91av

Many of us entered this so-called super-election year with a sense of foreboding. So far, not much has happened to allay those fears. Russia’s war on Ukraine is exacerbating a perception that democracy is threatened in Europe and beyond. In the US, Donald Trump, a presidential candidate with , has faced two assassination attempts. And more broadly, people seem to be losing faith in politics. “Most people from a diverse array of countries around the world lack confidence in the performance of their political institutions,” says a 2024 report by the International Institute for Democracy and Electoral Assistance.

On many objective measures, too, democracy isn’t functioning as it should. The systems we call democracies tend to favour the rich. Political violence is growing, as is legislative gridlock, and worldwide, elections are becoming less free and fair. Some 30 years after commentators crowed about the triumph of Western liberal democracy, their prediction seems further than ever from being realised. What happened?

According to at the University Mohammed VI Polytechnic in Rabat, Morocco, we have lost sight of what democracy is. “We have made a terrible confusion between the system known as a republic – which relies on elections, parties and a permanent governing class – and the system known as a democracy, in which citizens directly participate in decisions and rotate power.” The good news, he says, is that we can revive the original dream of government by the people for the people. That is what he and other researchers are trying to do. With insights from complexity theory, innovative online platforms and help from artificial intelligence, they are reimagining democracy for the 21st century by returning collective intelligence to the core of politics.

A modern Athens

The word and concept of democracy dates from around 500 BC, when the people of Athens in ancient Greece devised a radical new system of government to harness collective know-how to find the best solutions to their shared problems. In doing so, they grasped intuitively what a now confirms: a cognitively diverse group produces better solutions than even the most brilliant individual. Admittedly, they left a lot of people out of their political problem-solving – notably women and enslaved people – but their system was far more inclusive than anything that had gone before, and it paid dividends. Among the city state’s many achievements were important advances in astronomy, maths, literature and medicine.

Today, most of us who think we are living in democracies are in fact living in systems closer to oligarchies, where the governing is done by small, usually . These elites are not only buffered from most people’s problems, but from the problem-solving capacity of the crowd, too. The result is that many people feel they aren’t being well served. In many places, there has been a populist backlash. But populism is even worse at solving our increasingly complex problems. “A lot of the problems we face as a species are the result of poor large-scale decision-making processes,” says at the Massachusetts Institute of Technology.

Klein and Paulson are among a growing number of academics advocating a return to the Athenian model, scaled up for the modern world. For Paulson, who worked on campaigns for presidents Barack Obama of the US and Emmanuel Macron of France, the solution is three-pronged. First, revert to the Athenian practice of sortition or “lottocracy”, where policy-makers are chosen by random, representative selection, as juries are in many countries. Second, capture as much collective intelligence as possible through AI-augmented deliberation. And third, reinstate civic education to produce citizens who are willing and useful participants of a true democracy.

Political scientists and tech experts have been discussing these ideas for 50 years, but only in the past decade have they begun talking to each other. And funding is now available to explore democratic innovation – for instance under the European Union’s and through a , the makers of ChatGPT. Momentum is building, says Paulson. The question is, can such innovations revamp democracy and restore stability in places where social crises and political violence are looming?

It is a big ask, because an electorate is a complex system that can behave in non-intuitive ways, becoming dysfunctional when pushed through so-called tipping points. “It’s like freezing water,” says Alexander Siegenfeld, a physicist at MIT. “If I keep lowering the temperature, nothing happens, nothing happens, and then I lower it a bit further and all of a sudden the water freezes.”

and his colleague at the New England Complex Systems Institute in Massachusetts used mathematical modelling to . Political polarisation waxes and wanes, but they found that when high polarisation combines with low voter turnout, the system becomes unstable, producing violent swings in outcomes. This is to do with people who have more extreme views being more likely to vote and politicians chasing these votes. As a result, the system produces governments and policies that most people don’t want, frustrated voters then abstain and the instability is exacerbated. The US went through such a period in the 1970s, the researchers found, and elections have become increasingly unstable ever since – culminating in the 2020 presidential election and the political violence that followed. “This isn’t just a patch of disturbance,” says Bar Yam. “Exit requires change.”

Siegenfeld and Bar-Yam believe replacing the first-past-the-post electoral system, which prevails in most US states, with ranked-choice voting and making voting compulsory would help restore stability. Others, however, argue it would still fail to harness collective intelligence. Political scientist at Yale University is among them. She wants to see the return of sortition in the form of – effectively citizens’ assemblies with randomly selected, demographically representative members. Although unlikely to be the shape of national governance anytime soon, in recent decades such assemblies have become more influential in Western countries. In , for instance, one contributed to the legalisation of abortion in 2018. And in France, Landemore guided another that culminated in now before parliament. “[Mini-publics] are showing us a way of doing politics differently,” says at the University of Westminster, UK. He thinks they are better suited to solving 21st-century problems than the structures that govern us now.

