attributes: direct voting, vote delegation, transitivity (delegated votes can be further delegated), recall at any level, vote modification and topic‑based delegation. Its founders hoped that reforming internal party processes would eventually help to void the iron law of oligarchy and therefore provide voters a more desirable choice. The concept gained prominence in several European Pirate Parties in the 2010s, who failed to secure lasting political influence. Liquid Democracy continued to evolve.11
In the same decade, another experiment emerged in Taiwan, where the open‑source tool Pol.is was used by the Sunflower Movement to contest an unfavorable trade agreement with mainland China. Best described as “wiki polling,” Pol.is lets participants agree or disagree with existing statements or add new ones; a machine‑learning system then maps opinion clusters and nudges participants toward broadly supported “bridging” statements. This enables crowd-sourced solutions and even draft laws at scales of thousands of participants. In a context of external threats from China and deep distrust in parties, the system quickly gained legitimacy in the young democracy. The Sunflower Movement succeeded, and one of its leading figures, Audrey Tang, later became Taiwan’s first minister for digital affairs. During her tenure through COVID‑19, Taiwan used similar crowd‑sourcing approaches for crisis management, contributing to low infection and death rates alongside relatively high civil freedoms.12 Since then, over twenty new laws have been crowdsourced while testing a widening array of digital technologies.
Another important line of experimentation, participatory budgeting, began in Porto Alegre, Brazil, in the late 2000s. Citizens used simple digital tools to collectively decide how to spend portions of the city budget. The success of this model led to adoption in cities worldwide, notably Barcelona and Madrid, inspiring more advanced platforms: the Decidim software became a kind of Swiss‑army knife for civic participation. Cities like Stockholm use it as a central interface where citizens can debate, propose spending priorities and help administer the process itself. Building on these experiences, Brazil’s second Lula administration launched Participativo Brasil, inviting more than a million people to weigh in on key societal questions through Decidim and public campaigns in major squares and media.13 The response exceeded what software and institutions could fully process, revealing both the limits of current capacity but also a strong public appetite for participation.
The use of lotteries in political theory and process re‑emerged since the late 1970s in the US and Europe under the umbrella term Sortition. In Germany, “planning cells” brought randomly selected people to deliberate under expert advice on contested local infrastructure and co‑design acceptable solutions since the seventies. This experimentation expanded into what the OECD later called the deliberative wave, with allotted citizens’ assemblies held across Europe at local and national levels.14 The most influential example was Ireland’s twenty-sixteen assembly on abortion, which paved the way for a constitutional referendum in a deeply Catholic country.
Climate assemblies were adopted by multiple governments in a consultative function, sometimes combined with digital participation processes for the agenda setting . The process gained by far the most public recognition in France.15 Other national assemblies were held on an array of topics, from Nutrition in Germany to Norway’s Future Assembly on Oil Fund.
Eastern Belgium created a permanent citizens’ council to set priority topics and commission ad‑hoc assemblies whose recommendations feed into politics; Paris recently adopted a similar model. In Oregon, the Citizens’ Initiative Review convenes a randomly selected group during referendum campaigns to evaluate proposals. Their findings are summarized in voter pamphlets, improving both debate quality and participation.16
Researchers have started asking in what dimensions citizens’ assemblies can be scaled with the help of AI.17 Big players like Anthropic, Meta and OpenAI have experimented with Sortition-based alignment-assemblies for either their constitutional models or inputs towards platform governance.18
Alongside the assemblies, deliberative polling developed as another way to harness random selection. Larger samples of citizens are polled on key issues, then split into small rotating groups to deliberate on a defined policy set before being polled again. Research led by James Fishkin at Stanford showed since the 1990s that participants often significantly shifted their views on heavy polarizing topics. To scale further, Stanford’s team created an AI‑supported video platform and achieved an on-par moderation compared to human moderators.19 The experiments have been run with impact around the world, deliberative polls are now even a formal step before constitutional changes in Mongolia.20
Despite their positive trajectory, many of these trends come with a profound catch: the government’s willingness to fully implement the recommendations and hand more power to its citizens.
Towards a new democratic operating system
Political experimentation has been a profound human trait for at least twenty thousand years. Prehistoric hunter-gatherer societies were not strictly egalitarian in every period, nor did the advent of agriculture, the creation of mega-settlements or the emergence of the nation state impose a predestined top-down organizational mode. Alternatives existed, were debated, fought for – and sometimes won. New social technologies enabled or enhanced them. The capacity for political reinvention is not an anomaly in human history. It is one of its defining features.
