Citizens’ assemblies are having a moment. Growing distrust in governments and experts has revived the ancient Greek method of empowering the public to participate in politics.
But selecting the members of these bodies is a complicated task. Ideally, citizens’ assemblies should be both representative and randomly selected. Balancing these two requirements is challenging as the volunteers tend to be unrepresentative of the whole population.
A team of computer scientists from Harvard and Carnegie Mellon universities has devised a potential solution: selection algorithms.
The team’s system finds panels that satisfy quota requirements and give potential members as equal a chance of selection as is mathematically possible.
The algorithm first constructs a set of quota-dependent panels. These are developed by iteratively building an “optimal portfolio” of panels and computing the fairest distribution of participants. A single panel is then randomly drawn from the distribution.
The open-source algorithm has already been used to select more than 40 citizens’ assemblies around the world. In Michigan, the system was used to pick a panel of 30 residents to make recommendations about COVID-19.
The process was a success, according to a June report by Fast Company:
Their chosen panelists were demographically representative, gender-balanced, aged 20 to 87, and had wide variations in race, education, and political views. The panel emerged with 12 policy recommendations on handling COVID-19 and the economy, including on mask mandates, unemployment benefits, and home relief grants.
The researchers will now explore new ways that computer science can contribute to democratic practices.
In a time of declining respect for professional politicians, the team’s algorithms could enhance an alternative form of representative democracy.
You can read the open-access study paper in Nature.
Greetings Humanoids! Did you know we have a newsletter all about AI? You can subscribe to it right here.