Many problems we face are the result of poor large-scale decision-making

But citizens’ assemblies have drawbacks: they are vulnerable to influence from whoever sets the question being examined or hires experts to advise them, and can also be time-consuming. In the UK, a charity called Nesta is trying to minimise these problems in an ongoing experiment called the . Twelve people sit around a table and watch a factual presentation on the problem delivered via a tablet. Their initial views are polled, also by tablet, and the results kick off a facilitated discussion. They are polled again before leaving and in under 2 hours “we see a small but significant shift in people’s preferences”, says , director of Nesta’s Centre for Collective Intelligence Design. She says this indicates that people are thinking collaboratively and listening to each other. It also chimes with evidence that people reason most effectively through argument. Participants report enjoying the experience too.

However, this approach doesn’t address perhaps the biggest problem with mini-publics: there are limits to how many people can take part before the debate quality suffers. Research suggests the – seven people or fewer. That is difficult to square with ensuring cognitive diversity. But the solution could be under our noses.

Large groups of people talk to each other all the time on social media. Far from halting the democratic backslide, however, this dialogue . At present, online discussion platforms leave our collective intelligence almost untapped, says Klein. The essential problem, in his view, is that the conversation is structureless. “We’ve used internet technology to take regular, unstructured conversations and try to do them with hundreds of thousands of people, and it just doesn’t scale,” he says, resulting in a huge comment pile where ideas and voices get lost. People resort to unproductive strategies to get heard, including clique formation and attention wars, while others self-censor or drop out. The result is groupthink and toxicity – emergent dysfunction, once again. However, capturing collective intelligence online is possible, as Wikipedia’s success shows. Despite being compiled by unpaid volunteers, it is written by paid experts, and more frequently consulted.

Better online tools

The challenge is to improve online platforms and find ways of combining them with face-to-face discussion for a more representative debate. Klein and his colleagues have developed a tool called the , which guides users through a process that is transparent and structured by topic (see Disagreeing agreeably, below). Engineers are familiar with the concept of the , the subset of all possible solutions that score optimally on multiple criteria. At the moment, we spend a lot of time far from that front. But the Deliberatorium is designed to get people closer to it – and it seems to work. In real-life situations where Klein has tested the tool – including at corporations Siemens and BP, and the Democratic party in Italy – several hundred people have found an optimal solution in as little as a day. “If the medium is the message, this is about changing the medium to change the message,” he says.

Others are scaling up the debate in different ways. In Tilburg, the Netherlands, is using large language models (LLMs) – the technology that powers ChatGPT – to summarise small-group discussions in close to real time. The goal is to then share these AI-generated summaries between groups to “cross-pollinate” the debate among a larger number of people.

Meanwhile, another , also in the Netherlands, took the recommendations of a citizens’ assembly on energy sustainability and presented them to a larger group of people via a digital platform. That larger group ranked them, explaining their arguments; this information was then fed on to a second citizens’ assembly, which assimilated it before presenting adjusted recommendations to the municipal council – which approved them unanimously. Peach describes this approach as a “gold standard” and wants to see it adopted by the UK government. However, although local authorities have shown enthusiasm for the Strategy Room model, she has been told there is no money to roll it out nationally.

Perhaps it is unsurprising that governments don’t tend to see democratic reform as a priority. “When you are struggling to pay for all your statutory responsibilities… public engagement falls to the bottom of the pile,” says Peach. Arguably, though, governments are resisting their own salvation. For instance, since embracing mass participatory democracy just a decade ago, Taiwan, which many cite as a model, has seen .

Change will also need support from citizens. This is where the third pillar of Paulson’s solution, civic education, comes in. We need to relearn the value of collective intelligence. The Athenians learned democracy by doing it, and through such communal activities as civic festivals, sport and theatre, that helped them understand themselves as part of something bigger. In Taiwan, the recently overhauled curriculum emphasises co-creation and the common good from the age of 6.

But even the best deliberative technology won’t get you far if people are resistant to new ideas and perspectives, says at the University of Technology Sydney, in Australia. He spent years developing tools for online collaboration before turning his attention to education, and thinks what is required is a certain mindset. “The fancy words would be ‘epistemic humility’, which is to hold your beliefs lightly and to be ready and open [to other people’s],” he says. He has built an LLM-powered chatbot called that, rather than answer a student’s question, challenges the assumptions underlying it.

Political change requires bottom-up and top-down inputs, says Paulson – both a participatory movement and creative leadership. Once those are in place, it takes a spark – a kairos in ancient Greek, or “opportune moment” – to mobilise them. That could be an acute crisis to which government is perceived to have responded inadequately. Covid-19 didn’t do it, and climate change is unfolding too slowly to qualify. If and when it does happen, however, there will at least be a blueprint for restoring an idea of demoracy that works for everyone.