The crisis of democracy now sits at the center of the poly-crisis. Its original blueprint was never designed to activate public participation and collective intelligence. Sortition based concepts and new technologies to scale collective decision making have not yet gained as much traction,21 but they already showcase several well researched and practiced elements of a potential new operating system for the twenty-first century.
Most deep social changes have been shaped or accelerated by social struggle. However a future digital democracy system based on collective intelligence will look like in detail, at its heart must always be humans steering via intelligent augmentation (IA), instead of being steered by AI.22 To keep on par with the speed of technological development, the question of democracy ultimately becomes a question of democratizing AI.
References
- Graeber, D., & Wengrow, D. (2021). The dawn of everything: A new history of humanity, Chapter 3,. Farrar, Straus and Giroux.
- Graeber, D., & Wengrow, D. (2021). The dawn of everything: A new history of humanity, Chapter 8,. Farrar, Straus and Giroux.
- Dar, D. (2016). From experience to imagination: Language and its evolution as a social communication technology. Journal of Neurolinguistics.
- Manin, B. (1997). The principles of representative government, Chapter 1, Cambridge University Press.
- Graeber, D., & Wengrow, D. (2021). The dawn of everything: A new history of humanity, Chapter 9, Farrar, Straus and Giroux.
- Graeber, D., & Wengrow, D. (2021). The dawn of everything: A new history of humanity, Chapter 2, Farrar, Straus and Giroux.
- Sintomer, Y. (2023). The government of chance, Chapter 2. Cambridge University Press.
- Madison, J. (1787). Federalist No. 14. The Federalist Papers. https://avalon.law.yale.edu/18th_century/fed14.asp
- Manin, B. (1997). The principles of representative government, Chapter 6, Cambridge University Press.
- Drochon, H. (2020). Robert Michels, the iron law of oligarchy and dynamic democracy. Constellations, 27(2), 185–198.
- Valsangiacomo, C. (2021). Clarifying and defining the concept of liquid democracy. Journal of Deliberative Democracy.
- How to Save Democracy [Podcast]. Taiwan’s transformation
- People Powered. (n.d.). Bridging the digital divide: Lessons from Brazil’s national participatory planning process. https://www.peoplepowered.org/news-content/bridging-the-digital-divide-lessons-from-brazils-national-participatory-planning-process
- OECD. (2020). Innovative citizen participation and new democratic institutions. OECD Publishing. https://www.oecd.org/en/publications/innovative-citizen-participation-and-new-democratic-institutions_339306da-en.html
- Landemore, H. (2026). Politics without politicians. Allen Lane
- Gastil, John (2023), The Oregon Citizens’ Initiative Review: Long-term impacts in the context of direct democracy. In: the power of democratic innovations: The impact of participatory and deliberative institutions (pp.95-116) https://www.researchgate.net/publication/384968182_The_Oregon_Citizens%27_Initiative_Review_Long-term_impacts_in_the_context_of_direct_democracy
- Chwalisz Claudi, McKinney, Sammy (2025.). Five dimensions of scaling democratic deliberation with and beyond AI. https://www.demnext.org/projects/five-dimensions-of-scaling-democratic-deliberation-with-and-beyond-ai
- Siddarth, D. (2024, July 23). Can alignment assemblies bring democracy to Silicon Valley? Democracy Technologies. https://democracy-technologies.org/participation/can-alignment-assemblies-bring-democracy-to-silicon-valley/
- Gelauff, L., Nikolenko, L., Sakshuwong, S., Fishkin, J., Goel, A., Munagala, K., & Siu, A. (year unknown). Achieving parity with human moderators. In (Eds.), [The Routledge Handbook of Collective Intelligence for Democracy and Governance]. Taylor & Francis. https://www.taylorfrancis.com/chapters/oa-edit/10.4324/9781003215929-15/
- Stanford Deliberation. (2023). Mongolia holds its second ever national deliberative poll. https://deliberation.stanford.edu/mongolia-holds-its-second-ever-national-deliberative-poll
- People Powered. (2026). Impacts of citizens’ assemblies. People Powered: Global Hub for Participatory Democracy. https://www.peoplepowered.org/research-data
- Lin, S. Y. (2025). Amplifying transformative potential while designing augmented deliberative systems. AI Objectives Institute. https://ai.objectives.institute/blog/amplifying-transformative-potential-while-designing-augmented-deliberative-systems
